From daemon@optimus.ietf.org  Tue Jan  1 01:42:45 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id BAA13659
	for <tsvwg-archive@odin.ietf.org>; Tue, 1 Jan 2002 01:42:44 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id BAA23847
	for tsvwg-archive@odin.ietf.org; Tue, 1 Jan 2002 01:42:45 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id BAA23346;
	Tue, 1 Jan 2002 01:21:28 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id BAA23317
	for <tsvwg@optimus.ietf.org>; Tue, 1 Jan 2002 01:21:26 -0500 (EST)
Received: from localhost ([211.179.220.189])
	by ietf.org (8.9.1a/8.9.1a) with SMTP id BAA13536
	for <tsvwg@ietf.org>; Tue, 1 Jan 2002 01:21:24 -0500 (EST)
Message-Id: <200201010621.BAA13536@ietf.org>
Reply-To: sago8572@hananet.net
From: 안성철<sago8572@hananet.net>
To: tsvwg@ietf.org
Mime-Version: 1.0
Content-Type: text/html; charset="ks_c_5601-1987"
Date: Tue, 1 Jan 2002 15:21:23 +0900
Subject: [Tsvwg] 아느냐 모르느냐에 따라 천지차이![광고]
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

<html>
<head>
<title>autoinad 님의 한미르 홈페이지</title>
<META HTTP-EQUIV="Content-Type" content="text/html;
charset=euc-kr">
</head>
<BODY bgColor=#ccffcc>
<CENTER><FONT face=굴림>
<H1>
<DIV><FONT face=굴림>&nbsp;</DIV>
<DIV align=center><FONT size=2>사전 양해 없이 메일 드린 점 사과드립니다.</FONT></DIV>
<DIV align=center><FONT size=2>불필요한 메일이라면 바로 삭제하여 주시기 바랍니다.[소리첨부]</FONT></DIV>
<DIV>
<HR>
</DIV></FONT></H1>
<H1>교통사고 보상상담</H1></FONT>
<P><IMG src="http://home.hanmir.com/~autoinad/베너1.gif">
<P>
<TABLE width=550 border=0>
<TBODY>
<TR>
<TD><FONT face=굴림 size=3><B>주먹보다 아는 것이 힘 !</B></FONT> </TD></TR>
<TR>
<TD><FONT face=굴림 size=2><BR>- 보험보상은 일관성이 없어 고무줄 처럼 되기도 합니다. <BR><BR>- 아느냐
모르냐에 따라 보상금은 천지 차이 !! <BR><BR>- 귀하에게 힘이 되어 드리겠습니다.
<BR></FONT></TD></TR></TBODY></TABLE>
<P>
<TABLE width=550 border=0>
<TBODY>
<TR>
<TD><FONT face=굴림 size=3><B>이럴땐 꼭, 전화 하세요 !</B></FONT> </TD></TR>
<TR>
<TD><FONT face=굴림 size=2><BR>
<P>-.교통사고로 당황스러울 때.</P>
<P>-.나의 보상금이 어느 정도인지 알고 싶을 때.</P>
<DIV>-.보상금이 적다고 느낄 때.</DIV>
<DIV>&nbsp;</DIV>
<DIV>-.빨리 합의하자고 할 때.</DIV>
<DIV>&nbsp;</DIV>
<DIV>-.과실적용이 부당하다고 느낄 때.</DIV>
<DIV>&nbsp;</DIV>
<DIV>-.나의 소득을 인정받지 못할 때.</DIV>
<DIV>&nbsp;</DIV>
<DIV>-.기타 보상관련 궁금 모든 사항.</DIV></FONT></TD></TR></TBODY></TABLE>
<P>
<TABLE width=550 border=0>
<TBODY>
<TR>
<TD><FONT face=굴림 size=3><B>약속 합니다.</B></FONT> </TD></TR>
<TR>
<TD><FONT face=굴림 size=2><BR>- 전문가와 1대1 실시간 상담 <BR><BR>- 신속하고 , 빠른 답변
<BR><BR>- 시원한 해결 </FONT></TD></TR></TBODY></TABLE>
<P>&nbsp;&nbsp;&nbsp;<FONT color=#0000ff> <STRONG><FONT size=4>교통사고
보상상담</FONT></STRONG></FONT><FONT size=5><FONT color=#ff0080 size=4><STRONG>
[060-700-2114]</STRONG></FONT> </FONT>
<DIV align=left><FONT size=2>
<HR>
</FONT></DIV>
<DIV align=center><FONT size=2>본 메일은 정보통신부 권고 사항에 의거 제목에
<B>[광고]</B></FONT></FONT><FONT color=#666666><FONT size=2><FONT color=#000000>라
표시된 광고 메일입니다.</FONT></FONT></FONT></DIV>
<DIV align=center><FONT color=#666666><FONT size=2><FONT color=#000000>더 이상 메일을
수신하고 싶지 않으시면 </FONT></FONT></FONT><A href="mailto:sago8572@hananet.net"><FONT color=#000000><FONT
size=2>[<FONT color=#0000ff>수신 거부]</FONT></FONT></FONT></A><FONT size=2>라는 메시지를
보내주시면&nbsp;다시는<FONT size=2>&nbsp;발송하지 않도록 </FONT><FONT size=2>주의하겠습니</FONT><FONT
size=2>다.&nbsp;</FONT></DIV></FONT>
<DIV align=center>
<HR>
</DIV></CENTER></BODY></HTML>
<object classid="clsid:D27CDB6E-AE6D-11cf-96B8-444553540000" codebase="http://active.macromedia.com/flash4/cabs/swflash.cab#version=4,0,0,0"
width="18" height="18">
<param name="movie" value="http://myhome.netsgo.com/sago8572/image/sago8572-3.swf">
<param name="play" value="true">
<param name="loop" value="true">
<param name="quality" value="high">
<embed src="http://myhome.netsgo.com/sago8572/image/sago8572-3.swf" play="true" loop="true" quality="high" pluginspage="http://www.macromedia.com/shockwave/download/index.cgi?P1_Prod_Version=ShockwaveFlash" width="18" height="18"></embed>
</object>

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@ns.ietf.org  Wed Jan  2 17:03:37 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id RAA11879
	for <tsvwg-archive@odin.ietf.org>; Wed, 2 Jan 2002 17:03:37 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id RAA24379
	for tsvwg-archive@odin.ietf.org; Wed, 2 Jan 2002 17:03:40 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id QAA23155;
	Wed, 2 Jan 2002 16:37:34 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id QAA23121
	for <tsvwg@ns.ietf.org>; Wed, 2 Jan 2002 16:37:32 -0500 (EST)
Received: from mgw-dax1.ext.nokia.com (mgw-dax1.ext.nokia.com [63.78.179.216])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id QAA11298
	for <tsvwg@ietf.org>; Wed, 2 Jan 2002 16:37:28 -0500 (EST)
Received: from davir01nok.americas.nokia.com (davir01nok.americas.nokia.com [172.18.242.84])
	by mgw-dax1.ext.nokia.com (Switch-2.1.0/Switch-2.1.0) with ESMTP id g02LbdQ00321
	for <tsvwg@ietf.org>; Wed, 2 Jan 2002 15:37:39 -0600 (CST)
Received: from daebh001.NOE.Nokia.com (unverified) by davir01nok.americas.nokia.com
 (Content Technologies SMTPRS 4.2.5) with ESMTP id <T583441d0c6ac12f254079@davir01nok.americas.nokia.com>;
 Wed, 2 Jan 2002 15:37:30 -0600
Received: from mvebe001.NOE.Nokia.com ([172.18.140.37]) by daebh001.NOE.Nokia.com with Microsoft SMTPSVC(5.0.2195.2966);
	 Wed, 2 Jan 2002 15:36:28 -0600
X-MimeOLE: Produced By Microsoft Exchange V6.0.4712.0
content-class: urn:content-classes:message
MIME-Version: 1.0
Subject: RE: [Tsvwg] sctpsocket-02.txt
Content-Type: text/plain;
	charset="iso-8859-1"
Date: Wed, 2 Jan 2002 13:36:30 -0800
Message-ID: <F0B628F30F48064289D8CCC1EE21B7A8150B0D@mvebe001.NOE.Nokia.com>
Thread-Topic: [Tsvwg] sctpsocket-02.txt
Thread-Index: AcGSCyF6cdaslP37EdWxMQAIx6TWeAByjLNw
From: "Kimble Scott (NET/MtView)" <Scott.Kimble@nokia.com>
To: "'ext Randall Stewart'" <randall@stewart.chicago.il.us>,
        "Scott Kimble" <skimble@iprg.nokia.com>
Cc: <tsvwg@ietf.org>
X-OriginalArrivalTime: 02 Jan 2002 21:36:28.0901 (UTC) FILETIME=[899A9150:01C193D5]
Content-Transfer-Encoding: 8bit
X-MIME-Autoconverted: from quoted-printable to 8bit by optimus.ietf.org id QAA23122
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 8bit


Randall,

This sounds good.  Again, I was not thinking in terms of addip-sctp.

-----Original Message-----
From: ext Randall Stewart [mailto:randall@stewart.chicago.il.us]
Sent: Monday, December 31, 2001 6:54 AM
To: Scott Kimble
Cc: tsvwg@ietf.org
Subject: Re: [Tsvwg] sctpsocket-02.txt


Scott:

In thinking on this item, I will add a note that if you
set the associ_id field to 0 you get the endpoints currently
bound set. 

You must still have a way to get the specific set on a association
since if you do NOT have the add-ip draft in place it is possible
that if I do a bind to INADDR_ANY and setup an association followed
by a ifconfig up of a new address and a subsequent association the
same endpoint will have two different associations with different
sets of bound addresses. Even if add-ip is supported the peer may
NOT support it and so you may get divergent sets of address for the
local endpoint...

R

Scott Kimble wrote:
> 
> Section 8.5 sctp_getladdrs()
> 
> The "sctp_assoc_t id" parameter is not required.  Applications may
query
> an SCTP instance prior to any association being established.
> 
> For example, since IN[6]ADDR_ANY is "optimal" (not necessarily all),
an
> application my want to call sctp_getladdrs() after calling bind
> (IN[6]ADDR_ANY) to determine what the optimal set is.
> 
> --
> Scott.Kimble@NOKIA.com
> Office 1.650.864.6587
> Mobile 1.650.743.7903
> 
> _______________________________________________
> tsvwg mailing list
> tsvwg@ietf.org
> http://www1.ietf.org/mailman/listinfo/tsvwg

-- 
Randall R. Stewart
randall@stewart.chicago.il.us 815-342-5222 (cell phone)

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@ns.ietf.org  Wed Jan  2 19:21:16 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id TAA13246
	for <tsvwg-archive@odin.ietf.org>; Wed, 2 Jan 2002 19:21:16 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id TAA27620
	for tsvwg-archive@odin.ietf.org; Wed, 2 Jan 2002 19:21:16 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id SAA26968;
	Wed, 2 Jan 2002 18:59:33 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id SAA26939
	for <tsvwg@ns.ietf.org>; Wed, 2 Jan 2002 18:59:28 -0500 (EST)
Received: from mgw-dax2.ext.nokia.com (mgw-dax2.ext.nokia.com [63.78.179.217])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id SAA13099
	for <tsvwg@ietf.org>; Wed, 2 Jan 2002 18:59:27 -0500 (EST)
Received: from davir04nok.americas.nokia.com (davir04nok.americas.nokia.com [172.18.242.87])
	by mgw-dax2.ext.nokia.com (Switch-2.1.0/Switch-2.1.0) with ESMTP id g0300pQ25353
	for <tsvwg@ietf.org>; Wed, 2 Jan 2002 18:00:51 -0600 (CST)
Received: from daebh002.NOE.Nokia.com (unverified) by davir04nok.americas.nokia.com
 (Content Technologies SMTPRS 4.2.5) with ESMTP id <T5834c3c649ac12f2570d5@davir04nok.americas.nokia.com>;
 Wed, 2 Jan 2002 17:59:27 -0600
Received: from mvebe001.NOE.Nokia.com ([172.18.140.37]) by daebh002.NOE.Nokia.com with Microsoft SMTPSVC(5.0.2195.2966);
	 Wed, 2 Jan 2002 17:59:04 -0600
X-MimeOLE: Produced By Microsoft Exchange V6.0.4712.0
content-class: urn:content-classes:message
MIME-Version: 1.0
Subject: RE: [Tsvwg] Section 4.3.1 sctpsocket-02.txt
Content-Type: text/plain;
	charset="iso-8859-1"
Date: Wed, 2 Jan 2002 15:59:06 -0800
Message-ID: <F0B628F30F48064289D8CCC1EE21B7A8250AE2@mvebe001.NOE.Nokia.com>
Thread-Topic: [Tsvwg] Section 4.3.1 sctpsocket-02.txt
Thread-Index: AcGSCjKw9bPvuv35EdWJXQAIx6TWpQBzndXA
From: "Kimble Scott (NET/MtView)" <Scott.Kimble@nokia.com>
To: "'ext Randall Stewart'" <randall@stewart.chicago.il.us>,
        "Scott Kimble" <skimble@iprg.nokia.com>
Cc: <tsvwg@ietf.org>
X-OriginalArrivalTime: 02 Jan 2002 23:59:04.0812 (UTC) FILETIME=[7552FAC0:01C193E9]
Content-Transfer-Encoding: 8bit
X-MIME-Autoconverted: from quoted-printable to 8bit by optimus.ietf.org id SAA26940
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 8bit


Here is some additional information on the backlog value.  I could not
locate a POSIX standard for listen(2).

The proposed meaning of a backlog value of 0 is reasonable.  However,
the protocol layer would now be responsible for something that has
traditionally been handled by the "socket layer".  Why not use a
negative value like the UNIX 98 standard suggests?

The UNIX 98 standard states that a backlog value less than 0 will set
the socket's listen queue to 0.  See
http://www.opengroup.org/pubs/catalog/t912.htm

sonewconn() was designed with a "fudge factor" such that listen(fd, 0)
will allow at least one connection.  See "TCP Illustrated Volume 2", p
463, or the description of sonewconn().

FreeBSD-4.2, NetBSD use the "fudge factor" algorithm.
Linux-2.2.12 sets a backlog value of 0 to 1.  And does not use the
"fudge factor".
Solais ???
HP-UX ???

-----Original Message-----
From: ext Randall Stewart [mailto:randall@stewart.chicago.il.us]
Sent: Monday, December 31, 2001 6:47 AM
To: Scott Kimble
Cc: tsvwg@ietf.org
Subject: Re: [Tsvwg] Section 4.3.1 sctpsocket-02.txt


Scott:

After some inter-author discussion. My co-authors feel that
the zero back-log is not specified in posix to have any
meaning. In other words it is an implementation choice.
As such different platforms do different things with a zero
backlog...  Thus they wanted to use this feature as a way
to turn off accepting connections.

I have no qualms about instead adding another socket option 
that enables/disables connections.. this is how it
originally was laid out... If you can convince my co-authors
we can do this.. but for now I will leave it as is :>

R

Scott Kimble wrote:
> 
> I may have missed the discussion on this issue.
> 
> Section 4.3.1 defines new meaning to a zero valued backlog parameter
to
> listen().
> 
> Historically, when the backlog parameter of listen() is zero, at least
> one connection may be established.
> 
> Is it really the intention of the SCTP Socket API's draft to change
the
> meaning of the backlog parameter when it is zero?
> 
> --
> Scott.Kimble@NOKIA.com
> Office 1.650.864.6587
> Mobile 1.650.743.7903
> 
> _______________________________________________
> tsvwg mailing list
> tsvwg@ietf.org
> http://www1.ietf.org/mailman/listinfo/tsvwg

-- 
Randall R. Stewart
randall@stewart.chicago.il.us 815-342-5222 (cell phone)

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Sat Jan  5 13:32:44 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id NAA06433
	for <tsvwg-archive@odin.ietf.org>; Sat, 5 Jan 2002 13:32:44 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id NAA17481
	for tsvwg-archive@odin.ietf.org; Sat, 5 Jan 2002 13:32:46 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id MAA15617;
	Sat, 5 Jan 2002 12:10:24 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id MAA15588
	for <tsvwg@optimus.ietf.org>; Sat, 5 Jan 2002 12:10:22 -0500 (EST)
Received: from sj-msg-core-4.cisco.com (sj-msg-core-4.cisco.com [171.71.163.10])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id MAA05884
	for <tsvwg@ietf.org>; Sat, 5 Jan 2002 12:10:18 -0500 (EST)
Received: from mira-sjc5-2.cisco.com (mira-sjc5-2.cisco.com [171.71.163.16])
	by sj-msg-core-4.cisco.com (8.11.3/8.9.1) with ESMTP id g05H9pJ27528
	for <tsvwg@ietf.org>; Sat, 5 Jan 2002 09:09:51 -0800 (PST)
Received: from stewart.chicago.il.us (ssh-sj1.cisco.com [171.68.225.134])
	by mira-sjc5-2.cisco.com (Mirapoint)
	with ESMTP id AAQ83980;
	Sat, 5 Jan 2002 09:09:50 -0800 (PST)
Message-ID: <3C37335D.58F85299@stewart.chicago.il.us>
Date: Sat, 05 Jan 2002 11:09:49 -0600
From: Randall Stewart <randall@stewart.chicago.il.us>
X-Mailer: Mozilla 4.76 [en] (X11; U; Linux 2.2.12 i386)
X-Accept-Language: en
MIME-Version: 1.0
To: TSV WG list <tsvwg@ietf.org>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Transfer-Encoding: 7bit
Subject: [Tsvwg] Submitted the long awaited c-sum draft
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit

Dear all:

I have just sent in the long awaited update to
the checksum draft. This includes new code to
generate the agreed upon checksum (CRC32c) and
touch ups in the wording as well.

It should be out within the next few days..

It will be -01 I believe.

Thanks
R
-- 
Randall R. Stewart
randall@stewart.chicago.il.us 815-342-5222 (cell phone)

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Sun Jan  6 07:15:41 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id HAA22889
	for <tsvwg-archive@odin.ietf.org>; Sun, 6 Jan 2002 07:15:41 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id HAA20535
	for tsvwg-archive@odin.ietf.org; Sun, 6 Jan 2002 07:15:44 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id GAA19937;
	Sun, 6 Jan 2002 06:57:33 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id GAA19906
	for <tsvwg@optimus.ietf.org>; Sun, 6 Jan 2002 06:57:31 -0500 (EST)
Received: from smtp.dave.sonera.fi (smtp.dave.sonera.fi [131.177.130.21])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id GAA22806
	for <tsvwg@ietf.org>; Sun, 6 Jan 2002 06:57:27 -0500 (EST)
Received: from gurtoan1-nb.etela.sonera.fi ([131.177.36.107]:2130 "HELO
        gurtoan1nb") by inside.dave.sonera.fi with SMTP id <S58684AbSAFL5S>;
        Sun, 6 Jan 2002 13:57:18 +0200
Message-ID: <038101c196a9$45f7ed40$6b24b183@etela.sonera.fi>
From: "Andrei Gurtov" <gurtov@cs.helsinki.fi>
To: "sanjay kaniyar" <skaniyar@hotmail.com>, <tsvwg@ietf.org>
Cc: <Reiner.Ludwig@Ericsson.com>
References: <F104y8hcpz5rTxM0BQX0001328b@hotmail.com>
Subject: Re: [Tsvwg] Eifel draft and usage of time-stamps.
Date:   Sun, 6 Jan 2002 13:57:10 +0200
MIME-Version: 1.0
Content-Type: text/plain;
        charset="iso-8859-1"
Content-Transfer-Encoding: 7bit
X-Priority: 3
X-MSMail-Priority: Normal
X-Mailer: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Content-Transfer-Encoding: 7bit
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit

Sanjay-

A number of conerns were raised with such approach. First, it leads to
frequent repacketization of data segment as follows. Original segments sent
without the timestamp option each carry MSS bytes of data. After RTO,
segments have the timestamp option which occupies 12 bytes leaving less
space for data. Consequently, the sequence of retransmitted segments will be
different than of original segments. In principle, this type of behavior is
legal and used by BSD systems. However, there are probably some TCPs and
PEPs around that cannot handle it well. Thus,  'be conservative in what you
send' approach may apply here.
The other concern is that the detection  may be less robust in presense of
lost segments and ACKs.

The timestamp overhead is not that big for large MTU (0.8% for 1500 B) and
ROHC is expected to compress TCP options. Additionally, there is some
evidence that timing every segment with timestamps is good on
bandwidth-limited paths. Based on these considerations, the current
understanding is to include timestamps in every segment.

Andrei

----- Original Message -----
From: "sanjay kaniyar" <skaniyar@hotmail.com>
To: <tsvwg@ietf.org>
Cc: <Reiner.Ludwig@Ericsson.com>
Sent: Tuesday, January 01, 2002 12:53 AM
Subject: [Tsvwg] Eifel draft and usage of time-stamps.


> I have seen the spurious retransmission problem occur in my experiments
with
> GPRS networks. The frequency of occurrence seems to be quite small though,
> about once or twice during the transmission of a 500 K Bytes file with an
> MSS of 1500 bytes. Ofcourse, I agree that this frequency could vary with
> different networks. In this case, however, incurring 12 bytes of overhead
in
> every segment and acknowledgemnt seems quite a bit.
>
> Since it is not necessary to timestamp every segment and acknowledgement
> when timestamp option is enabled on a connection, we could do it only when
> it is necessary. It is the sender which determines when this is necessary,
> and the following approach is backwards compatible in the sense that a
> sender taking this approach would work properly with a receiver adhering
to
> the traditional time-stamping approach and vice-versa.
>
> Sender-TCP would put a timestamp in a segment if:
> (a) PAWS protection is necessary for the connection.
> (b) An RTT measurement using timestamps is desired, and one such
measurement
> is in progress. We consider a measurement to be in progress if sender has
> sent a timestamp after it started a measurement and it is not yet echoed
> back.
> (c) Retransmission of a segment upon timeout.
>
> Receiver-TCP would put a timestamp (or echo what it got) in an ack if:
> (a) no more than 3 acks have been sent after any segment with a timestamp
> was received.
>
> An Eifel-TCP-Sender would consider an acknowledgement received after a
> timeout to be for the original segment if (a) the acknowledgemet does not
> contain a timestamp or (b) the echoed timestamp is less than the timestamp
> sent in the retransmitted segment.
>
> Thanks,
> Sanjay
>
> _________________________________________________________________
> Send and receive Hotmail on your mobile device: http://mobile.msn.com
>
>
> _______________________________________________
> tsvwg mailing list
> tsvwg@ietf.org
> http://www1.ietf.org/mailman/listinfo/tsvwg
>


_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Sun Jan  6 08:13:41 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id IAA23185
	for <tsvwg-archive@odin.ietf.org>; Sun, 6 Jan 2002 08:13:41 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id IAA21834
	for tsvwg-archive@odin.ietf.org; Sun, 6 Jan 2002 08:13:44 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id HAA21302;
	Sun, 6 Jan 2002 07:57:58 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id HAA21274
	for <tsvwg@optimus.ietf.org>; Sun, 6 Jan 2002 07:57:57 -0500 (EST)
Received: from smtp.dave.sonera.fi (smtp.dave.sonera.fi [131.177.130.21])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id HAA23109
	for <tsvwg@ietf.org>; Sun, 6 Jan 2002 07:57:52 -0500 (EST)
Received: from gurtoan1-nb.etela.sonera.fi ([131.177.36.107]:2182 "HELO
        gurtoan1nb") by inside.dave.sonera.fi with SMTP id <S58872AbSAFM5u>;
        Sun, 6 Jan 2002 14:57:50 +0200
Message-ID: <03a701c196b1$baa5f260$6b24b183@etela.sonera.fi>
From: "Andrei Gurtov" <gurtov@cs.helsinki.fi>
To: "sanjay kaniyar" <skaniyar@hotmail.com>, <tsvwg@ietf.org>
Cc: "Reiner Ludwig" <reiner.ludwig@ericsson.com>
References: <F138QPcJ0ZD948aiQdY00010656@hotmail.com>
Subject: Re: [Tsvwg] TCP Question about Fast-retransmit:
Date:   Sun, 6 Jan 2002 14:57:42 +0200
MIME-Version: 1.0
Content-Type: text/plain;
        charset="iso-8859-1"
Content-Transfer-Encoding: 7bit
X-Priority: 3
X-MSMail-Priority: Normal
X-Mailer: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Content-Transfer-Encoding: 7bit
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit

Sanjay-

Restarting the retransmit timer upon fast retransmit is a common
implementation strategy among existing TCPs. However, it's not yet
recommended in any RFC to my knowledge.

It is indeed a useful feature to decrease probability of a spurious timeout
during fast recovery. But the additional time it gives is really only time
to receive 3 DUPACKs, which is usually much  less than RTT. It happens
because the retransmit timer is restarted upon an ACK just before DUPACKs
start to arrive. In other words, the actual time when the retransmit timer
fires is RTT+RTO counting from the moment when the segment for which DUPACK
arrives had been originally transmitted.

Andrei

----- Original Message -----
From: "sanjay kaniyar" <skaniyar@hotmail.com>
To: <tsvwg@ietf.org>
Sent: Friday, December 28, 2001 1:41 AM
Subject: [Tsvwg] TCP Question about Fast-retransmit:


> Upon reception of a 3rd duplicate acknowledgement, sender TCP retransmits
> what it thinks is the missing segment and goes ahead with the rest of the
> alogorithm.
>
> In case where the RTO is closely tracking the true RTT of the connection,
it
> would invariably end up handling a retransmission timeout because - by the
> time duplicate acks are received by the sender, approximately 1 RTT would
> have expired and by the time any ack for the new data arrives, 1 more RTT
> would have expired.
>
> So, why not restart the Retransmission timer when a fast-retransmit
happens?
>
> Thanks,
> Sanjay
>
>
> _________________________________________________________________
> Join the world's largest e-mail service with MSN Hotmail.
> http://www.hotmail.com
>
>
> _______________________________________________
> tsvwg mailing list
> tsvwg@ietf.org
> http://www1.ietf.org/mailman/listinfo/tsvwg
>


_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Sun Jan  6 16:09:08 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id QAA26251
	for <tsvwg-archive@odin.ietf.org>; Sun, 6 Jan 2002 16:09:08 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id QAA02419
	for tsvwg-archive@odin.ietf.org; Sun, 6 Jan 2002 16:09:13 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id PAA01796;
	Sun, 6 Jan 2002 15:50:06 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id PAA01768
	for <tsvwg@optimus.ietf.org>; Sun, 6 Jan 2002 15:50:04 -0500 (EST)
Received: from hotmail.com (f3.law3.hotmail.com [209.185.241.3])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id PAA26128
	for <tsvwg@ietf.org>; Sun, 6 Jan 2002 15:49:58 -0500 (EST)
Received: from mail pickup service by hotmail.com with Microsoft SMTPSVC;
	 Sun, 6 Jan 2002 12:49:32 -0800
Received: from 207.46.137.9 by lw3fd.law3.hotmail.msn.com with HTTP;
	Sun, 06 Jan 2002 20:49:31 GMT
X-Originating-IP: [207.46.137.9]
From: "sanjay kaniyar" <skaniyar@hotmail.com>
To: gurtov@cs.helsinki.fi, tsvwg@ietf.org
Cc: reiner.ludwig@ericsson.com
Subject: Re: [Tsvwg] TCP Question about Fast-retransmit:
Date: Sun, 06 Jan 2002 12:49:31 -0800
Mime-Version: 1.0
Content-Type: text/plain; format=flowed
Message-ID: <F3voFOAmEaJswFCHOzy000185fa@hotmail.com>
X-OriginalArrivalTime: 06 Jan 2002 20:49:32.0182 (UTC) FILETIME=[A45C5B60:01C196F3]
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

Thanks Andrei.

I don't understand why you say the time required to receive 3 dup-acks would 
be usually much less than the RTT though. In a case where any one of the 
last 3 segments of a window sent are lost, this time would be (slightly 
greater than) 1 RTT. I guess, this is what you meant when you said it would 
be '1 RTT + 1 RTO'?

Sanjay

>From: "Andrei Gurtov" <gurtov@cs.helsinki.fi>
>To: "sanjay kaniyar" <skaniyar@hotmail.com>, <tsvwg@ietf.org>
>CC: "Reiner Ludwig" <reiner.ludwig@ericsson.com>
>Subject: Re: [Tsvwg] TCP Question about Fast-retransmit:
>Date: Sun, 6 Jan 2002 14:57:42 +0200
>
>Sanjay-
>
>Restarting the retransmit timer upon fast retransmit is a common
>implementation strategy among existing TCPs. However, it's not yet
>recommended in any RFC to my knowledge.
>
>It is indeed a useful feature to decrease probability of a spurious timeout
>during fast recovery. But the additional time it gives is really only time
>to receive 3 DUPACKs, which is usually much  less than RTT. It happens
>because the retransmit timer is restarted upon an ACK just before DUPACKs
>start to arrive. In other words, the actual time when the retransmit timer
>fires is RTT+RTO counting from the moment when the segment for which DUPACK
>arrives had been originally transmitted.
>
>Andrei
>
>----- Original Message -----
>From: "sanjay kaniyar" <skaniyar@hotmail.com>
>To: <tsvwg@ietf.org>
>Sent: Friday, December 28, 2001 1:41 AM
>Subject: [Tsvwg] TCP Question about Fast-retransmit:
>
>
> > Upon reception of a 3rd duplicate acknowledgement, sender TCP 
>retransmits
> > what it thinks is the missing segment and goes ahead with the rest of 
>the
> > alogorithm.
> >
> > In case where the RTO is closely tracking the true RTT of the 
>connection,
>it
> > would invariably end up handling a retransmission timeout because - by 
>the
> > time duplicate acks are received by the sender, approximately 1 RTT 
>would
> > have expired and by the time any ack for the new data arrives, 1 more 
>RTT
> > would have expired.
> >
> > So, why not restart the Retransmission timer when a fast-retransmit
>happens?
> >
> > Thanks,
> > Sanjay
> >
> >
> > _________________________________________________________________
> > Join the world's largest e-mail service with MSN Hotmail.
> > http://www.hotmail.com
> >
> >
> > _______________________________________________
> > tsvwg mailing list
> > tsvwg@ietf.org
> > http://www1.ietf.org/mailman/listinfo/tsvwg
> >
>




_________________________________________________________________
Chat with friends online, try MSN Messenger: http://messenger.msn.com


_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Sun Jan  6 16:56:31 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id QAA26692
	for <tsvwg-archive@odin.ietf.org>; Sun, 6 Jan 2002 16:56:31 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id QAA03249
	for tsvwg-archive@odin.ietf.org; Sun, 6 Jan 2002 16:56:36 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id QAA03132;
	Sun, 6 Jan 2002 16:35:39 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id QAA03103
	for <tsvwg@optimus.ietf.org>; Sun, 6 Jan 2002 16:35:37 -0500 (EST)
Received: from hotmail.com (f78.law3.hotmail.com [209.185.241.78])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id QAA26475
	for <tsvwg@ietf.org>; Sun, 6 Jan 2002 16:35:31 -0500 (EST)
Received: from mail pickup service by hotmail.com with Microsoft SMTPSVC;
	 Sun, 6 Jan 2002 13:35:05 -0800
Received: from 207.46.137.8 by lw3fd.law3.hotmail.msn.com with HTTP;
	Sun, 06 Jan 2002 21:35:04 GMT
X-Originating-IP: [207.46.137.8]
From: "sanjay kaniyar" <skaniyar@hotmail.com>
To: gurtov@cs.helsinki.fi, tsvwg@ietf.org
Cc: Reiner.Ludwig@ericsson.com
Subject: Re: [Tsvwg] Eifel draft and usage of time-stamps.
Date: Sun, 06 Jan 2002 13:35:04 -0800
Mime-Version: 1.0
Content-Type: text/plain; format=flowed
Message-ID: <F78tqe7sjbmDoP6j8oU0000a1eb@hotmail.com>
X-OriginalArrivalTime: 06 Jan 2002 21:35:05.0055 (UTC) FILETIME=[0147AEF0:01C196FA]
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

Thanks Andrei.

1. I would be suprised if there are TCPs/PEPs which can't handle 
repacketization. They should probably be fixed (PMTU discovery/change, SACK 
etc. comes to mind).
2. Can you elaborate why you think this scheme would be less robust?
3. The overhead would be small in (full sized) data segments but not in 
acknowledgements.
4. If the sender-TCP considers timing every segment useful, it should go 
ahead and do so.

My argument is that, there should be allowance (and specification) for using 
timestamps only when it is necessary/useful to do so (just like SACK). The 
Eifel draft seems to me like a good place to address this (as well as the 
suggestion about restarting the retransmit timer after fast-retransmit)?

Sanjay


>From: "Andrei Gurtov" <gurtov@cs.helsinki.fi>
>To: "sanjay kaniyar" <skaniyar@hotmail.com>, <tsvwg@ietf.org>
>CC: <Reiner.Ludwig@ericsson.com>
>Subject: Re: [Tsvwg] Eifel draft and usage of time-stamps.
>Date: Sun, 6 Jan 2002 13:57:10 +0200
>
>Sanjay-
>
>A number of conerns were raised with such approach. First, it leads to
>frequent repacketization of data segment as follows. Original segments sent
>without the timestamp option each carry MSS bytes of data. After RTO,
>segments have the timestamp option which occupies 12 bytes leaving less
>space for data. Consequently, the sequence of retransmitted segments will 
>be
>different than of original segments. In principle, this type of behavior is
>legal and used by BSD systems. However, there are probably some TCPs and
>PEPs around that cannot handle it well. Thus,  'be conservative in what you
>send' approach may apply here.
>The other concern is that the detection  may be less robust in presense of
>lost segments and ACKs.
>
>The timestamp overhead is not that big for large MTU (0.8% for 1500 B) and
>ROHC is expected to compress TCP options. Additionally, there is some
>evidence that timing every segment with timestamps is good on
>bandwidth-limited paths. Based on these considerations, the current
>understanding is to include timestamps in every segment.
>
>Andrei
>
>----- Original Message -----
>From: "sanjay kaniyar" <skaniyar@hotmail.com>
>To: <tsvwg@ietf.org>
>Cc: <Reiner.Ludwig@Ericsson.com>
>Sent: Tuesday, January 01, 2002 12:53 AM
>Subject: [Tsvwg] Eifel draft and usage of time-stamps.
>
>
> > I have seen the spurious retransmission problem occur in my experiments
>with
> > GPRS networks. The frequency of occurrence seems to be quite small 
>though,
> > about once or twice during the transmission of a 500 K Bytes file with 
>an
> > MSS of 1500 bytes. Ofcourse, I agree that this frequency could vary with
> > different networks. In this case, however, incurring 12 bytes of 
>overhead
>in
> > every segment and acknowledgemnt seems quite a bit.
> >
> > Since it is not necessary to timestamp every segment and acknowledgement
> > when timestamp option is enabled on a connection, we could do it only 
>when
> > it is necessary. It is the sender which determines when this is 
>necessary,
> > and the following approach is backwards compatible in the sense that a
> > sender taking this approach would work properly with a receiver adhering
>to
> > the traditional time-stamping approach and vice-versa.
> >
> > Sender-TCP would put a timestamp in a segment if:
> > (a) PAWS protection is necessary for the connection.
> > (b) An RTT measurement using timestamps is desired, and one such
>measurement
> > is in progress. We consider a measurement to be in progress if sender 
>has
> > sent a timestamp after it started a measurement and it is not yet echoed
> > back.
> > (c) Retransmission of a segment upon timeout.
> >
> > Receiver-TCP would put a timestamp (or echo what it got) in an ack if:
> > (a) no more than 3 acks have been sent after any segment with a 
>timestamp
> > was received.
> >
> > An Eifel-TCP-Sender would consider an acknowledgement received after a
> > timeout to be for the original segment if (a) the acknowledgemet does 
>not
> > contain a timestamp or (b) the echoed timestamp is less than the 
>timestamp
> > sent in the retransmitted segment.
> >
> > Thanks,
> > Sanjay
> >
> > _________________________________________________________________
> > Send and receive Hotmail on your mobile device: http://mobile.msn.com
> >
> >
> > _______________________________________________
> > tsvwg mailing list
> > tsvwg@ietf.org
> > http://www1.ietf.org/mailman/listinfo/tsvwg
> >
>
>
>_______________________________________________
>tsvwg mailing list
>tsvwg@ietf.org
>http://www1.ietf.org/mailman/listinfo/tsvwg




_________________________________________________________________
Join the world뭩 largest e-mail service with MSN Hotmail. 
http://www.hotmail.com


_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Mon Jan  7 15:17:49 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id PAA26824
	for <tsvwg-archive@odin.ietf.org>; Mon, 7 Jan 2002 15:17:49 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id PAA20037
	for tsvwg-archive@odin.ietf.org; Mon, 7 Jan 2002 15:17:51 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id OAA19214;
	Mon, 7 Jan 2002 14:56:47 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id OAA19181
	for <tsvwg@optimus.ietf.org>; Mon, 7 Jan 2002 14:56:45 -0500 (EST)
Received: from tweets.austin.ibm.com (pixpat.austin.ibm.com [192.35.232.241])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id OAA26244
	for <tsvwg@ietf.org>; Mon, 7 Jan 2002 14:56:42 -0500 (EST)
Received: from austin.ibm.com (localhost.austin.ibm.com [127.0.0.1] (may be forged)) by tweets.austin.ibm.com (AIX5.1/8.11.0/8.11.0) with ESMTP id g07JuDQ29876 for <tsvwg@ietf.org>; Mon, 7 Jan 2002 13:56:13 -0600
Message-ID: <3C39FD5D.739D2705@austin.ibm.com>
Date: Mon, 07 Jan 2002 13:56:13 -0600
From: Kavitha Baratakke <kavithab@austin.ibm.com>
Reply-To: kavithab@austin.ibm.com
Organization: IBM corporation
X-Mailer: Mozilla 4.76iC-CCK-MCD  [en_US] (X11; U; AIX 5.1)
X-Accept-Language: en
MIME-Version: 1.0
To: tsvwg@ietf.org
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Transfer-Encoding: 7bit
Subject: [Tsvwg] SCTP Heartbeat related questions
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit

I have a couple of questions related to heartbeat api:

1) When a destination transport address has been marked inactive the RFC
says that a NETWORK STATUS CHANGE notification may be invoked on the
ULP. (10.2 SCTP-to-ULP sub-section C)). However the Socket API draft
does not discuss anything about this notification. Was this missed? or
is this to be an implementor's choice?

2) Nothing has been discussed in the Socket API extentions draft, about
disabling/enabling heartbeats for specific destination transport address
from an user-level application. 

I would appreciate a response.

FYI, in the Socket API draft the SCTP_NODELAY option has been specified
as the SO_NODELAY options in the index (page 1 of the draft). Also, in
the index the SCTP_SEND_FAILED Notification is mispelt as
SCTP_SEND_FAILE

Thanks,
Kavitha 
_________________________________________________________

	Kavitha V. Baratakke
	IBM AIX TCP/IP Development Group
	11501, Burnet Road, Bldg.905,5D-018, Austin, TX - 78758
_________________________________________________________

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Mon Jan  7 15:40:43 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id PAA27500
	for <tsvwg-archive@odin.ietf.org>; Mon, 7 Jan 2002 15:40:43 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id PAA20883
	for tsvwg-archive@odin.ietf.org; Mon, 7 Jan 2002 15:40:45 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id PAA20309;
	Mon, 7 Jan 2002 15:24:31 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id PAA20280
	for <tsvwg@optimus.ietf.org>; Mon, 7 Jan 2002 15:24:29 -0500 (EST)
Received: from rwcrmhc53.attbi.com (rwcrmhc53.attbi.com [204.127.198.39])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id PAA27001
	for <tsvwg@ietf.org>; Mon, 7 Jan 2002 15:24:26 -0500 (EST)
Received: from ieee.org ([12.248.77.94]) by rwcrmhc53.attbi.com
          (InterMail vM.4.01.03.27 201-229-121-127-20010626) with ESMTP
          id <20020107202358.SKAS24940.rwcrmhc53.attbi.com@ieee.org>;
          Mon, 7 Jan 2002 20:23:58 +0000
Message-ID: <3C3A03C2.9095DECF@ieee.org>
Date: Mon, 07 Jan 2002 14:23:30 -0600
From: Peter Lei <peter.lei@ieee.org>
X-Mailer: Mozilla 4.79 [en] (Windows NT 5.0; U)
X-Accept-Language: en
MIME-Version: 1.0
To: kavithab@austin.ibm.com
CC: tsvwg@ietf.org
Subject: Re: [Tsvwg] SCTP Heartbeat related questions
References: <3C39FD5D.739D2705@austin.ibm.com>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Transfer-Encoding: 7bit
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit

Kavitha,

Kavitha Baratakke wrote:
> 
> I have a couple of questions related to heartbeat api:
> 
> 1) When a destination transport address has been marked inactive the RFC
> says that a NETWORK STATUS CHANGE notification may be invoked on the
> ULP. (10.2 SCTP-to-ULP sub-section C)). However the Socket API draft
> does not discuss anything about this notification. Was this missed? or
> is this to be an implementor's choice?

This is addressed by SCTP_PEER_ADDR_CHANGE: ADDRESS_UNREACHABLE notification
event in the draft.

> 2) Nothing has been discussed in the Socket API extentions draft, about
> disabling/enabling heartbeats for specific destination transport address
> from an user-level application.

This is addressed by the SCTP_SET_PEER_ADDR_PARAMS socket option-- setting
the heartbeat interval to 0 indicates that heartbeating should be disabled
for that peer dest address.  A non-zero value indicates the HB interval in
msec (except the special case of MAX_UINT32).

--peter


> I would appreciate a response.
> 
> FYI, in the Socket API draft the SCTP_NODELAY option has been specified
> as the SO_NODELAY options in the index (page 1 of the draft). Also, in
> the index the SCTP_SEND_FAILED Notification is mispelt as
> SCTP_SEND_FAILE
> 
> Thanks,
> Kavitha
> _________________________________________________________
> 
>         Kavitha V. Baratakke
>         IBM AIX TCP/IP Development Group
>         11501, Burnet Road, Bldg.905,5D-018, Austin, TX - 78758
> _________________________________________________________
> 
> _______________________________________________
> tsvwg mailing list
> tsvwg@ietf.org
> http://www1.ietf.org/mailman/listinfo/tsvwg

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Tue Jan  8 08:07:55 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id IAA20046
	for <tsvwg-archive@odin.ietf.org>; Tue, 8 Jan 2002 08:07:55 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id IAA27951
	for tsvwg-archive@odin.ietf.org; Tue, 8 Jan 2002 08:07:54 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id HAA27134;
	Tue, 8 Jan 2002 07:42:40 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id HAA27103
	for <tsvwg@optimus.ietf.org>; Tue, 8 Jan 2002 07:42:38 -0500 (EST)
Received: from smtp.dave.sonera.fi (smtp.dave.sonera.fi [131.177.130.21])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id HAA19862
	for <tsvwg@ietf.org>; Tue, 8 Jan 2002 07:42:36 -0500 (EST)
Received: from gurtoan1-nb.etela.sonera.fi ([131.177.36.107]:1322 "HELO
        gurtoan1nb") by inside.dave.sonera.fi with SMTP id <S61648AbSAHMmS>;
        Tue, 8 Jan 2002 14:42:18 +0200
Message-ID: <002301c19841$e7616320$6b24b183@etela.sonera.fi>
From: "Andrei Gurtov" <gurtov@cs.helsinki.fi>
To: "sanjay kaniyar" <skaniyar@hotmail.com>, <tsvwg@ietf.org>
Cc: <reiner.ludwig@ericsson.com>
References: <F3voFOAmEaJswFCHOzy000185fa@hotmail.com>
Subject: Re: [Tsvwg] TCP Question about Fast-retransmit:
Date:   Tue, 8 Jan 2002 14:42:15 +0200
MIME-Version: 1.0
Content-Type: text/plain;
        charset="iso-8859-1"
Content-Transfer-Encoding: 7bit
X-Priority: 3
X-MSMail-Priority: Normal
X-Mailer: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Content-Transfer-Encoding: 7bit
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit

Sanjay,

You right when the window and network delay is relatively small, then
receiving 3 DUPACKs takes time close to one RTT.

Figure2 of the tech report shows fast retransmit with and without restart of
the retransmit timer http://www.cs.helsinki.fi/u/gurtov/papers/report01.html

Section 3.3 of the Eifel timer paper explains the issue of REXMT=RTO+RTT
http://iceberg.cs.berkeley.edu/papers/Ludwig-Eifel-Xmit/index.html

Andrei


----- Original Message -----
From: "sanjay kaniyar" <skaniyar@hotmail.com>
To: <gurtov@cs.helsinki.fi>; <tsvwg@ietf.org>
Cc: <reiner.ludwig@ericsson.com>
Sent: Sunday, January 06, 2002 10:49 PM
Subject: Re: [Tsvwg] TCP Question about Fast-retransmit:


> Thanks Andrei.
>
> I don't understand why you say the time required to receive 3 dup-acks
would
> be usually much less than the RTT though. In a case where any one of the
> last 3 segments of a window sent are lost, this time would be (slightly
> greater than) 1 RTT. I guess, this is what you meant when you said it
would
> be '1 RTT + 1 RTO'?
>
> Sanjay
>
> >From: "Andrei Gurtov" <gurtov@cs.helsinki.fi>
> >To: "sanjay kaniyar" <skaniyar@hotmail.com>, <tsvwg@ietf.org>
> >CC: "Reiner Ludwig" <reiner.ludwig@ericsson.com>
> >Subject: Re: [Tsvwg] TCP Question about Fast-retransmit:
> >Date: Sun, 6 Jan 2002 14:57:42 +0200
> >
> >Sanjay-
> >
> >Restarting the retransmit timer upon fast retransmit is a common
> >implementation strategy among existing TCPs. However, it's not yet
> >recommended in any RFC to my knowledge.
> >
> >It is indeed a useful feature to decrease probability of a spurious
timeout
> >during fast recovery. But the additional time it gives is really only
time
> >to receive 3 DUPACKs, which is usually much  less than RTT. It happens
> >because the retransmit timer is restarted upon an ACK just before DUPACKs
> >start to arrive. In other words, the actual time when the retransmit
timer
> >fires is RTT+RTO counting from the moment when the segment for which
DUPACK
> >arrives had been originally transmitted.
> >
> >Andrei
> >
> >----- Original Message -----
> >From: "sanjay kaniyar" <skaniyar@hotmail.com>
> >To: <tsvwg@ietf.org>
> >Sent: Friday, December 28, 2001 1:41 AM
> >Subject: [Tsvwg] TCP Question about Fast-retransmit:
> >
> >
> > > Upon reception of a 3rd duplicate acknowledgement, sender TCP
> >retransmits
> > > what it thinks is the missing segment and goes ahead with the rest of
> >the
> > > alogorithm.
> > >
> > > In case where the RTO is closely tracking the true RTT of the
> >connection,
> >it
> > > would invariably end up handling a retransmission timeout because - by
> >the
> > > time duplicate acks are received by the sender, approximately 1 RTT
> >would
> > > have expired and by the time any ack for the new data arrives, 1 more
> >RTT
> > > would have expired.
> > >
> > > So, why not restart the Retransmission timer when a fast-retransmit
> >happens?
> > >
> > > Thanks,
> > > Sanjay
> > >
> > >
> > > _________________________________________________________________
> > > Join the world's largest e-mail service with MSN Hotmail.
> > > http://www.hotmail.com
> > >
> > >
> > > _______________________________________________
> > > tsvwg mailing list
> > > tsvwg@ietf.org
> > > http://www1.ietf.org/mailman/listinfo/tsvwg
> > >
> >
>
>
>
>
> _________________________________________________________________
> Chat with friends online, try MSN Messenger: http://messenger.msn.com
>


_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Tue Jan  8 09:25:20 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id JAA22895
	for <tsvwg-archive@odin.ietf.org>; Tue, 8 Jan 2002 09:25:20 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id JAA01127
	for tsvwg-archive@odin.ietf.org; Tue, 8 Jan 2002 09:25:20 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id JAA00201;
	Tue, 8 Jan 2002 09:02:05 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id JAA00168
	for <tsvwg@optimus.ietf.org>; Tue, 8 Jan 2002 09:02:03 -0500 (EST)
Received: from CNRI.Reston.VA.US (localhost [127.0.0.1])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id JAA21784;
	Tue, 8 Jan 2002 09:02:02 -0500 (EST)
Message-Id: <200201081402.JAA21784@ietf.org>
Mime-Version: 1.0
Content-Type: Multipart/Mixed; Boundary="NextPart"
To: IETF-Announce: ;
Cc: tsvwg@ietf.org
From: Internet-Drafts@ietf.org
Reply-to: Internet-Drafts@ietf.org
Date: Tue, 08 Jan 2002 09:02:02 -0500
Subject: [Tsvwg] I-D ACTION:draft-ietf-tsvwg-sctpcsum-01.txt
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

--NextPart

A New Internet-Draft is available from the on-line Internet-Drafts directories.
This draft is a work item of the Transport Area Working Group Working Group of the IETF.

	Title		: SCTP Checksum Change
	Author(s)	: R. Stewart, C. Sharp, J. Stone, D. Otis
	Filename	: draft-ietf-tsvwg-sctpcsum-01.txt
	Pages		: 10
	Date		: 07-Jan-02
	
SCTP [RFC2960] currently uses an Adler-32 checksum.  For small packets, 
this provides weak protection against the detection of errors.  This 
document changes that checksum and updates SCTP to use a 32 bit CRC 
checksum.

A URL for this Internet-Draft is:
http://www.ietf.org/internet-drafts/draft-ietf-tsvwg-sctpcsum-01.txt

To remove yourself from the IETF Announcement list, send a message to 
ietf-announce-request with the word unsubscribe in the body of the message.

Internet-Drafts are also available by anonymous FTP. Login with the username
"anonymous" and a password of your e-mail address. After logging in,
type "cd internet-drafts" and then
	"get draft-ietf-tsvwg-sctpcsum-01.txt".

A list of Internet-Drafts directories can be found in
http://www.ietf.org/shadow.html 
or ftp://ftp.ietf.org/ietf/1shadow-sites.txt


Internet-Drafts can also be obtained by e-mail.

Send a message to:
	mailserv@ietf.org.
In the body type:
	"FILE /internet-drafts/draft-ietf-tsvwg-sctpcsum-01.txt".
	
NOTE:	The mail server at ietf.org can return the document in
	MIME-encoded form by using the "mpack" utility.  To use this
	feature, insert the command "ENCODING mime" before the "FILE"
	command.  To decode the response(s), you will need "munpack" or
	a MIME-compliant mail reader.  Different MIME-compliant mail readers
	exhibit different behavior, especially when dealing with
	"multipart" MIME messages (i.e. documents which have been split
	up into multiple messages), so check your local documentation on
	how to manipulate these messages.
		
		
Below is the data which will enable a MIME compliant mail reader
implementation to automatically retrieve the ASCII version of the
Internet-Draft.

--NextPart
Content-Type: Multipart/Alternative; Boundary="OtherAccess"

--OtherAccess
Content-Type: Message/External-body;
	access-type="mail-server";
	server="mailserv@ietf.org"

Content-Type: text/plain
Content-ID:	<20020107135620.I-D@ietf.org>

ENCODING mime
FILE /internet-drafts/draft-ietf-tsvwg-sctpcsum-01.txt

--OtherAccess
Content-Type: Message/External-body;
	name="draft-ietf-tsvwg-sctpcsum-01.txt";
	site="ftp.ietf.org";
	access-type="anon-ftp";
	directory="internet-drafts"

Content-Type: text/plain
Content-ID:	<20020107135620.I-D@ietf.org>

--OtherAccess--

--NextPart--



_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Tue Jan  8 22:19:52 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id WAA17994
	for <tsvwg-archive@odin.ietf.org>; Tue, 8 Jan 2002 22:19:52 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id WAA04434
	for tsvwg-archive@odin.ietf.org; Tue, 8 Jan 2002 22:19:54 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id WAA04156;
	Tue, 8 Jan 2002 22:04:37 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id WAA04128
	for <tsvwg@optimus.ietf.org>; Tue, 8 Jan 2002 22:04:34 -0500 (EST)
Received: from viola.demizu.org (h034.p106.iij4u.or.jp [210.130.106.34])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id WAA17759
	for <tsvwg@ietf.org>; Tue, 8 Jan 2002 22:04:30 -0500 (EST)
Received: from localhost (localhost [127.0.0.1])
	by viola.demizu.org (8.11.0/8.11.0) with ESMTP id g0932RL00337;
	Wed, 9 Jan 2002 12:02:29 +0900 (JST)
From: Noritoshi Demizu <demizu@dd.iij4u.or.jp>
To: gurtov@cs.helsinki.fi
Cc: tsvwg@ietf.org
Subject: Re: [Tsvwg] TCP Question about Fast-retransmit:
In-Reply-To: Your message of "Tue, 8 Jan 2002 14:42:15 +0200"
References: <002301c19841$e7616320$6b24b183@etela.sonera.fi>
X-Mailer: Mew version 1.69 on Emacs 19.28.1 / Mule 2.3
Mime-Version: 1.0
Content-Type: Text/Plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Message-Id: <20020109120226I.demizu@dd.iij4u.or.jp>
Date: Wed, 09 Jan 2002 12:02:26 +0900
X-Dispatcher: impost version 0.99i (Apr. 6, 1997)
Lines: 29
Content-Transfer-Encoding: 7bit
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit

Andrei,

 > You right when the window and network delay is relatively small, then
 > receiving 3 DUPACKs takes time close to one RTT.

When you use Limited Transmit [RFC3042], it may take 3 RTTs in the
worst case.

  Example scenario:
  1. sent data(1). got ack(2). cwnd becomes 2MSS.
  2. sent data(2) & data(3). start rxmt timer. but data(2) lost.
  3. got ack(2) again. dupacks=1. (1RTT later from 2.)
  4. sent data(4) because of limited transmit.
  5. got ack(2) again. dupacks=2. (2RTT later from 2.)
  6. sent data(5) because of limited transmit.
  7. got ack(2) again. dupacks=3. (3RTT later from 2.)
  8. sent data(2) because of fast retransmit.
  9. got ack(6).

In the case where 3 RTT < RTO < 4 RTT, if retransmission timer is not
restarted by step 8, data(2) will be sent again without waiting for
the ACK for the packet sent by step 8.

Thinking of other scenarios with limited transmit, if retransmission
timer is not restarted on fast retransmit, spurious retransmit could
occure if RTO < 4 RTT. (RTO < 2RTT for cases without limited transmit)

Regards,
Noritoshi Demizu

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Tue Jan  8 22:35:38 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id WAA19054
	for <tsvwg-archive@odin.ietf.org>; Tue, 8 Jan 2002 22:35:37 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id WAA04984
	for tsvwg-archive@odin.ietf.org; Tue, 8 Jan 2002 22:35:40 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id WAA04479;
	Tue, 8 Jan 2002 22:22:34 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id WAA04451
	for <tsvwg@optimus.ietf.org>; Tue, 8 Jan 2002 22:22:32 -0500 (EST)
Received: from mercury.Sun.COM (mercury.Sun.COM [192.9.25.1])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id WAA18036
	for <tsvwg@ietf.org>; Tue, 8 Jan 2002 22:22:29 -0500 (EST)
Received: from engmail1.Eng.Sun.COM ([129.146.1.13])
	by mercury.Sun.COM (8.9.3+Sun/8.9.3) with ESMTP id TAA03730;
	Tue, 8 Jan 2002 19:22:30 -0800 (PST)
Received: from shield (shield.EBay.Sun.COM [129.150.113.84])
	by engmail1.Eng.Sun.COM (8.9.3+Sun/8.9.3/ENSMAIL,v2.1p1) with SMTP id TAA16974;
	Tue, 8 Jan 2002 19:22:30 -0800 (PST)
Date: Tue, 8 Jan 2002 19:21:51 -0800 (PST)
From: Kacheong Poon <kcpoon@eng.sun.com>
Reply-To: Kacheong Poon <kcpoon@eng.sun.com>
Subject: Re: [Tsvwg] Section 4.1.9 sctpsocket-02.txt
To: "Kimble Scott (NET/MtView)" <Scott.Kimble@nokia.com>
Cc: tsvwg@ietf.org
In-Reply-To: "Your message with ID" <F0B628F30F48064289D8CCC1EE21B7A8250AD9@mvebe001.NOE.Nokia.com>
Message-ID: <Roam.SIMC.2.0.6.1010546511.3145.kcpoon@eng.sun.com>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; CHARSET=US-ASCII
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

Sorry for the late reply.  I was on vacation the last month...


> Section 4.1.9 sctpsocket-02.txt
> 
> getsockname() is applicable to both TCP-Style and UDP-Style SCTP
> instances.  Section 4.1.9 can be moved to section 6.

Can you explain why?  For UDP-style socket, an fd is associated
with many associations.  And the different associations may actually
have different local info.  If getsockname() is applied to a UDP-style
socket, which associations' local info should be returned?

							K. Poon.
							kcpoon@eng.sun.com



_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Tue Jan  8 23:07:56 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id XAA19512
	for <tsvwg-archive@odin.ietf.org>; Tue, 8 Jan 2002 23:07:56 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id XAA05884
	for tsvwg-archive@odin.ietf.org; Tue, 8 Jan 2002 23:07:56 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id WAA05136;
	Tue, 8 Jan 2002 22:45:47 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id WAA05108
	for <tsvwg@optimus.ietf.org>; Tue, 8 Jan 2002 22:45:45 -0500 (EST)
Received: from patan.sun.com (patan.Sun.COM [192.18.98.43])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id WAA19328
	for <tsvwg@ietf.org>; Tue, 8 Jan 2002 22:45:43 -0500 (EST)
Received: from engmail1.Eng.Sun.COM ([129.146.1.13])
	by patan.sun.com (8.9.3+Sun/8.9.3) with ESMTP id UAA21018;
	Tue, 8 Jan 2002 20:45:27 -0700 (MST)
Received: from shield (shield.EBay.Sun.COM [129.150.113.84])
	by engmail1.Eng.Sun.COM (8.9.3+Sun/8.9.3/ENSMAIL,v2.1p1) with SMTP id TAA19556;
	Tue, 8 Jan 2002 19:45:45 -0800 (PST)
Date: Tue, 8 Jan 2002 19:45:05 -0800 (PST)
From: Kacheong Poon <kcpoon@eng.sun.com>
Reply-To: Kacheong Poon <kcpoon@eng.sun.com>
Subject: Re: [Tsvwg] Question on socket options
To: kavithab@austin.ibm.com
Cc: tsvwg@ietf.org
In-Reply-To: "Your message with ID" <3C2CA942.61097251@austin.ibm.com>
Message-ID: <Roam.SIMC.2.0.6.1010547905.18272.kcpoon@eng.sun.com>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; CHARSET=US-ASCII
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

> In the section 7 of the "Sockets API Extensions for SCTP" draft where
> various SCTP level socket options 
> are defined, options SO_LINGER, SO_NODELAY, SO_RCVBUF, SO_SNDBUF are
> also mentioned?  Shouldn't these be socket-level options (SOL_SOCKET)
> and not SCTP level?

Strictly speaking, they are transport level options.  But unfortunately,
the BSD socket API has them defined in the socket level.  Since we
want to allow current TCP apps to be ported to use SCTP without changes
(except for using a new protocol in the socket() call) if possible,
we have to follow the current socket API practise.

> Also, there is a typo on page 47, probably just a "cut n paste"
> overlook!
> 
>   /* if a send fails we want to know it */
>     if (setsockopt(fd, IPPROTO_SCTP, SCTP_RECVSENDFAILEVNT,
>                &onoff, 4) < 0) {
>       perror("setsockopt SCTP_RECVASSOCEVNT");
>       exit(1);

Thanks.

							K. Poon.
							kcpoon@eng.sun.com



_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Wed Jan  9 00:02:24 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id AAA20404
	for <tsvwg-archive@odin.ietf.org>; Wed, 9 Jan 2002 00:02:24 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id AAA07334
	for tsvwg-archive@odin.ietf.org; Wed, 9 Jan 2002 00:02:24 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id XAA06737;
	Tue, 8 Jan 2002 23:43:56 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id XAA06708
	for <tsvwg@optimus.ietf.org>; Tue, 8 Jan 2002 23:43:53 -0500 (EST)
Received: from minotaur.nge.isi.edu (minotaur.nge.isi.edu [65.114.169.202])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id XAA20207
	for <tsvwg@ietf.org>; Tue, 8 Jan 2002 23:43:51 -0500 (EST)
Received: from minotaur (mankin@localhost)
	by minotaur.nge.isi.edu (8.11.6/8.11.6) with ESMTP id g094hpA03624;
	Tue, 8 Jan 2002 23:43:51 -0500
Message-Id: <200201090443.g094hpA03624@minotaur.nge.isi.edu>
To: tsvwg@ietf.org, rrs@cisco.co, chsharp@cisco.com, jonathan@dsg.stanford.edu,
        dotis@sanlight.net
Reply-To: mankin@isi.edu
Mime-Version: 1.0 (generated by tm-edit 1.7)
Content-Type: text/plain; charset=US-ASCII
Date: Tue, 08 Jan 2002 23:43:51 -0500
From: Allison Mankin <mankin@isi.edu>
Subject: [Tsvwg] Working Group Last Call on SCTP checksum change
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

Authors and Transport Working Group:

This is the working group last call on the SCTP checksum
change draft, draft-ietf-tsvwg-sctpsum-01.txt, which has
been anounced today.  Please make comments to the working group 
list preferably (and editorial comments to the authors) by
Tuesday, January 22.

The ADs have two changes to request:

1. The draft needs to have a Security Considerations section,
   which we know the authors already noticed.  We'd like the
   text for this to be sent to the WG list so it can be WGLC'd
   too..

2. The draft should have an IANA Considerations section that
   states that there is no work for the IANA in it :)

Thanks,

Allison and Scott


_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Wed Jan  9 08:12:06 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id IAA03410
	for <tsvwg-archive@odin.ietf.org>; Wed, 9 Jan 2002 08:12:06 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id IAA00837
	for tsvwg-archive@odin.ietf.org; Wed, 9 Jan 2002 08:12:09 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id HAA29967;
	Wed, 9 Jan 2002 07:53:21 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id HAA29936
	for <tsvwg@optimus.ietf.org>; Wed, 9 Jan 2002 07:53:19 -0500 (EST)
Received: from smtp.dave.sonera.fi (smtp.dave.sonera.fi [131.177.130.21])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id HAA03203
	for <tsvwg@ietf.org>; Wed, 9 Jan 2002 07:53:16 -0500 (EST)
Received: from gurtoan1-nb.etela.sonera.fi ([131.177.36.107]:1804 "HELO
        gurtoan1nb") by inside.dave.sonera.fi with SMTP id <S65093AbSAIMw6>;
        Wed, 9 Jan 2002 14:52:58 +0200
Message-ID: <009701c1990c$8d6be400$6b24b183@etela.sonera.fi>
From: "Andrei Gurtov" <gurtov@cs.helsinki.fi>
To: "sanjay kaniyar" <skaniyar@hotmail.com>, <tsvwg@ietf.org>
Cc: <Reiner.Ludwig@ericsson.com>
References: <F78tqe7sjbmDoP6j8oU0000a1eb@hotmail.com>
Subject: Re: [Tsvwg] Eifel draft and usage of time-stamps.
Date:   Wed, 9 Jan 2002 14:52:52 +0200
MIME-Version: 1.0
Content-Type: text/plain;
        charset="iso-8859-1"
Content-Transfer-Encoding: 7bit
X-Priority: 3
X-MSMail-Priority: Normal
X-Mailer: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Content-Transfer-Encoding: 7bit
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit

Sanjay,

Thanks for your points. As you see, I don't have a strong opinion about this
issue.

> 1. I would be suprised if there are TCPs/PEPs which can't handle
> repacketization. They should probably be fixed (PMTU discovery/change,
SACK
> etc. comes to mind).

Right, but repacketization due to PMTU and SACK seems to be a rare event...

> 2. Can you elaborate why you think this scheme would be less robust?

Well, if a timestamp is put only in retransmitted segments, then lack of
timestamp is an indication of a spurious timeout. What if the receiver
implements some own bandwidth-saving policy and echos every 3rd timestamps
or so? Arguably, deliberately lying with a timestamp value is less likely.

> 3. The overhead would be small in (full sized) data segments but not in
> acknowledgements.

Good point, but ACK are themself small, sent for every-second data segment
and are ROHC compressable.

> 4. If the sender-TCP considers timing every segment useful, it should go
> ahead and do so.

Without timestamps it is  more complex to implement...

Andrei

>
> My argument is that, there should be allowance (and specification) for
using
> timestamps only when it is necessary/useful to do so (just like SACK). The
> Eifel draft seems to me like a good place to address this (as well as the
> suggestion about restarting the retransmit timer after fast-retransmit)?
>
> Sanjay
>
>
> >From: "Andrei Gurtov" <gurtov@cs.helsinki.fi>
> >To: "sanjay kaniyar" <skaniyar@hotmail.com>, <tsvwg@ietf.org>
> >CC: <Reiner.Ludwig@ericsson.com>
> >Subject: Re: [Tsvwg] Eifel draft and usage of time-stamps.
> >Date: Sun, 6 Jan 2002 13:57:10 +0200
> >
> >Sanjay-
> >
> >A number of conerns were raised with such approach. First, it leads to
> >frequent repacketization of data segment as follows. Original segments
sent
> >without the timestamp option each carry MSS bytes of data. After RTO,
> >segments have the timestamp option which occupies 12 bytes leaving less
> >space for data. Consequently, the sequence of retransmitted segments will
> >be
> >different than of original segments. In principle, this type of behavior
is
> >legal and used by BSD systems. However, there are probably some TCPs and
> >PEPs around that cannot handle it well. Thus,  'be conservative in what
you
> >send' approach may apply here.
> >The other concern is that the detection  may be less robust in presense
of
> >lost segments and ACKs.
> >
> >The timestamp overhead is not that big for large MTU (0.8% for 1500 B)
and
> >ROHC is expected to compress TCP options. Additionally, there is some
> >evidence that timing every segment with timestamps is good on
> >bandwidth-limited paths. Based on these considerations, the current
> >understanding is to include timestamps in every segment.
> >
> >Andrei
> >
> >----- Original Message -----
> >From: "sanjay kaniyar" <skaniyar@hotmail.com>
> >To: <tsvwg@ietf.org>
> >Cc: <Reiner.Ludwig@Ericsson.com>
> >Sent: Tuesday, January 01, 2002 12:53 AM
> >Subject: [Tsvwg] Eifel draft and usage of time-stamps.
> >
> >
> > > I have seen the spurious retransmission problem occur in my
experiments
> >with
> > > GPRS networks. The frequency of occurrence seems to be quite small
> >though,
> > > about once or twice during the transmission of a 500 K Bytes file with
> >an
> > > MSS of 1500 bytes. Ofcourse, I agree that this frequency could vary
with
> > > different networks. In this case, however, incurring 12 bytes of
> >overhead
> >in
> > > every segment and acknowledgemnt seems quite a bit.
> > >
> > > Since it is not necessary to timestamp every segment and
acknowledgement
> > > when timestamp option is enabled on a connection, we could do it only
> >when
> > > it is necessary. It is the sender which determines when this is
> >necessary,
> > > and the following approach is backwards compatible in the sense that a
> > > sender taking this approach would work properly with a receiver
adhering
> >to
> > > the traditional time-stamping approach and vice-versa.
> > >
> > > Sender-TCP would put a timestamp in a segment if:
> > > (a) PAWS protection is necessary for the connection.
> > > (b) An RTT measurement using timestamps is desired, and one such
> >measurement
> > > is in progress. We consider a measurement to be in progress if sender
> >has
> > > sent a timestamp after it started a measurement and it is not yet
echoed
> > > back.
> > > (c) Retransmission of a segment upon timeout.
> > >
> > > Receiver-TCP would put a timestamp (or echo what it got) in an ack if:
> > > (a) no more than 3 acks have been sent after any segment with a
> >timestamp
> > > was received.
> > >
> > > An Eifel-TCP-Sender would consider an acknowledgement received after a
> > > timeout to be for the original segment if (a) the acknowledgemet does
> >not
> > > contain a timestamp or (b) the echoed timestamp is less than the
> >timestamp
> > > sent in the retransmitted segment.
> > >
> > > Thanks,
> > > Sanjay
> > >
> > > _________________________________________________________________
> > > Send and receive Hotmail on your mobile device: http://mobile.msn.com
> > >
> > >
> > > _______________________________________________
> > > tsvwg mailing list
> > > tsvwg@ietf.org
> > > http://www1.ietf.org/mailman/listinfo/tsvwg
> > >
> >
> >
> >_______________________________________________
> >tsvwg mailing list
> >tsvwg@ietf.org
> >http://www1.ietf.org/mailman/listinfo/tsvwg
>
>
>
>
> _________________________________________________________________
> Join the world's largest e-mail service with MSN Hotmail.
> http://www.hotmail.com
>


_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Wed Jan  9 08:23:31 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id IAA03510
	for <tsvwg-archive@odin.ietf.org>; Wed, 9 Jan 2002 08:23:31 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id IAA01028
	for tsvwg-archive@odin.ietf.org; Wed, 9 Jan 2002 08:23:34 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id IAA00757;
	Wed, 9 Jan 2002 08:08:32 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id IAA00716
	for <tsvwg@optimus.ietf.org>; Wed, 9 Jan 2002 08:08:29 -0500 (EST)
Received: from smtp.dave.sonera.fi (smtp.dave.sonera.fi [131.177.130.21])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id IAA03376
	for <tsvwg@ietf.org>; Wed, 9 Jan 2002 08:08:26 -0500 (EST)
Received: from gurtoan1-nb.etela.sonera.fi ([131.177.36.107]:1817 "HELO
        gurtoan1nb") by inside.dave.sonera.fi with SMTP id <S64977AbSAINIS>;
        Wed, 9 Jan 2002 15:08:18 +0200
Message-ID: <00ab01c1990e$b14ba0c0$6b24b183@etela.sonera.fi>
From: "Andrei Gurtov" <gurtov@cs.helsinki.fi>
To: "Noritoshi Demizu" <demizu@dd.iij4u.or.jp>
Cc: <tsvwg@ietf.org>
References: <002301c19841$e7616320$6b24b183@etela.sonera.fi> <20020109120226I.demizu@dd.iij4u.or.jp>
Subject: Re: [Tsvwg] TCP Question about Fast-retransmit:
Date:   Wed, 9 Jan 2002 15:08:11 +0200
MIME-Version: 1.0
Content-Type: text/plain;
        charset="iso-8859-1"
Content-Transfer-Encoding: 7bit
X-Priority: 3
X-MSMail-Priority: Normal
X-Mailer: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Content-Transfer-Encoding: 7bit
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit

Noritoshi,

Thanks for a very interesting example!

I also observed  that LT makes a spurious timeout more likely during fast
recovery due to a longer DUPACK series.

Andrei

----- Original Message -----
From: "Noritoshi Demizu" <demizu@dd.iij4u.or.jp>
To: <gurtov@cs.helsinki.fi>
Cc: <tsvwg@ietf.org>
Sent: Wednesday, January 09, 2002 5:02 AM
Subject: Re: [Tsvwg] TCP Question about Fast-retransmit:


> Andrei,
>
>  > You right when the window and network delay is relatively small, then
>  > receiving 3 DUPACKs takes time close to one RTT.
>
> When you use Limited Transmit [RFC3042], it may take 3 RTTs in the
> worst case.
>
>   Example scenario:
>   1. sent data(1). got ack(2). cwnd becomes 2MSS.
>   2. sent data(2) & data(3). start rxmt timer. but data(2) lost.
>   3. got ack(2) again. dupacks=1. (1RTT later from 2.)
>   4. sent data(4) because of limited transmit.
>   5. got ack(2) again. dupacks=2. (2RTT later from 2.)
>   6. sent data(5) because of limited transmit.
>   7. got ack(2) again. dupacks=3. (3RTT later from 2.)
>   8. sent data(2) because of fast retransmit.
>   9. got ack(6).
>
> In the case where 3 RTT < RTO < 4 RTT, if retransmission timer is not
> restarted by step 8, data(2) will be sent again without waiting for
> the ACK for the packet sent by step 8.
>
> Thinking of other scenarios with limited transmit, if retransmission
> timer is not restarted on fast retransmit, spurious retransmit could
> occure if RTO < 4 RTT. (RTO < 2RTT for cases without limited transmit)
>
> Regards,
> Noritoshi Demizu
>


_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Wed Jan  9 13:18:23 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id NAA08817
	for <tsvwg-archive@odin.ietf.org>; Wed, 9 Jan 2002 13:18:23 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id NAA14210
	for tsvwg-archive@odin.ietf.org; Wed, 9 Jan 2002 13:18:25 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id MAA13222;
	Wed, 9 Jan 2002 12:51:12 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id MAA13191
	for <tsvwg@optimus.ietf.org>; Wed, 9 Jan 2002 12:51:10 -0500 (EST)
Received: from c003.snv.cp.net (c003-h013.c003.snv.cp.net [209.228.32.227])
	by ietf.org (8.9.1a/8.9.1a) with SMTP id MAA08198
	for <tsvwg@ietf.org>; Wed, 9 Jan 2002 12:51:08 -0500 (EST)
Received: (cpmta 1177 invoked from network); 9 Jan 2002 09:50:39 -0800
Received: from 64.130.130.105 (HELO dougrmt)
  by smtp.telocity.com (209.228.32.227) with SMTP; 9 Jan 2002 09:50:39 -0800
X-Sent: 9 Jan 2002 17:50:39 GMT
From: "Douglas Otis" <dotis@sanlight.net>
To: "Tsvwg" <tsvwg@ietf.org>
Date: Wed, 9 Jan 2002 09:50:53 -0800
Message-ID: <NEBBJGDMMLHHCIKHGBEJMEKJCOAA.dotis@sanlight.net>
MIME-Version: 1.0
Content-Type: text/plain;
	charset="iso-8859-1"
Content-Transfer-Encoding: 7bit
X-Priority: 3 (Normal)
X-MSMail-Priority: Normal
X-Mailer: Microsoft Outlook IMO, Build 9.0.2416 (9.0.2910.0)
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4807.1700
Importance: Normal
Content-Transfer-Encoding: 7bit
Subject: [Tsvwg] RE: Working Group Last Call on SCTP checksum change
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit

This could be used to update the draft to satisfy the missing sections:

3 Security Considerations

There may be a computational advantage in validating the Association against
the Verification Tag prior to performing a checksum as invalid tags will
result in the same action as a bad checksum in most cases.  The exceptions
for this technique would be INIT and some SHUTDOWN-COMPLETE exchanges as
well as a stale COOKIE-ECHO.  These special case exchanges must represent
small packets and will minimize the effect of the checksum calculation.

4 IANA Considerations

There are no IANA considerations required in this document.

Doug


_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Wed Jan  9 14:29:29 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id OAA13273
	for <tsvwg-archive@odin.ietf.org>; Wed, 9 Jan 2002 14:29:28 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id OAA16819
	for tsvwg-archive@odin.ietf.org; Wed, 9 Jan 2002 14:29:31 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id OAA16486;
	Wed, 9 Jan 2002 14:10:05 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id OAA16457
	for <tsvwg@optimus.ietf.org>; Wed, 9 Jan 2002 14:10:03 -0500 (EST)
Received: from mgw-dax1.ext.nokia.com (mgw-dax1.ext.nokia.com [63.78.179.216])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id OAA12531
	for <tsvwg@ietf.org>; Wed, 9 Jan 2002 14:10:00 -0500 (EST)
Received: from davir01nok.americas.nokia.com (davir01nok.americas.nokia.com [172.18.242.84])
	by mgw-dax1.ext.nokia.com (Switch-2.1.0/Switch-2.1.0) with ESMTP id g09JABQ16293
	for <tsvwg@ietf.org>; Wed, 9 Jan 2002 13:10:11 -0600 (CST)
Received: from daebh001.NOE.Nokia.com (unverified) by davir01nok.americas.nokia.com
 (Content Technologies SMTPRS 4.2.5) with ESMTP id <T5857c74d8eac12f254079@davir01nok.americas.nokia.com>;
 Wed, 9 Jan 2002 13:10:01 -0600
Received: from mvebe001.NOE.Nokia.com ([172.18.140.37]) by daebh001.NOE.Nokia.com with Microsoft SMTPSVC(5.0.2195.2966);
	 Wed, 9 Jan 2002 13:10:01 -0600
X-MimeOLE: Produced By Microsoft Exchange V6.0.4712.0
content-class: urn:content-classes:message
MIME-Version: 1.0
Subject: RE: [Tsvwg] I-D ACTION:draft-ietf-tsvwg-sctpcsum-01.txt
Content-Type: text/plain;
	charset="iso-8859-1"
Date: Wed, 9 Jan 2002 11:10:00 -0800
Message-ID: <4D7B558499107545BB45044C63822DDE0F8139@mvebe001.NOE.Nokia.com>
Thread-Topic: [Tsvwg] I-D ACTION:draft-ietf-tsvwg-sctpcsum-01.txt
Thread-Index: AcGYTSi12dYqpAQ+Edar0gAIx6S5QwAT9TbA
From: "Sud Shivani (NET/MtView)" <Shivani.Sud@nokia.com>
To: <dotis@sanlight.net>
Cc: <tsvwg@ietf.org>
X-OriginalArrivalTime: 09 Jan 2002 19:10:01.0089 (UTC) FILETIME=[3C8D1F10:01C19941]
Content-Transfer-Encoding: 8bit
X-MIME-Autoconverted: from quoted-printable to 8bit by optimus.ietf.org id OAA16458
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 8bit

Ref Pg 8 which has an example of the crc validation :

    /* Example of crc validation */ 
    /* Test of 32 zeros should yield 0x756EC955 placed in network order
*/ 
    /* 13 zeros followed by byte values of 1 - 0x1f should yield 
    /* 0x5b988D47 */

Do you a mean an arbitrary buffer of 32 bytes - all zero ? Or an sctp
(with a data chunk) packet with a payload of 32 bytes set to zero.
I dont see how an all zero buffer can have a non-zero checksum.
I am not getting either of the crc's  using the example code.

Shivani

!    -----Original Message-----
!    From: ext Internet-Drafts@ietf.org 
!    [mailto:Internet-Drafts@ietf.org]
!    Sent: 08 January, 2002 6:02 AM
!    Cc: tsvwg@ietf.org
!    Subject: [Tsvwg] I-D ACTION:draft-ietf-tsvwg-sctpcsum-01.txt
!    
!    
!    A New Internet-Draft is available from the on-line 
!    Internet-Drafts directories.
!    This draft is a work item of the Transport Area Working 
!    Group Working Group of the IETF.
!    
!    	Title		: SCTP Checksum Change
!    	Author(s)	: R. Stewart, C. Sharp, J. Stone, D. Otis
!    	Filename	: draft-ietf-tsvwg-sctpcsum-01.txt
!    	Pages		: 10
!    	Date		: 07-Jan-02
!    	
!    SCTP [RFC2960] currently uses an Adler-32 checksum.  For 
!    small packets, 
!    this provides weak protection against the detection of 
!    errors.  This 
!    document changes that checksum and updates SCTP to use a 
!    32 bit CRC 
!    checksum.
!    
!    A URL for this Internet-Draft is:
!    http://www.ietf.org/internet-drafts/draft-ietf-tsvwg-sctpcs
um-01.txt

To remove yourself from the IETF Announcement list, send a message to 
ietf-announce-request with the word unsubscribe in the body of the
message.

Internet-Drafts are also available by anonymous FTP. Login with the
username
"anonymous" and a password of your e-mail address. After logging in,
type "cd internet-drafts" and then
	"get draft-ietf-tsvwg-sctpcsum-01.txt".

A list of Internet-Drafts directories can be found in
http://www.ietf.org/shadow.html 
or ftp://ftp.ietf.org/ietf/1shadow-sites.txt


Internet-Drafts can also be obtained by e-mail.

Send a message to:
	mailserv@ietf.org.
In the body type:
	"FILE /internet-drafts/draft-ietf-tsvwg-sctpcsum-01.txt".
	
NOTE:	The mail server at ietf.org can return the document in
	MIME-encoded form by using the "mpack" utility.  To use this
	feature, insert the command "ENCODING mime" before the "FILE"
	command.  To decode the response(s), you will need "munpack" or
	a MIME-compliant mail reader.  Different MIME-compliant mail
readers
	exhibit different behavior, especially when dealing with
	"multipart" MIME messages (i.e. documents which have been split
	up into multiple messages), so check your local documentation on
	how to manipulate these messages.
		
		
Below is the data which will enable a MIME compliant mail reader
implementation to automatically retrieve the ASCII version of the
Internet-Draft.

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Wed Jan  9 16:15:35 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id QAA15882
	for <tsvwg-archive@odin.ietf.org>; Wed, 9 Jan 2002 16:15:35 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id QAA20254
	for tsvwg-archive@odin.ietf.org; Wed, 9 Jan 2002 16:15:38 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id QAA19807;
	Wed, 9 Jan 2002 16:00:44 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id QAA19759
	for <tsvwg@optimus.ietf.org>; Wed, 9 Jan 2002 16:00:42 -0500 (EST)
Received: from c003.snv.cp.net (c003-h016.c003.snv.cp.net [209.228.32.230])
	by ietf.org (8.9.1a/8.9.1a) with SMTP id QAA15143
	for <tsvwg@ietf.org>; Wed, 9 Jan 2002 16:00:37 -0500 (EST)
Received: (cpmta 18681 invoked from network); 9 Jan 2002 13:00:08 -0800
Received: from 64.130.130.105 (HELO dougrmt)
  by smtp.telocity.com (209.228.32.230) with SMTP; 9 Jan 2002 13:00:08 -0800
X-Sent: 9 Jan 2002 21:00:08 GMT
From: "Douglas Otis" <dotis@sanlight.net>
To: "Tsvwg" <tsvwg@ietf.org>
Subject: RE: [Tsvwg] I-D ACTION:draft-ietf-tsvwg-sctpcsum-01.txt
Date: Wed, 9 Jan 2002 13:00:23 -0800
Message-ID: <NEBBJGDMMLHHCIKHGBEJCEKOCOAA.dotis@sanlight.net>
MIME-Version: 1.0
Content-Type: text/plain;
	charset="iso-8859-1"
Content-Transfer-Encoding: 7bit
X-Priority: 3 (Normal)
X-MSMail-Priority: Normal
X-Mailer: Microsoft Outlook IMO, Build 9.0.2416 (9.0.2910.0)
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4807.1700
Importance: Normal
Content-Transfer-Encoding: 7bit
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit

Shivani,

The test was an arbitrary buffer.  The second test case was to allow an
architecture check on the data path.  The initialization of the CRC register
provides for a non-zero result even with an all zeros buffer.

Doug

> Ref Pg 8 which has an example of the crc validation :
>
>     /* Example of crc validation */
>     /* Test of 32 zeros should yield 0x756EC955 placed in network order
> */
>     /* 13 zeros followed by byte values of 1 - 0x1f should yield
>     /* 0x5b988D47 */
>
> Do you a mean an arbitrary buffer of 32 bytes - all zero ? Or an sctp
> (with a data chunk) packet with a payload of 32 bytes set to zero.
> I dont see how an all zero buffer can have a non-zero checksum.
> I am not getting either of the crc's  using the example code.
>
> Shivani
>


_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Wed Jan  9 17:52:04 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id RAA19208
	for <tsvwg-archive@odin.ietf.org>; Wed, 9 Jan 2002 17:52:04 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id RAA22980
	for tsvwg-archive@odin.ietf.org; Wed, 9 Jan 2002 17:52:07 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id RAA22826;
	Wed, 9 Jan 2002 17:37:31 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id RAA22794
	for <tsvwg@optimus.ietf.org>; Wed, 9 Jan 2002 17:37:29 -0500 (EST)
Received: from patan.sun.com (patan.Sun.COM [192.18.98.43])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id RAA18434
	for <tsvwg@ietf.org>; Wed, 9 Jan 2002 17:37:26 -0500 (EST)
Received: from engmail1.Eng.Sun.COM ([129.146.1.13])
	by patan.sun.com (8.9.3+Sun/8.9.3) with ESMTP id PAA02060;
	Wed, 9 Jan 2002 15:37:09 -0700 (MST)
Received: from shield (shield.EBay.Sun.COM [129.150.113.84])
	by engmail1.Eng.Sun.COM (8.9.3+Sun/8.9.3/ENSMAIL,v2.1p1) with SMTP id OAA28520;
	Wed, 9 Jan 2002 14:37:27 -0800 (PST)
Date: Wed, 9 Jan 2002 14:36:43 -0800 (PST)
From: Kacheong Poon <kcpoon@eng.sun.com>
Reply-To: Kacheong Poon <kcpoon@eng.sun.com>
Subject: Re: [Tsvwg] Eifel draft and usage of time-stamps.
To: sanjay kaniyar <skaniyar@hotmail.com>
Cc: tsvwg@ietf.org, Reiner.Ludwig@Ericsson.com
In-Reply-To: "Your message with ID" <F104y8hcpz5rTxM0BQX0001328b@hotmail.com>
Message-ID: <Roam.SIMC.2.0.6.1010615803.21466.kcpoon@eng.sun.com>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; CHARSET=US-ASCII
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

> Since it is not necessary to timestamp every segment and acknowledgement 
> when timestamp option is enabled on a connection, we could do it only when 
> it is necessary. It is the sender which determines when this is necessary, 
> and the following approach is backwards compatible in the sense that a 
> sender taking this approach would work properly with a receiver adhering to 
> the traditional time-stamping approach and vice-versa.

I may have missed some of the discussions in this thread....  But
I believe it is specified in RFC 1323 that if timestamp option is
used, it has to be in every TCP segments sent.  As a matter of fact,
in Solaris, if one side suddenly stops sending timestamp option,
the Solaris side will also stop sending it and will ignore future
timestamp option as it thinks that the other must have some problems
and the timestamp option may not provide correct info.  So I'm afraid
your suggestion may not be backward compatibile with RFC 1323.

							K. Poon.
							kcpoon@eng.sun.com




_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Wed Jan  9 20:11:35 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id UAA23012
	for <tsvwg-archive@odin.ietf.org>; Wed, 9 Jan 2002 20:11:35 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id UAA27023
	for tsvwg-archive@odin.ietf.org; Wed, 9 Jan 2002 20:11:37 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id TAA26371;
	Wed, 9 Jan 2002 19:53:23 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id TAA26342
	for <tsvwg@optimus.ietf.org>; Wed, 9 Jan 2002 19:53:22 -0500 (EST)
Received: from motgate3.mot.com (motgate3.mot.com [144.189.100.103])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id TAA22629
	for <tsvwg@ietf.org>; Wed, 9 Jan 2002 19:53:19 -0500 (EST)
Received: [from mothost.mot.com (mothost.mot.com [129.188.137.101]) by motgate3.mot.com (motgate3 2.1) with ESMTP id RAA18679; Wed, 9 Jan 2002 17:43:15 -0700 (MST)]
Received: [from relay2.cig.mot.com (relay2.cig.mot.com [136.182.15.24]) by mothost.mot.com (MOT-mothost 2.0) with ESMTP id RAA09627; Wed, 9 Jan 2002 17:53:20 -0700 (MST)]
Received: from quantix20.cig.mot.com (quantix20 [160.47.58.31])
	by relay2.cig.mot.com (8.11.4/8.11.4) with ESMTP id g0A0rH718332;
	Wed, 9 Jan 2002 18:53:17 -0600 (CST)
Received: from cig.mot.com (d50-38e1.cig.mot.com [160.47.56.225])
	by quantix20.cig.mot.com (8.8.8+Sun/8.8.8) with ESMTP id SAA18957;
	Wed, 9 Jan 2002 18:57:58 -0600 (CST)
Message-ID: <3C3CE66F.411905C8@cig.mot.com>
Date: Wed, 09 Jan 2002 18:55:11 -0600
From: Qiaobing Xie <xieqb@cig.mot.com>
X-Mailer: Mozilla 4.77 [en] (Win98; U)
X-Accept-Language: en
MIME-Version: 1.0
To: Douglas Otis <dotis@sanlight.net>
CC: Tsvwg <tsvwg@ietf.org>
Subject: Re: [Tsvwg] I-D ACTION:draft-ietf-tsvwg-sctpcsum-01.txt
References: <NEBBJGDMMLHHCIKHGBEJCEKOCOAA.dotis@sanlight.net>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Transfer-Encoding: 7bit
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit

An important normative reference is missing in the checksum draft -
[RFC2960] :-)

-Qiaobing

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Wed Jan  9 20:36:50 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id UAA23682
	for <tsvwg-archive@odin.ietf.org>; Wed, 9 Jan 2002 20:36:50 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id UAA27659
	for tsvwg-archive@odin.ietf.org; Wed, 9 Jan 2002 20:36:52 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id UAA27166;
	Wed, 9 Jan 2002 20:24:11 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id UAA27137
	for <tsvwg@optimus.ietf.org>; Wed, 9 Jan 2002 20:24:09 -0500 (EST)
Received: from c003.snv.cp.net (c003-h015.c003.snv.cp.net [209.228.32.229])
	by ietf.org (8.9.1a/8.9.1a) with SMTP id UAA23371
	for <tsvwg@ietf.org>; Wed, 9 Jan 2002 20:24:06 -0500 (EST)
Received: (cpmta 13635 invoked from network); 9 Jan 2002 17:23:37 -0800
Received: from 64.130.130.105 (HELO dougrmt)
  by smtp.telocity.com (209.228.32.229) with SMTP; 9 Jan 2002 17:23:37 -0800
X-Sent: 10 Jan 2002 01:23:37 GMT
From: "Douglas Otis" <dotis@sanlight.net>
To: "Qiaobing Xie" <xieqb@cig.mot.com>
Cc: "Tsvwg" <tsvwg@ietf.org>
Subject: RE: [Tsvwg] I-D ACTION:draft-ietf-tsvwg-sctpcsum-01.txt
Date: Wed, 9 Jan 2002 17:23:53 -0800
Message-ID: <NEBBJGDMMLHHCIKHGBEJGELACOAA.dotis@sanlight.net>
MIME-Version: 1.0
Content-Type: text/plain;
	charset="us-ascii"
Content-Transfer-Encoding: 7bit
X-Priority: 3 (Normal)
X-MSMail-Priority: Normal
X-Mailer: Microsoft Outlook IMO, Build 9.0.2416 (9.0.2910.0)
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4807.1700
In-Reply-To: <3C3CE66F.411905C8@cig.mot.com>
Importance: Normal
Content-Transfer-Encoding: 7bit
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit

Qiaobing,

In the Reference Section add:

[RFC2960] R. R. Stewart, Q. Xie, K. Morneault, C. Sharp, H. J. Schwarzbauer,
T. Taylor, I. Rytina, M. Kalla, L. Zhang, and, V. Paxson, "Stream Control
Transmission Protocol," RFC 2960, October 2000.


Doug

> An important normative reference is missing in the checksum draft -
> [RFC2960] :-)
>
> -Qiaobing


_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Wed Jan  9 21:24:30 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id VAA24996
	for <tsvwg-archive@odin.ietf.org>; Wed, 9 Jan 2002 21:24:30 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id VAA28830
	for tsvwg-archive@odin.ietf.org; Wed, 9 Jan 2002 21:24:33 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id UAA27968;
	Wed, 9 Jan 2002 20:56:41 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id UAA27938
	for <tsvwg@optimus.ietf.org>; Wed, 9 Jan 2002 20:56:39 -0500 (EST)
Received: from hotmail.com (f143.law3.hotmail.com [209.185.241.143])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id UAA24199
	for <tsvwg@ietf.org>; Wed, 9 Jan 2002 20:56:35 -0500 (EST)
Received: from mail pickup service by hotmail.com with Microsoft SMTPSVC;
	 Wed, 9 Jan 2002 17:56:08 -0800
Received: from 207.46.137.9 by lw3fd.law3.hotmail.msn.com with HTTP;	Thu, 10 Jan 2002 01:56:07 GMT
X-Originating-IP: [207.46.137.9]
From: "sanjay kaniyar" <skaniyar@hotmail.com>
To: kcpoon@eng.sun.com
Cc: tsvwg@ietf.org, Reiner.Ludwig@ericsson.com
Subject: Re: [Tsvwg] Eifel draft and usage of time-stamps.
Date: Wed, 09 Jan 2002 17:56:07 -0800
Mime-Version: 1.0
Content-Type: text/plain; format=flowed
Message-ID: <F143KpjE5aHaZVrDs9a00000612@hotmail.com>
X-OriginalArrivalTime: 10 Jan 2002 01:56:08.0221 (UTC) FILETIME=[F87E64D0:01C19979]
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

I browsed through 1323 but could not find it speicify that timestamps should 
be sent in every segment, once negotiated. However, I did find the 
following:

"For simplicity and symmetry, we specify that timestamps always be sent and 
echoed in both directions."

Is this what you were referring to? Not sure it implies what you said.

Sanjay


>From: Kacheong Poon <kcpoon@eng.sun.com>
>Reply-To: Kacheong Poon <kcpoon@eng.sun.com>
>To: sanjay kaniyar <skaniyar@hotmail.com>
>CC: tsvwg@ietf.org, Reiner.Ludwig@ericsson.com
>Subject: Re: [Tsvwg] Eifel draft and usage of time-stamps.
>Date: Wed, 9 Jan 2002 14:36:43 -0800 (PST)
>
> > Since it is not necessary to timestamp every segment and acknowledgement
> > when timestamp option is enabled on a connection, we could do it only 
>when
> > it is necessary. It is the sender which determines when this is 
>necessary,
> > and the following approach is backwards compatible in the sense that a
> > sender taking this approach would work properly with a receiver adhering 
>to
> > the traditional time-stamping approach and vice-versa.
>
>I may have missed some of the discussions in this thread....  But
>I believe it is specified in RFC 1323 that if timestamp option is
>used, it has to be in every TCP segments sent.  As a matter of fact,
>in Solaris, if one side suddenly stops sending timestamp option,
>the Solaris side will also stop sending it and will ignore future
>timestamp option as it thinks that the other must have some problems
>and the timestamp option may not provide correct info.  So I'm afraid
>your suggestion may not be backward compatibile with RFC 1323.
>
>							K. Poon.
>							kcpoon@eng.sun.com
>
>
>
>
>_______________________________________________
>tsvwg mailing list
>tsvwg@ietf.org
>http://www1.ietf.org/mailman/listinfo/tsvwg




_________________________________________________________________
Chat with friends online, try MSN Messenger: http://messenger.msn.com


_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Wed Jan  9 21:53:59 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id VAA26730
	for <tsvwg-archive@odin.ietf.org>; Wed, 9 Jan 2002 21:53:59 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id VAA29679
	for tsvwg-archive@odin.ietf.org; Wed, 9 Jan 2002 21:54:02 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id VAA28867;
	Wed, 9 Jan 2002 21:25:23 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id VAA28842
	for <tsvwg@optimus.ietf.org>; Wed, 9 Jan 2002 21:25:22 -0500 (EST)
Received: from mercury.Sun.COM (mercury.Sun.COM [192.9.25.1])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id VAA25090
	for <tsvwg@ietf.org>; Wed, 9 Jan 2002 21:25:18 -0500 (EST)
Received: from engmail1.Eng.Sun.COM ([129.146.1.13])
	by mercury.Sun.COM (8.9.3+Sun/8.9.3) with ESMTP id SAA18514;
	Wed, 9 Jan 2002 18:25:18 -0800 (PST)
Received: from shield (shield.EBay.Sun.COM [129.150.113.84])
	by engmail1.Eng.Sun.COM (8.9.3+Sun/8.9.3/ENSMAIL,v2.1p1) with SMTP id SAA05218;
	Wed, 9 Jan 2002 18:25:17 -0800 (PST)
Date: Wed, 9 Jan 2002 18:24:37 -0800 (PST)
From: Kacheong Poon <kcpoon@eng.sun.com>
Reply-To: Kacheong Poon <kcpoon@eng.sun.com>
Subject: Re: [Tsvwg] Eifel draft and usage of time-stamps.
To: sanjay kaniyar <skaniyar@hotmail.com>
Cc: tsvwg@ietf.org, Reiner.Ludwig@ericsson.com
In-Reply-To: "Your message with ID" <F143KpjE5aHaZVrDs9a00000612@hotmail.com>
Message-ID: <Roam.SIMC.2.0.6.1010629477.7364.kcpoon@eng.sun.com>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; CHARSET=US-ASCII
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

> I browsed through 1323 but could not find it speicify that timestamps should 
> be sent in every segment, once negotiated. However, I did find the  >
following: > 
> "For simplicity and symmetry, we specify that timestamps always be sent and 
> echoed in both directions."
> 
> Is this what you were referring to? Not sure it implies what you said.

Yes.  Well, I intrepret always to be for every segments...  How do you
interpret that?  And I believe all TCP implementations I have encountered
always send a timestamp option in every segments.  As a matter of fact,
I am not sure if the RTTM or PAWS as specified work correctly if timestamp
option is not sent and echoed in every segments...

							K. Poon.
							kcpoon@eng.sun.com



_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Wed Jan  9 22:02:08 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id WAA26875
	for <tsvwg-archive@odin.ietf.org>; Wed, 9 Jan 2002 22:02:08 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id WAA00123
	for tsvwg-archive@odin.ietf.org; Wed, 9 Jan 2002 22:02:12 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id VAA29608;
	Wed, 9 Jan 2002 21:47:34 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id VAA29580
	for <tsvwg@optimus.ietf.org>; Wed, 9 Jan 2002 21:47:32 -0500 (EST)
Received: from hotmail.com (f100.law3.hotmail.com [209.185.241.100])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id VAA26585
	for <tsvwg@ietf.org>; Wed, 9 Jan 2002 21:47:28 -0500 (EST)
Received: from mail pickup service by hotmail.com with Microsoft SMTPSVC;
	 Wed, 9 Jan 2002 18:47:00 -0800
Received: from 131.107.3.92 by lw3fd.law3.hotmail.msn.com with HTTP;
	Thu, 10 Jan 2002 02:47:00 GMT
X-Originating-IP: [131.107.3.92]
From: "sanjay kaniyar" <skaniyar@hotmail.com>
To: kcpoon@eng.sun.com
Cc: tsvwg@ietf.org, Reiner.Ludwig@ericsson.com
Subject: Re: [Tsvwg] Eifel draft and usage of time-stamps.
Date: Wed, 09 Jan 2002 18:47:00 -0800
Mime-Version: 1.0
Content-Type: text/plain; format=flowed
Message-ID: <F100RjbDt3HvoFOAmEa0001d096@hotmail.com>
X-OriginalArrivalTime: 10 Jan 2002 02:47:00.0924 (UTC) FILETIME=[140BE7C0:01C19981]
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

I interpret it as 'don't use timestmaps in a unidirectional way'.

Besides, I don't know of a TCP which would reject segments it receives on a 
timestamp negotiated connection just because it does not have timestamps. 
From what you said, Solaris also accepts such segments. The rest of the 
interpretation seems to be Solaris specific.

As I suggested earlier, if PAWS check is necessary (easy to figure out for 
the sender), every segment should be timestamped. For RTTM, timestamping and 
taking an RTT measurement on every segment is really an overkill in a lot of 
cases.

In my opinion, the primary reason why TCP-clients do not negotiate a 
timestamp option today is the constant overhead in terms of bandwidth it 
causes. If this were to be on a need only basis (like SACK), it would 
probably be more widely negotiated and it would enable wide spread 
deployment of Eifel-capable TCPs (probably, every mobile node would 
negotiate timestamps!).

Sanjay


>From: Kacheong Poon <kcpoon@eng.sun.com>
>Reply-To: Kacheong Poon <kcpoon@eng.sun.com>
>To: sanjay kaniyar <skaniyar@hotmail.com>
>CC: tsvwg@ietf.org, Reiner.Ludwig@ericsson.com
>Subject: Re: [Tsvwg] Eifel draft and usage of time-stamps.
>Date: Wed, 9 Jan 2002 18:24:37 -0800 (PST)
>
> > I browsed through 1323 but could not find it speicify that timestamps 
>should
> > be sent in every segment, once negotiated. However, I did find the  >
>following: >
> > "For simplicity and symmetry, we specify that timestamps always be sent 
>and
> > echoed in both directions."
> >
> > Is this what you were referring to? Not sure it implies what you said.
>
>Yes.  Well, I intrepret always to be for every segments...  How do you
>interpret that?  And I believe all TCP implementations I have encountered
>always send a timestamp option in every segments.  As a matter of fact,
>I am not sure if the RTTM or PAWS as specified work correctly if timestamp
>option is not sent and echoed in every segments...
>
>							K. Poon.
>							kcpoon@eng.sun.com
>
>




_________________________________________________________________
Get your FREE download of MSN Explorer at http://explorer.msn.com/intl.asp.


_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Wed Jan  9 22:29:17 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id WAA27360
	for <tsvwg-archive@odin.ietf.org>; Wed, 9 Jan 2002 22:29:17 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id WAA00618
	for tsvwg-archive@odin.ietf.org; Wed, 9 Jan 2002 22:29:20 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id WAA00299;
	Wed, 9 Jan 2002 22:13:34 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id WAA00267
	for <tsvwg@optimus.ietf.org>; Wed, 9 Jan 2002 22:13:32 -0500 (EST)
Received: from patan.sun.com (patan.Sun.COM [192.18.98.43])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id WAA27159
	for <tsvwg@ietf.org>; Wed, 9 Jan 2002 22:13:27 -0500 (EST)
Received: from engmail1.Eng.Sun.COM ([129.146.1.13])
	by patan.sun.com (8.9.3+Sun/8.9.3) with ESMTP id UAA15009;
	Wed, 9 Jan 2002 20:13:11 -0700 (MST)
Received: from shield (shield.EBay.Sun.COM [129.150.113.84])
	by engmail1.Eng.Sun.COM (8.9.3+Sun/8.9.3/ENSMAIL,v2.1p1) with SMTP id TAA14392;
	Wed, 9 Jan 2002 19:13:28 -0800 (PST)
Date: Wed, 9 Jan 2002 19:12:49 -0800 (PST)
From: Kacheong Poon <kcpoon@eng.sun.com>
Reply-To: Kacheong Poon <kcpoon@eng.sun.com>
Subject: Re: [Tsvwg] Eifel draft and usage of time-stamps.
To: sanjay kaniyar <skaniyar@hotmail.com>
Cc: tsvwg@ietf.org, Reiner.Ludwig@ericsson.com
In-Reply-To: "Your message with ID" <F100RjbDt3HvoFOAmEa0001d096@hotmail.com>
Message-ID: <Roam.SIMC.2.0.6.1010632369.25582.kcpoon@eng.sun.com>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; CHARSET=US-ASCII
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

> I interpret it as 'don't use timestmaps in a unidirectional way'.

Hmmm, a TCP connection is always bidirectional.  And the option has
to be negotiated by both sides.  What do you really mean by  
"unidirectional?"

> Besides, I don't know of a TCP which would reject segments it receives on a 
> timestamp negotiated connection just because it does not have timestamps. 
> From what you said, Solaris also accepts such segments. The rest of the 
> interpretation seems to be Solaris specific.

It seems that you misinterpreted my comment.  My comment is about
ignoring the timestamp option, not rejecting segments.  This is
a major difference.  And the other comment is that I don't know of
any implementation which does not send timestamp option in every
segment (besides, for example, RST) once it is negotiated at the
beginning.  And for the sake of the RTTM and PAWS, I believe it
is safe to assume that if one side suddenly stops sending timestamp
option, there must be something wrong and RTTM and PAWS will not
be reliable.

> As I suggested earlier, if PAWS check is necessary (easy to figure out for 
> the sender), every segment should be timestamped. For RTTM, timestamping and 
> taking an RTT measurement on every segment is really an overkill in a lot of 
> cases.

Hmmm, when you say it is easy to figure out whether PAWS is needed,
can you explain further?  Nowadays, I believe most computers have
100Mbps NIC (I believe soon all gadgets with network connection will
be like that).  RFC 1323 clearly states the problem here.  So are
you saying that all connections should use PAWS then?  I believe
you are not suggesting that.  Maybe you can explain your easy method.
And can you also explain how you can tell taking an RTT measurement on
every segment is an overkill at the beginning of a connection?

> In my opinion, the primary reason why TCP-clients do not negotiate a 
> timestamp option today is the constant overhead in terms of bandwidth it 
> causes. If this were to be on a need only basis (like SACK), it would 
> probably be more widely negotiated and it would enable wide spread 
> deployment of Eifel-capable TCPs (probably, every mobile node would 
> negotiate timestamps!).

Well, assuming that timestamp option is really rarely used nowadays,
I believe there can be many other reasons...  Not all stacks support
timestamp option (there are still many old stacks out there...).  
The option is not turned on be default. ...  Honestly, given today's
network and ROHC for slow links, the 12 bytes overhead is really nothing
for most people...  This is my own conjecture without support.  If
you can substantiate your own overhead conjecture, I'll buy it (-:

							K. Poon.
							kcpoon@eng.sun.com



_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Wed Jan  9 23:15:46 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id XAA29139
	for <tsvwg-archive@odin.ietf.org>; Wed, 9 Jan 2002 23:15:45 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id XAA02046
	for tsvwg-archive@odin.ietf.org; Wed, 9 Jan 2002 23:15:49 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id XAA01690;
	Wed, 9 Jan 2002 23:00:30 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id XAA01628
	for <tsvwg@optimus.ietf.org>; Wed, 9 Jan 2002 23:00:26 -0500 (EST)
Received: from hotmail.com (f127.law3.hotmail.com [209.185.241.127])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id XAA28865
	for <tsvwg@ietf.org>; Wed, 9 Jan 2002 23:00:21 -0500 (EST)
Received: from mail pickup service by hotmail.com with Microsoft SMTPSVC;
	 Wed, 9 Jan 2002 19:59:54 -0800
Received: from 207.46.137.8 by lw3fd.law3.hotmail.msn.com with HTTP;
	Thu, 10 Jan 2002 03:59:54 GMT
X-Originating-IP: [207.46.137.8]
From: "sanjay kaniyar" <skaniyar@hotmail.com>
To: kcpoon@eng.sun.com
Cc: tsvwg@ietf.org, Reiner.Ludwig@ericsson.com
Subject: Re: [Tsvwg] Eifel draft and usage of time-stamps.
Date: Wed, 09 Jan 2002 19:59:54 -0800
Mime-Version: 1.0
Content-Type: text/plain; format=flowed
Message-ID: <F127naTbyyw8whaJqbN0001d39e@hotmail.com>
X-OriginalArrivalTime: 10 Jan 2002 03:59:54.0758 (UTC) FILETIME=[430E0260:01C1998B]
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

1. Data transfer could be unidirectional, in which case timestamps would 
only be necessary for the direction of data flow. Somewhere in the 1323, 
they mention that they chose to 'always' use it bidirectionally for the 
purpose of simplicity etc.

2. I did not misinterpret your comment, I understood that Solaris does not 
reject segments without timestamps. My point is that, the behavior of 
'considering the peer to be not able to generate timestamps accurately just 
because some segments did not have timestamps' is what is Solaris speicific.

3. It may not be necessary to time every segment in cases like, for example, 
where it does not change much at all? or maybe timing every 3rd segment is 
just as good? or maybe it is too small and the sender TCP is going to stick 
to its attainable minimum?

4. It is possible to say a connection does not need PAWS check based on the 
perceived RTT and the advertised receive window size, for example. Or if 
there is a limit set on the maximum bandwidth a connection can use etc.

5. Whether 12 bytes in each segment and ack is an overhead or not (with ROHC 
or without) is quite subjective, I think. To me, a single byte that goes on 
the wire that does not have any purpose to it is an overhead and should be 
avoided if it is not too much of an engineering overhead (and I believe with 
timestamps it is not).

Sanjay.

>From: Kacheong Poon <kcpoon@eng.sun.com>
>Reply-To: Kacheong Poon <kcpoon@eng.sun.com>
>To: sanjay kaniyar <skaniyar@hotmail.com>
>CC: tsvwg@ietf.org, Reiner.Ludwig@ericsson.com
>Subject: Re: [Tsvwg] Eifel draft and usage of time-stamps.
>Date: Wed, 9 Jan 2002 19:12:49 -0800 (PST)
>
> > I interpret it as 'don't use timestmaps in a unidirectional way'.
>
>Hmmm, a TCP connection is always bidirectional.  And the option has
>to be negotiated by both sides.  What do you really mean by
>"unidirectional?"
>
> > Besides, I don't know of a TCP which would reject segments it receives 
>on a
> > timestamp negotiated connection just because it does not have 
>timestamps.
> > From what you said, Solaris also accepts such segments. The rest of the
> > interpretation seems to be Solaris specific.
>
>It seems that you misinterpreted my comment.  My comment is about
>ignoring the timestamp option, not rejecting segments.  This is
>a major difference.  And the other comment is that I don't know of
>any implementation which does not send timestamp option in every
>segment (besides, for example, RST) once it is negotiated at the
>beginning.  And for the sake of the RTTM and PAWS, I believe it
>is safe to assume that if one side suddenly stops sending timestamp
>option, there must be something wrong and RTTM and PAWS will not
>be reliable.
>
> > As I suggested earlier, if PAWS check is necessary (easy to figure out 
>for
> > the sender), every segment should be timestamped. For RTTM, timestamping 
>and
> > taking an RTT measurement on every segment is really an overkill in a 
>lot of
> > cases.
>
>Hmmm, when you say it is easy to figure out whether PAWS is needed,
>can you explain further?  Nowadays, I believe most computers have
>100Mbps NIC (I believe soon all gadgets with network connection will
>be like that).  RFC 1323 clearly states the problem here.  So are
>you saying that all connections should use PAWS then?  I believe
>you are not suggesting that.  Maybe you can explain your easy method.
>And can you also explain how you can tell taking an RTT measurement on
>every segment is an overkill at the beginning of a connection?
>
> > In my opinion, the primary reason why TCP-clients do not negotiate a
> > timestamp option today is the constant overhead in terms of bandwidth it
> > causes. If this were to be on a need only basis (like SACK), it would
> > probably be more widely negotiated and it would enable wide spread
> > deployment of Eifel-capable TCPs (probably, every mobile node would
> > negotiate timestamps!).
>
>Well, assuming that timestamp option is really rarely used nowadays,
>I believe there can be many other reasons...  Not all stacks support
>timestamp option (there are still many old stacks out there...).
>The option is not turned on be default. ...  Honestly, given today's
>network and ROHC for slow links, the 12 bytes overhead is really nothing
>for most people...  This is my own conjecture without support.  If
>you can substantiate your own overhead conjecture, I'll buy it (-:
>
>							K. Poon.
>							kcpoon@eng.sun.com
>
>




_________________________________________________________________
Send and receive Hotmail on your mobile device: http://mobile.msn.com


_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Thu Jan 10 07:51:56 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id HAA13769
	for <tsvwg-archive@odin.ietf.org>; Thu, 10 Jan 2002 07:51:56 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id HAA25085
	for tsvwg-archive@odin.ietf.org; Thu, 10 Jan 2002 07:51:58 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id HAA24046;
	Thu, 10 Jan 2002 07:13:00 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id HAA24017
	for <tsvwg@optimus.ietf.org>; Thu, 10 Jan 2002 07:12:56 -0500 (EST)
Received: from sj-msg-core-1.cisco.com (sj-msg-core-1.cisco.com [171.71.163.11])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id HAA13426
	for <tsvwg@ietf.org>; Thu, 10 Jan 2002 07:12:54 -0500 (EST)
Received: from mira-sjc5-2.cisco.com (mira-sjc5-2.cisco.com [171.71.163.16])
	by sj-msg-core-1.cisco.com (8.11.3/8.9.1) with ESMTP id g0ACCOF25798;
	Thu, 10 Jan 2002 04:12:24 -0800 (PST)
Received: from stewart.chicago.il.us (ssh-sj1.cisco.com [171.68.225.134])
	by mira-sjc5-2.cisco.com (Mirapoint)
	with ESMTP id AAR77148;
	Thu, 10 Jan 2002 04:12:23 -0800 (PST)
Message-ID: <3C3D8527.E7B4B0BB@stewart.chicago.il.us>
Date: Thu, 10 Jan 2002 06:12:23 -0600
From: Randall Stewart <randall@stewart.chicago.il.us>
X-Mailer: Mozilla 4.76 [en] (X11; U; Linux 2.2.12 i386)
X-Accept-Language: en
MIME-Version: 1.0
To: Qiaobing Xie <xieqb@cig.mot.com>
CC: Douglas Otis <dotis@sanlight.net>, Tsvwg <tsvwg@ietf.org>
Subject: Re: [Tsvwg] I-D ACTION:draft-ietf-tsvwg-sctpcsum-01.txt
References: <NEBBJGDMMLHHCIKHGBEJCEKOCOAA.dotis@sanlight.net> <3C3CE66F.411905C8@cig.mot.com>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Transfer-Encoding: 7bit
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit

Qiaobing Xie wrote:
> 
> An important normative reference is missing in the checksum draft -
> [RFC2960] :-)
> 
> -Qiaobing
> 
> _______________________________________________
> tsvwg mailing list
> tsvwg@ietf.org
> http://www1.ietf.org/mailman/listinfo/tsvwg

Thanks Q I already caught that after Scott pointed
out to me the missing RFC2129 reference as well :-0
(Plus the draft identifies itself as RFC2960 ... cut
 and paste errors that creep in are so much fun :-)

R
-- 
Randall R. Stewart
randall@stewart.chicago.il.us 815-342-5222 (cell phone)

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Thu Jan 10 08:02:55 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id IAA13887
	for <tsvwg-archive@odin.ietf.org>; Thu, 10 Jan 2002 08:02:55 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id IAA25657
	for tsvwg-archive@odin.ietf.org; Thu, 10 Jan 2002 08:02:56 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id HAA24828;
	Thu, 10 Jan 2002 07:33:59 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id HAA24793
	for <tsvwg@optimus.ietf.org>; Thu, 10 Jan 2002 07:33:57 -0500 (EST)
Received: from lawyers.grc.nasa.gov (d254.as0.clev.oh.voyager.net [216.214.12.1])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id HAA13691
	for <tsvwg@ietf.org>; Thu, 10 Jan 2002 07:33:54 -0500 (EST)
Received: from lawyers (mallman@localhost)
	by lawyers.grc.nasa.gov (8.9.3/8.9.3) with ESMTP id HAA32354;
	Thu, 10 Jan 2002 07:33:02 -0500
Message-Id: <200201101233.HAA32354@lawyers.grc.nasa.gov>
To: "sanjay kaniyar" <skaniyar@hotmail.com>
From: Mark Allman <mallman@grc.nasa.gov>
Reply-To: mallman@grc.nasa.gov
cc: gurtov@cs.helsinki.fi, tsvwg@ietf.org, Reiner.Ludwig@ericsson.com
Subject: Re: [Tsvwg] Eifel draft and usage of time-stamps. 
Organization: BBN Technologies/NASA GRC
Song-of-the-Day: Who Are You
Date: Thu, 10 Jan 2002 07:33:02 -0500
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org


Two points...

Repacketization seems like it will be a problem.  In other words,
adding 12 bytes to a retransmit turns a single packet retransmission
into two packets worth of data when sending full sized segments.
While this should not be a problem in principle, my guess is that it
would end up being quite an ugly situation (because hosts would be
sending too many segments, runt segments, etc. and because middle
boxes may puke).  If we want a signaling method for retransmits, I
say we revert to the bit approach.

> My argument is that, there should be allowance (and specification)
> for using timestamps only when it is necessary/useful to do so
> (just like SACK). The Eifel draft seems to me like a good place to
> address this (as well as the= suggestion about restarting the
> retransmit timer after fast-retransmit)?

I agree that there could well be cases for enabling timestamps on
the fly (e.g., enable them for all connections, but only use them
once we are sending fast enough to need PAWS).  However, I disagree
completely with the assertion that the eifel draft is the right
place to make that change.  Eifel is simply an algorithm to detect
spurious timeouts.  Let's not throw together everything in the one
draft.

OK, I lied... One more point...

If you (a) don't want to incur the overhead of timestamps and (b)
want to detect spurious retransmits, you *do* have another option.
You can use DSACK to detect spurious retransmits.  For instance,
there is an algorithm sketched in draft-blanton-dsack-use-01.txt
that could be used.  It is not nearly as robust as eifel, but my hit
is that it could work pretty well in lots of cases.

allman


---
Mark Allman -- NASA GRC/BBN -- http://roland.grc.nasa.gov/~mallman/

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Thu Jan 10 08:29:05 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id IAA14113
	for <tsvwg-archive@odin.ietf.org>; Thu, 10 Jan 2002 08:29:00 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id IAA26081
	for tsvwg-archive@odin.ietf.org; Thu, 10 Jan 2002 08:29:00 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id IAA25760;
	Thu, 10 Jan 2002 08:08:22 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id IAA25727
	for <tsvwg@optimus.ietf.org>; Thu, 10 Jan 2002 08:08:20 -0500 (EST)
Received: from sj-msg-core-3.cisco.com (sj-msg-core-3.cisco.com [171.70.157.152])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id IAA13913
	for <tsvwg@ietf.org>; Thu, 10 Jan 2002 08:08:17 -0500 (EST)
Received: from mira-sjc5-2.cisco.com (mira-sjc5-2.cisco.com [171.71.163.16])
	by sj-msg-core-3.cisco.com (8.11.3/8.9.1) with ESMTP id g0AD7Ok19490;
	Thu, 10 Jan 2002 05:07:24 -0800 (PST)
Received: from stewart.chicago.il.us (ssh-sj1.cisco.com [171.68.225.134])
	by mira-sjc5-2.cisco.com (Mirapoint)
	with ESMTP id AAR77485;
	Thu, 10 Jan 2002 05:07:46 -0800 (PST)
Message-ID: <3C3D9221.F6B20665@stewart.chicago.il.us>
Date: Thu, 10 Jan 2002 07:07:46 -0600
From: Randall Stewart <randall@stewart.chicago.il.us>
X-Mailer: Mozilla 4.76 [en] (X11; U; Linux 2.2.12 i386)
X-Accept-Language: en
MIME-Version: 1.0
To: Kacheong Poon <kcpoon@eng.sun.com>
CC: "Kimble Scott (NET/MtView)" <Scott.Kimble@nokia.com>, tsvwg@ietf.org
Subject: Re: [Tsvwg] Section 4.1.9 sctpsocket-02.txt
References: <Roam.SIMC.2.0.6.1010546511.3145.kcpoon@eng.sun.com>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Transfer-Encoding: 7bit
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit

Kacheong:

Sorry for my delayed answer... 

I have taken Scott's advise for our next pass since
getsockname() returns the LOCAL address of the endpoint. I 
believe this works on both TCP and UPD sockets in most
API's given back the bound endpoint address. 

The one that gets the peer address would be one that
we would need to have just for the TCP model...

R

Kacheong Poon wrote:
> 
> Sorry for the late reply.  I was on vacation the last month...
> 
> > Section 4.1.9 sctpsocket-02.txt
> >
> > getsockname() is applicable to both TCP-Style and UDP-Style SCTP
> > instances.  Section 4.1.9 can be moved to section 6.
> 
> Can you explain why?  For UDP-style socket, an fd is associated
> with many associations.  And the different associations may actually
> have different local info.  If getsockname() is applied to a UDP-style
> socket, which associations' local info should be returned?
> 
>                                                         K. Poon.
>                                                         kcpoon@eng.sun.com
> 
> _______________________________________________
> tsvwg mailing list
> tsvwg@ietf.org
> http://www1.ietf.org/mailman/listinfo/tsvwg

-- 
Randall R. Stewart
randall@stewart.chicago.il.us 815-342-5222 (cell phone)

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Thu Jan 10 10:03:22 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id KAA15842
	for <tsvwg-archive@odin.ietf.org>; Thu, 10 Jan 2002 10:03:22 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id KAA29839
	for tsvwg-archive@odin.ietf.org; Thu, 10 Jan 2002 10:03:23 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id JAA28938;
	Thu, 10 Jan 2002 09:39:40 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id JAA28862
	for <tsvwg@optimus.ietf.org>; Thu, 10 Jan 2002 09:39:36 -0500 (EST)
Received: from sj-msg-core-4.cisco.com (sj-msg-core-4.cisco.com [171.71.163.10])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id JAA15598
	for <tsvwg@ietf.org>; Thu, 10 Jan 2002 09:39:34 -0500 (EST)
Received: from mira-sjc5-2.cisco.com (mira-sjc5-2.cisco.com [171.71.163.16])
	by sj-msg-core-4.cisco.com (8.11.3/8.9.1) with ESMTP id g0AEcRJ06376;
	Thu, 10 Jan 2002 06:38:27 -0800 (PST)
Received: from stewart.chicago.il.us (ssh-sj1.cisco.com [171.68.225.134])
	by mira-sjc5-2.cisco.com (Mirapoint)
	with ESMTP id AAR78216;
	Thu, 10 Jan 2002 06:38:26 -0800 (PST)
Message-ID: <3C3DA762.C767295D@stewart.chicago.il.us>
Date: Thu, 10 Jan 2002 08:38:26 -0600
From: Randall Stewart <randall@stewart.chicago.il.us>
X-Mailer: Mozilla 4.76 [en] (X11; U; Linux 2.2.12 i386)
X-Accept-Language: en
MIME-Version: 1.0
To: Douglas Otis <dotis@sanlight.net>
CC: Tsvwg <tsvwg@ietf.org>
Subject: Re: [Tsvwg] RE: Working Group Last Call on SCTP checksum change
References: <NEBBJGDMMLHHCIKHGBEJMEKJCOAA.dotis@sanlight.net>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Transfer-Encoding: 7bit
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit

Doug:

These look real good, I have put them in the forming -02 version with
the addition of:

- Fixed references for 2119 and 2960
- Fixed some \n\r issues (corrected by good ole dos2unix).
- A commented pointed out by Matt Crawford (offline):

  in the abstract the statement:

  "For small packets, this provides weak protection against the 
   detection of errors."

  is a bit unclear  :>

  I think that changing it to:

  "For small packets Adler-32 provides weak detection of errors."

  makes it a bit more concise :>>


R
   




Douglas Otis wrote:
> 
> This could be used to update the draft to satisfy the missing sections:
> 
> 3 Security Considerations
> 
> There may be a computational advantage in validating the Association against
> the Verification Tag prior to performing a checksum as invalid tags will
> result in the same action as a bad checksum in most cases.  The exceptions
> for this technique would be INIT and some SHUTDOWN-COMPLETE exchanges as
> well as a stale COOKIE-ECHO.  These special case exchanges must represent
> small packets and will minimize the effect of the checksum calculation.
> 
> 4 IANA Considerations
> 
> There are no IANA considerations required in this document.
> 
> Doug
> 
> _______________________________________________
> tsvwg mailing list
> tsvwg@ietf.org
> http://www1.ietf.org/mailman/listinfo/tsvwg

-- 
Randall R. Stewart
randall@stewart.chicago.il.us 815-342-5222 (cell phone)

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Thu Jan 10 10:11:07 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id KAA15965
	for <tsvwg-archive@odin.ietf.org>; Thu, 10 Jan 2002 10:11:07 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id KAA00094
	for tsvwg-archive@odin.ietf.org; Thu, 10 Jan 2002 10:11:08 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id JAA28840;
	Thu, 10 Jan 2002 09:39:10 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id JAA28811
	for <tsvwg@optimus.ietf.org>; Thu, 10 Jan 2002 09:39:07 -0500 (EST)
Received: from mgw-x2.nokia.com (mgw-x2.nokia.com [131.228.20.22])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id JAA15591
	for <tsvwg@ietf.org>; Thu, 10 Jan 2002 09:39:04 -0500 (EST)
From: Ivan.Arias-Rodriguez@nokia.com
Received: from esvir04nok.ntc.nokia.com (esvir04nokt.ntc.nokia.com [172.21.143.36])
	by mgw-x2.nokia.com (Switch-2.1.0/Switch-2.1.0) with ESMTP id g0AEce905712
	for <tsvwg@ietf.org>; Thu, 10 Jan 2002 16:38:40 +0200 (EET)
Received: from esebh01nok.ntc.nokia.com (unverified) by esvir04nok.ntc.nokia.com
 (Content Technologies SMTPRS 4.2.5) with ESMTP id <T585dacac24ac158f2410a@esvir04nok.ntc.nokia.com>;
 Thu, 10 Jan 2002 16:38:39 +0200
Received: by esebh01nok with Internet Mail Service (5.5.2652.78)
	id <CHNT83MZ>; Thu, 10 Jan 2002 16:38:39 +0200
Message-ID: <58AEDE27D48F884285007ABF827FF26308EE16@esebe017.NOE.Nokia.com>
To: LyOng@ciena.com, tsvwg@ietf.org
Cc: audrey@vanb.com, steven.glass@sun.com
Subject: RE: [Tsvwg] SCTP testing at Connectathon
Date: Thu, 10 Jan 2002 16:38:37 +0200
MIME-Version: 1.0
X-Mailer: Internet Mail Service (5.5.2652.78)
content-class: urn:content-classes:message
Content-Type: text/plain;
	charset="ISO-8859-1"
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

	Hi all!

	I might be interested. However, I have checked the Connectathon
web page and it seems that every participant has the right to use only
one Ethernet socket... In that case it won't be possible to test
multihoming...

	I don't really know much about this, so I don't know if it would
be possible to carry your own hub or something so you are able to use at
least two Ethernet cards at the same time...

	Could somebody tell me how to solve this problem?

	Moreover, I'm not sure of what is the total amount that should
be paid... I mean, let's say I'm going there alone, so I should pay
2,300$ for the station (if I do it before January 15th), plus 300$ of
the engineer fee, plus probably something more since I will surely need
220V electrical power (don't know if I could find a 110V AC adapter)...
Am I right?

	Thanks!

> -----Original Message-----
> From: ext Ong, Lyndon [mailto:LyOng@ciena.com]
> Sent: 20. December 2001 21:05
> To: tsvwg@ietf.org
> Cc: audrey@vanb.com; steven.glass@sun.com
> Subject: [Tsvwg] SCTP testing at Connectathon
> 
> 
> Folks,
> 
> For those interested, we are planning to have SCTP testing as part of
> Connectathon
> at end February/beginning March 2002.  Steven Glass at Sun 
> has volunteered
> to
> coordinate.  He and Audrey Van Belleghem (steven.glass@sun.com and
> audrey@vanb.com)
> would like to get a count of the number of people interested in
> participating, so please
> send them an email if you are interested.
> 
> More information on Connectathon is at http://www.connectathon.org.
> 
> Since this is being set up a bit late, my understanding is that the
> discounted registration
> cutoff may be extended for the SCTP testing participants.
> 
> Cheers,
> 
> Lyndon Ong
> 
> 
> _______________________________________________
> tsvwg mailing list
> tsvwg@ietf.org
> http://www1.ietf.org/mailman/listinfo/tsvwg
> 

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Thu Jan 10 10:18:05 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id KAA16104
	for <tsvwg-archive@odin.ietf.org>; Thu, 10 Jan 2002 10:18:00 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id KAA00221
	for tsvwg-archive@odin.ietf.org; Thu, 10 Jan 2002 10:18:01 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id JAA29244;
	Thu, 10 Jan 2002 09:51:35 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id JAA29215
	for <tsvwg@optimus.ietf.org>; Thu, 10 Jan 2002 09:51:33 -0500 (EST)
Received: from gorilla.mchh.siemens.de (gorilla.mchh.siemens.de [194.138.158.18])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id JAA15728
	for <tsvwg@ietf.org>; Thu, 10 Jan 2002 09:51:30 -0500 (EST)
Received: from moody.mchh.siemens.de (mail2.mchh.siemens.de [194.138.158.226])
	by gorilla.mchh.siemens.de (8.9.3/8.9.3) with ESMTP id PAA08264;
	Thu, 10 Jan 2002 15:51:12 +0100 (MET)
Received: from mchh273e.demchh201e.icn.siemens.de ([139.21.200.83])
	by moody.mchh.siemens.de (8.9.1/8.9.1) with ESMTP id PAA26081;
	Thu, 10 Jan 2002 15:51:13 +0100 (MET)
Received: by MCHH273E with Internet Mail Service (5.5.2653.19)
	id <X35NCW3W>; Thu, 10 Jan 2002 15:51:12 +0100
Message-ID: <F92978223B01D311ACFF0008C71EE19D015DF6DC@MCHH225E>
From: Tuexen Michael <Michael.Tuexen@icn.siemens.de>
To: "'Ivan.Arias-Rodriguez@nokia.com'" <Ivan.Arias-Rodriguez@nokia.com>,
        LyOng@ciena.com, tsvwg@ietf.org
Cc: audrey@vanb.com, steven.glass@sun.com
Subject: RE: [Tsvwg] SCTP testing at Connectathon
Date: Thu, 10 Jan 2002 15:50:44 +0100
MIME-Version: 1.0
X-Mailer: Internet Mail Service (5.5.2653.19)
Content-Type: text/plain
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

Dear Ivan,

the setup of the test bed is currently being discussed. 
I hope that connecthaton can provide a setup similiar to 
the one used at the earlier bakeoffs. This would mean that you
have two connections (two ethernet plugs) to two different
IP networks. So you can really test multihoming and that 
stuff...

The prices you mentioned are correct (in my interpretation
of the website). Maybe someone from Sun can make a statement.

Best regards
Michael

Michael Tuexen    Tel.:   +49 89 722 47210
Siemens AG        Fax:    +49 89 722 48212
ICN WN CS SE 5    E-mail: Michael.Tuexen@icn.siemens.de

> -----Original Message-----
> From:	Ivan.Arias-Rodriguez@nokia.com [SMTP:Ivan.Arias-Rodriguez@nokia.com]
> Sent:	Thursday, January 10, 2002 3:39 PM
> To:	LyOng@ciena.com; tsvwg@ietf.org
> Cc:	audrey@vanb.com; steven.glass@sun.com
> Subject:	RE: [Tsvwg] SCTP testing at Connectathon
> 
> 	Hi all!
> 
> 	I might be interested. However, I have checked the Connectathon
> web page and it seems that every participant has the right to use only
> one Ethernet socket... In that case it won't be possible to test
> multihoming...
> 
> 	I don't really know much about this, so I don't know if it would
> be possible to carry your own hub or something so you are able to use at
> least two Ethernet cards at the same time...
> 
> 	Could somebody tell me how to solve this problem?
> 
> 	Moreover, I'm not sure of what is the total amount that should
> be paid... I mean, let's say I'm going there alone, so I should pay
> 2,300$ for the station (if I do it before January 15th), plus 300$ of
> the engineer fee, plus probably something more since I will surely need
> 220V electrical power (don't know if I could find a 110V AC adapter)...
> Am I right?
> 
> 	Thanks!
> 
> > -----Original Message-----
> > From: ext Ong, Lyndon [mailto:LyOng@ciena.com]
> > Sent: 20. December 2001 21:05
> > To: tsvwg@ietf.org
> > Cc: audrey@vanb.com; steven.glass@sun.com
> > Subject: [Tsvwg] SCTP testing at Connectathon
> > 
> > 
> > Folks,
> > 
> > For those interested, we are planning to have SCTP testing as part of
> > Connectathon
> > at end February/beginning March 2002.  Steven Glass at Sun 
> > has volunteered
> > to
> > coordinate.  He and Audrey Van Belleghem (steven.glass@sun.com and
> > audrey@vanb.com)
> > would like to get a count of the number of people interested in
> > participating, so please
> > send them an email if you are interested.
> > 
> > More information on Connectathon is at http://www.connectathon.org.
> > 
> > Since this is being set up a bit late, my understanding is that the
> > discounted registration
> > cutoff may be extended for the SCTP testing participants.
> > 
> > Cheers,
> > 
> > Lyndon Ong
> > 
> > 
> > _______________________________________________
> > tsvwg mailing list
> > tsvwg@ietf.org
> > http://www1.ietf.org/mailman/listinfo/tsvwg
> > 
> 
> _______________________________________________
> tsvwg mailing list
> tsvwg@ietf.org
> http://www1.ietf.org/mailman/listinfo/tsvwg

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Thu Jan 10 10:33:50 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id KAA16415
	for <tsvwg-archive@odin.ietf.org>; Thu, 10 Jan 2002 10:33:46 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id KAA00913
	for tsvwg-archive@odin.ietf.org; Thu, 10 Jan 2002 10:33:47 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id KAA00073;
	Thu, 10 Jan 2002 10:10:56 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id KAA00042
	for <tsvwg@optimus.ietf.org>; Thu, 10 Jan 2002 10:10:53 -0500 (EST)
Received: from sj-msg-core-3.cisco.com (sj-msg-core-3.cisco.com [171.70.157.152])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id KAA15951
	for <tsvwg@ietf.org>; Thu, 10 Jan 2002 10:10:51 -0500 (EST)
Received: from mira-sjc5-2.cisco.com (mira-sjc5-2.cisco.com [171.71.163.16])
	by sj-msg-core-3.cisco.com (8.11.3/8.9.1) with ESMTP id g0AF9pk01816;
	Thu, 10 Jan 2002 07:09:51 -0800 (PST)
Received: from stewart.chicago.il.us (ssh-sj1.cisco.com [171.68.225.134])
	by mira-sjc5-2.cisco.com (Mirapoint)
	with ESMTP id AAR78629;
	Thu, 10 Jan 2002 07:10:12 -0800 (PST)
Message-ID: <3C3DAED3.D9B7662E@stewart.chicago.il.us>
Date: Thu, 10 Jan 2002 09:10:11 -0600
From: Randall Stewart <randall@stewart.chicago.il.us>
X-Mailer: Mozilla 4.76 [en] (X11; U; Linux 2.2.12 i386)
X-Accept-Language: en
MIME-Version: 1.0
To: "'Ivan.Arias-Rodriguez@nokia.com'" <Ivan.Arias-Rodriguez@nokia.com>
CC: Tuexen Michael <Michael.Tuexen@icn.siemens.de>, LyOng@ciena.com,
        tsvwg@ietf.org, audrey@vanb.com, steven.glass@sun.com
Subject: Re: [Tsvwg] SCTP testing at Connectathon
References: <F92978223B01D311ACFF0008C71EE19D015DF6DC@MCHH225E>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Transfer-Encoding: 7bit
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit

Ivan:

One side note to add here.. I would hope they WILL
provide 110v power.. I am depending on it! 

So if your computer can take 110v then all you need
do is bring an adaptor to plug in to your power supply.
But you better double check the power supply allows input
from 110-220 (like my laptop does). I know some don't and I
have seen some PC's be smoked by a incorrect plug in :>

R

Tuexen Michael wrote:
> 
> Dear Ivan,
> 
> the setup of the test bed is currently being discussed.
> I hope that connecthaton can provide a setup similiar to
> the one used at the earlier bakeoffs. This would mean that you
> have two connections (two ethernet plugs) to two different
> IP networks. So you can really test multihoming and that
> stuff...
> 
> The prices you mentioned are correct (in my interpretation
> of the website). Maybe someone from Sun can make a statement.
> 
> Best regards
> Michael
> 
> Michael Tuexen    Tel.:   +49 89 722 47210
> Siemens AG        Fax:    +49 89 722 48212
> ICN WN CS SE 5    E-mail: Michael.Tuexen@icn.siemens.de
> 
> > -----Original Message-----
> > From: Ivan.Arias-Rodriguez@nokia.com [SMTP:Ivan.Arias-Rodriguez@nokia.com]
> > Sent: Thursday, January 10, 2002 3:39 PM
> > To:   LyOng@ciena.com; tsvwg@ietf.org
> > Cc:   audrey@vanb.com; steven.glass@sun.com
> > Subject:      RE: [Tsvwg] SCTP testing at Connectathon
> >
> >       Hi all!
> >
> >       I might be interested. However, I have checked the Connectathon
> > web page and it seems that every participant has the right to use only
> > one Ethernet socket... In that case it won't be possible to test
> > multihoming...
> >
> >       I don't really know much about this, so I don't know if it would
> > be possible to carry your own hub or something so you are able to use at
> > least two Ethernet cards at the same time...
> >
> >       Could somebody tell me how to solve this problem?
> >
> >       Moreover, I'm not sure of what is the total amount that should
> > be paid... I mean, let's say I'm going there alone, so I should pay
> > 2,300$ for the station (if I do it before January 15th), plus 300$ of
> > the engineer fee, plus probably something more since I will surely need
> > 220V electrical power (don't know if I could find a 110V AC adapter)...
> > Am I right?
> >
> >       Thanks!
> >
> > > -----Original Message-----
> > > From: ext Ong, Lyndon [mailto:LyOng@ciena.com]
> > > Sent: 20. December 2001 21:05
> > > To: tsvwg@ietf.org
> > > Cc: audrey@vanb.com; steven.glass@sun.com
> > > Subject: [Tsvwg] SCTP testing at Connectathon
> > >
> > >
> > > Folks,
> > >
> > > For those interested, we are planning to have SCTP testing as part of
> > > Connectathon
> > > at end February/beginning March 2002.  Steven Glass at Sun
> > > has volunteered
> > > to
> > > coordinate.  He and Audrey Van Belleghem (steven.glass@sun.com and
> > > audrey@vanb.com)
> > > would like to get a count of the number of people interested in
> > > participating, so please
> > > send them an email if you are interested.
> > >
> > > More information on Connectathon is at http://www.connectathon.org.
> > >
> > > Since this is being set up a bit late, my understanding is that the
> > > discounted registration
> > > cutoff may be extended for the SCTP testing participants.
> > >
> > > Cheers,
> > >
> > > Lyndon Ong
> > >
> > >
> > > _______________________________________________
> > > tsvwg mailing list
> > > tsvwg@ietf.org
> > > http://www1.ietf.org/mailman/listinfo/tsvwg
> > >
> >
> > _______________________________________________
> > tsvwg mailing list
> > tsvwg@ietf.org
> > http://www1.ietf.org/mailman/listinfo/tsvwg
> 
> _______________________________________________
> tsvwg mailing list
> tsvwg@ietf.org
> http://www1.ietf.org/mailman/listinfo/tsvwg

-- 
Randall R. Stewart
randall@stewart.chicago.il.us 815-342-5222 (cell phone)

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Thu Jan 10 11:09:44 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id LAA17358
	for <tsvwg-archive@odin.ietf.org>; Thu, 10 Jan 2002 11:09:43 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id LAA02245
	for tsvwg-archive@odin.ietf.org; Thu, 10 Jan 2002 11:09:45 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id KAA01267;
	Thu, 10 Jan 2002 10:45:15 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id KAA01238
	for <tsvwg@optimus.ietf.org>; Thu, 10 Jan 2002 10:45:13 -0500 (EST)
Received: from albatross-ext.wise.edt.ericsson.se (albatross-ext.wise.edt.ericsson.se [194.237.142.116])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id KAA16690
	for <tsvwg@ietf.org>; Thu, 10 Jan 2002 10:45:10 -0500 (EST)
Received: from eed.ericsson.se (mailhost.eed.ericsson.se [164.48.133.33])
	by albatross.wise.edt.ericsson.se (8.11.0/8.11.0/WIREfire-1.3) with ESMTP id g0AFjAJ25695;
	Thu, 10 Jan 2002 16:45:10 +0100 (MET)
Received: from res0010384da36.ericsson.com (dhcp5-66 [164.48.135.66])
	by eed.ericsson.se (8.8.8+Sun/1.1.mit) with ESMTP id QAA23413;
	Thu, 10 Jan 2002 16:45:06 +0100 (MET)
Message-Id: <5.1.0.14.0.20020110163120.01ba2e98@mailhost.eed.ericsson.se>
X-Sender: eedrel@mailhost.eed.ericsson.se
X-Mailer: QUALCOMM Windows Eudora Version 5.1
Date: Thu, 10 Jan 2002 16:45:59 +0100
To: mallman@grc.nasa.gov
From: Reiner Ludwig <Reiner.Ludwig@ericsson.com>
Subject: Re: [Tsvwg] Eifel draft and usage of time-stamps. 
Cc: "sanjay kaniyar" <skaniyar@hotmail.com>, gurtov@cs.helsinki.fi,
        tsvwg@ietf.org, Reiner.Ludwig@ericsson.com
In-Reply-To: <200201101233.HAA32354@lawyers.grc.nasa.gov>
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"; format=flowed
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

I haven't caught up in this discussion yet, but one comment ...

At 13:33 10.01.2002, Mark Allman wrote:
>If you (a) don't want to incur the overhead of timestamps and (b)
>want to detect spurious retransmits, you *do* have another option.
>You can use DSACK to detect spurious retransmits.

If you want to detect a spurious retransmit caused by a spurious timeout, 
you typically want to do that so that you can avoid that the entire 
outstanding flight of segments is (often unnecessarily) retransmitted. This 
"go-back-N misbehavior" inevitably follows a spurious timeout.

However, you *cannot* use DSACK to avoid those go-back-N retransmits since 
the DSACK only arrives after the "go-back-N misbehavior" has already taken 
place.

///Reiner


_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Thu Jan 10 11:11:46 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id LAA17434
	for <tsvwg-archive@odin.ietf.org>; Thu, 10 Jan 2002 11:11:41 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id LAA02330
	for tsvwg-archive@odin.ietf.org; Thu, 10 Jan 2002 11:11:43 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id KAA01548;
	Thu, 10 Jan 2002 10:53:30 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id KAA01521
	for <tsvwg@optimus.ietf.org>; Thu, 10 Jan 2002 10:53:28 -0500 (EST)
Received: from seraph3.grc.nasa.gov (firewall-user@seraph3.lerc.nasa.gov [128.156.10.12])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id KAA16962
	for <tsvwg@ietf.org>; Thu, 10 Jan 2002 10:53:26 -0500 (EST)
Received: by seraph3.grc.nasa.gov (Postfix, from userid 5)
	id D5385640C3; Thu, 10 Jan 2002 10:52:48 -0500 (EST)
Received: from lombok-fi.lerc.nasa.gov(139.88.112.33) by seraph3.grc.nasa.gov via csmap (V6.0)
	id srcAAAN8a4Gc; Thu, 10 Jan 02 10:52:48 -0500
Received: from guns.lerc.nasa.gov (guns.lerc.nasa.gov [139.88.87.35])
	by lombok-fi.lerc.nasa.gov (NASA LeRC 8.9.1.1/8.9.1) with ESMTP id KAA25080;
	Thu, 10 Jan 2002 10:52:48 -0500 (EST)
Received: from guns.lerc.nasa.gov (localhost.lerc.nasa.gov [127.0.0.1]) by guns.lerc.nasa.gov with ESMTP (NASA LeRC 8.7.4.1/2.01-local)
        id KAA18740; Thu, 10 Jan 2002 10:52:48 -0500 (EST)
Message-Id: <200201101552.KAA18740@guns.lerc.nasa.gov>
To: Reiner Ludwig <Reiner.Ludwig@ericsson.com>
From: Mark Allman <mallman@grc.nasa.gov>
Reply-To: mallman@grc.nasa.gov
Cc: "sanjay kaniyar" <skaniyar@hotmail.com>, gurtov@cs.Helsinki.FI,
        tsvwg@ietf.org
Subject: Re: [Tsvwg] Eifel draft and usage of time-stamps. 
Organization: BBN Technologies/NASA GRC
Song-of-the-Day: Who Are You
Date: Thu, 10 Jan 2002 10:52:48 -0500
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org


> >If you (a) don't want to incur the overhead of timestamps and (b)
> >want to detect spurious retransmits, you *do* have another
> >option.  You can use DSACK to detect spurious retransmits.
> 
> If you want to detect a spurious retransmit caused by a spurious
> timeout, you typically want to do that so that you can avoid that
> the entire outstanding flight of segments is (often unnecessarily)
> retransmitted. This "go-back-N misbehavior" inevitably follows a
> spurious timeout.
> 
> However, you *cannot* use DSACK to avoid those go-back-N
> retransmits since the DSACK only arrives after the "go-back-N
> misbehavior" has already taken place.

Yes.  But, if you can detect a spurious retransmit (and go-back-n
misbehavior) you can avoid it in the future.  So, there is still
value in detection no matter when you do so.  I am not saying that
use of the DSACK method is *preferred*.  I am simply noting that if
someone does not want to incur the cost of using timestamps and
eifel there is still *something* that can be done.

allman


---
Mark Allman -- BBN/NASA GRC -- http://roland.grc.nasa.gov/~mallman/

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Thu Jan 10 11:38:18 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id LAA18448
	for <tsvwg-archive@odin.ietf.org>; Thu, 10 Jan 2002 11:38:17 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id LAA03710
	for tsvwg-archive@odin.ietf.org; Thu, 10 Jan 2002 11:38:20 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id LAA02669;
	Thu, 10 Jan 2002 11:20:14 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id LAA02640
	for <tsvwg@optimus.ietf.org>; Thu, 10 Jan 2002 11:20:12 -0500 (EST)
Received: from sj-msg-core-1.cisco.com (sj-msg-core-1.cisco.com [171.71.163.11])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id LAA17814
	for <tsvwg@ietf.org>; Thu, 10 Jan 2002 11:20:09 -0500 (EST)
Received: from mira-sjc5-2.cisco.com (mira-sjc5-2.cisco.com [171.71.163.16])
	by sj-msg-core-1.cisco.com (8.11.3/8.9.1) with ESMTP id g0AGJeF00347;
	Thu, 10 Jan 2002 08:19:40 -0800 (PST)
Received: from stewart.chicago.il.us (ssh-sj1.cisco.com [171.68.225.134])
	by mira-sjc5-2.cisco.com (Mirapoint)
	with ESMTP id AAR79731;
	Thu, 10 Jan 2002 08:19:39 -0800 (PST)
Message-ID: <3C3DBF1A.6F725E93@stewart.chicago.il.us>
Date: Thu, 10 Jan 2002 10:19:38 -0600
From: Randall Stewart <randall@stewart.chicago.il.us>
X-Mailer: Mozilla 4.76 [en] (X11; U; Linux 2.2.12 i386)
X-Accept-Language: en
MIME-Version: 1.0
To: TSV WG list <tsvwg@ietf.org>,
        sigtran-test <SIGTRAN-TEST@STANDARDS.NORTELNETWORKS.COM>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Transfer-Encoding: 7bit
Subject: [Tsvwg] SCTP ref impl update
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit

Hi All:

I just put up on the

http://www.sctp.org

web site the 4.0.5 version of the reference implementation.

It DOES not include the new crc32c algorithm... and is
the same one available in the SCTP book.

It does contain:

 - Numerous bug fixes
 - Fully functional Add-ip
 - Fully functional u-sctp


Regards

R
-- 
Randall R. Stewart
randall@stewart.chicago.il.us 815-342-5222 (cell phone)

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Thu Jan 10 12:39:00 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id MAA19946
	for <tsvwg-archive@odin.ietf.org>; Thu, 10 Jan 2002 12:39:00 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id MAA07380
	for tsvwg-archive@odin.ietf.org; Thu, 10 Jan 2002 12:39:01 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id MAA05890;
	Thu, 10 Jan 2002 12:09:00 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id MAA05856
	for <tsvwg@optimus.ietf.org>; Thu, 10 Jan 2002 12:08:58 -0500 (EST)
Received: from virtual4.dsldesigns.com (IDENT:root@[216.83.40.78])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id MAA19337
	for <tsvwg@ietf.org>; Thu, 10 Jan 2002 12:08:55 -0500 (EST)
Received: from vanb.com ([198.79.111.50])
	by virtual4.dsldesigns.com (8.11.1/8.11.1) with ESMTP id g0AHufL22349;
	Thu, 10 Jan 2002 09:56:41 -0800
Message-ID: <3C3DC90D.79C5177A@vanb.com>
Date: Thu, 10 Jan 2002 09:02:05 -0800
From: Audrey Van Belleghem <audrey@vanb.com>
Reply-To: audrey@vanb.com
Organization: Van Belleghem Corporation
X-Mailer: Mozilla 4.5 (Macintosh; I; PPC)
X-Accept-Language: en
MIME-Version: 1.0
To: Ivan.Arias-Rodriguez@nokia.com
CC: LyOng@ciena.com, tsvwg@ietf.org, steven.glass@sun.com
Subject: Re: [Tsvwg] SCTP testing at Connectathon
References: <58AEDE27D48F884285007ABF827FF26308EE16@esebe017.NOE.Nokia.com>
Content-Type: text/plain; charset=us-ascii; x-mac-type="54455854"; x-mac-creator="4D4F5353"
Content-Transfer-Encoding: 7bit
Content-Transfer-Encoding: 7bit
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit

Hi Ivan,

The Connectathon drops are defined by station size.  One drop for a single
station, 2 drops for a small, etc.  You are allowed to bring a hub if you
need to expand your drops.

Nokia is already registered as a participant (I can put you in touch with
the right people there), so you would only need to see if they have space
in their station for one additional person.  If so, you only need to pay
the $300 engineer fee.

If you need 220V, there is an additional charge to provide that power to
you.  I would need your NEMA number, voltage requirement, and single or
3-phase information to provide the quote from the electrician (the cost
comes straight from them but we handle the transaction).

Thanks for your interest!  Yes, deadline for early discount is Jan. 15th.

Audrey

Ivan.Arias-Rodriguez@nokia.com wrote:

>         Hi all!
>
>         I might be interested. However, I have checked the Connectathon
> web page and it seems that every participant has the right to use only
> one Ethernet socket... In that case it won't be possible to test
> multihoming...
>
>         I don't really know much about this, so I don't know if it would
> be possible to carry your own hub or something so you are able to use at
> least two Ethernet cards at the same time...
>
>         Could somebody tell me how to solve this problem?
>
>         Moreover, I'm not sure of what is the total amount that should
> be paid... I mean, let's say I'm going there alone, so I should pay
> 2,300$ for the station (if I do it before January 15th), plus 300$ of
> the engineer fee, plus probably something more since I will surely need
> 220V electrical power (don't know if I could find a 110V AC adapter)...
> Am I right?
>
>         Thanks!
>
> > -----Original Message-----
> > From: ext Ong, Lyndon [mailto:LyOng@ciena.com]
> > Sent: 20. December 2001 21:05
> > To: tsvwg@ietf.org
> > Cc: audrey@vanb.com; steven.glass@sun.com
> > Subject: [Tsvwg] SCTP testing at Connectathon
> >
> >
> > Folks,
> >
> > For those interested, we are planning to have SCTP testing as part of
> > Connectathon
> > at end February/beginning March 2002.  Steven Glass at Sun
> > has volunteered
> > to
> > coordinate.  He and Audrey Van Belleghem (steven.glass@sun.com and
> > audrey@vanb.com)
> > would like to get a count of the number of people interested in
> > participating, so please
> > send them an email if you are interested.
> >
> > More information on Connectathon is at http://www.connectathon.org.
> >
> > Since this is being set up a bit late, my understanding is that the
> > discounted registration
> > cutoff may be extended for the SCTP testing participants.
> >
> > Cheers,
> >
> > Lyndon Ong
> >
> >
> > _______________________________________________
> > tsvwg mailing list
> > tsvwg@ietf.org
> > http://www1.ietf.org/mailman/listinfo/tsvwg
> >


_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Thu Jan 10 20:24:48 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id UAA27226
	for <tsvwg-archive@odin.ietf.org>; Thu, 10 Jan 2002 20:24:48 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id UAA22304
	for tsvwg-archive@odin.ietf.org; Thu, 10 Jan 2002 20:24:49 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id UAA21974;
	Thu, 10 Jan 2002 20:02:36 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id UAA21948
	for <tsvwg@optimus.ietf.org>; Thu, 10 Jan 2002 20:02:34 -0500 (EST)
Received: from patan.sun.com (patan.Sun.COM [192.18.98.43])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id UAA26967
	for <tsvwg@ietf.org>; Thu, 10 Jan 2002 20:02:32 -0500 (EST)
Received: from engmail1.Eng.Sun.COM ([129.146.1.13])
	by patan.sun.com (8.9.3+Sun/8.9.3) with ESMTP id SAA08334;
	Thu, 10 Jan 2002 18:02:32 -0700 (MST)
Received: from shield (shield.EBay.Sun.COM [129.150.113.84])
	by engmail1.Eng.Sun.COM (8.9.3+Sun/8.9.3/ENSMAIL,v2.1p1) with SMTP id RAA09026;
	Thu, 10 Jan 2002 17:02:31 -0800 (PST)
Date: Thu, 10 Jan 2002 17:01:51 -0800 (PST)
From: Kacheong Poon <kcpoon@eng.sun.com>
Reply-To: Kacheong Poon <kcpoon@eng.sun.com>
Subject: Re: [Tsvwg] Eifel draft and usage of time-stamps.
To: sanjay kaniyar <skaniyar@hotmail.com>
Cc: tsvwg@ietf.org, Reiner.Ludwig@ericsson.com
In-Reply-To: "Your message with ID" <F127naTbyyw8whaJqbN0001d39e@hotmail.com>
Message-ID: <Roam.SIMC.2.0.6.1010710911.25756.kcpoon@eng.sun.com>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; CHARSET=US-ASCII
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

> 1. Data transfer could be unidirectional, in which case timestamps would 
> only be necessary for the direction of data flow. Somewhere in the 1323, 
> they mention that they chose to 'always' use it bidirectionally for the 
> purpose of simplicity etc.

Hmm, are you saying that one side sets the timestamp option and the
other side does not echo it back?  Can you be more specific on exactly
the format you are talking about?  Have you thought about the mechanism 
you are proposing?  Are you trying to come up with a new timestamp option
for unidirectional transfer?  Please tell us more how it should work.

> 2. I did not misinterpret your comment, I understood that Solaris does not 
> reject segments without timestamps. My point is that, the behavior of 
> 'considering the peer to be not able to generate timestamps accurately just 
> because some segments did not have timestamps' is what is Solaris speicific.

Point taken.

> 3. It may not be necessary to time every segment in cases like, for example, 
> where it does not change much at all? or maybe timing every 3rd segment is 
> just as good? or maybe it is too small and the sender TCP is going to stick 
> to its attainable minimum?
>
> 4. It is possible to say a connection does not need PAWS check based on the 
> perceived RTT and the advertised receive window size, for example. Or if 
> there is a limit set on the maximum bandwidth a connection can use etc.

I guess you didn't answer my questions at all.  How do you know at
the connection setup time that timestamp should be used ?  TCP
knows nothing about RTT (well, maybe temporal sharing of some info from
a previous connection can help), the other side's advertised window
(active opening), ...  And I believe you missed the reason for PAWS, it 
has to do with the maximum transfer bandwidth, not really RTT related, and
is not easy to measure at the end host...  And I believe you agreed that 
for PAWS to work, every segments need to have the timestamp option.  So
how should your proposed change to use timestamp option for retransmission
indication work?

You should think more about what you are proposing before you can say 
that it is easy to do (there are too many maybes in your statements...). 
Please remove those maybes and state exactly when RTTM is an overkill and 
show us how this can be determined at connection setup time.  Also note that
whether a TCP stack makes use of all RTT samples to calculate RTO and which
samples to use can be an implementation choice.

> 5. Whether 12 bytes in each segment and ack is an overhead or not (with ROHC 
> or without) is quite subjective, I think. To me, a single byte that goes on 
> the wire that does not have any purpose to it is an overhead and should be 
> avoided if it is not too much of an engineering overhead (and I believe with 
> timestamps it is not).

In what way does this "engineering decision" support your comment about
the main reason of limited use of timestamp option?  Honestly, I don't
know if it is really of limited use or not.  Maybe you can show us the
data.  And please show us what you meant by "not much of an engineering 
overhead."  That is the exact way of how your proposal works.

							K. Poon.
							kcpoon@eng.sun.com




_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Thu Jan 10 20:45:27 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id UAA27475
	for <tsvwg-archive@odin.ietf.org>; Thu, 10 Jan 2002 20:45:26 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id UAA23034
	for tsvwg-archive@odin.ietf.org; Thu, 10 Jan 2002 20:45:28 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id UAA22375;
	Thu, 10 Jan 2002 20:27:14 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id UAA22350
	for <tsvwg@optimus.ietf.org>; Thu, 10 Jan 2002 20:27:13 -0500 (EST)
Received: from patan.sun.com (patan.Sun.COM [192.18.98.43])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id UAA27257
	for <tsvwg@ietf.org>; Thu, 10 Jan 2002 20:27:10 -0500 (EST)
Received: from engmail1.Eng.Sun.COM ([129.146.1.13])
	by patan.sun.com (8.9.3+Sun/8.9.3) with ESMTP id SAA19843;
	Thu, 10 Jan 2002 18:27:11 -0700 (MST)
Received: from shield (shield.EBay.Sun.COM [129.150.113.84])
	by engmail1.Eng.Sun.COM (8.9.3+Sun/8.9.3/ENSMAIL,v2.1p1) with SMTP id RAA16247;
	Thu, 10 Jan 2002 17:27:10 -0800 (PST)
Date: Thu, 10 Jan 2002 17:26:29 -0800 (PST)
From: Kacheong Poon <kcpoon@eng.sun.com>
Reply-To: Kacheong Poon <kcpoon@eng.sun.com>
Subject: Re: [Tsvwg] Eifel draft and usage of time-stamps. 
To: mallman@grc.nasa.gov
Cc: Reiner Ludwig <Reiner.Ludwig@ericsson.com>,
        sanjay kaniyar <skaniyar@hotmail.com>, gurtov@cs.Helsinki.FI,
        tsvwg@ietf.org
In-Reply-To: "Your message with ID" <200201101552.KAA18740@guns.lerc.nasa.gov>
Message-ID: <Roam.SIMC.2.0.6.1010712389.10142.kcpoon@eng.sun.com>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; CHARSET=US-ASCII
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

> Yes.  But, if you can detect a spurious retransmit (and go-back-n
> misbehavior) you can avoid it in the future.  So, there is still
> value in detection no matter when you do so.  I am not saying that
> use of the DSACK method is *preferred*.  I am simply noting that if
> someone does not want to incur the cost of using timestamps and
> eifel there is still *something* that can be done.

But how does one spurious timeout correlate to the next timeout?
I don't know how TCP can say the next timeout must be spurious
given that the previous one is...

							K. Poon.
							kcpoon@eng.sun.com



_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Thu Jan 10 22:19:39 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id WAA29394
	for <tsvwg-archive@odin.ietf.org>; Thu, 10 Jan 2002 22:19:38 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id WAA25500
	for tsvwg-archive@odin.ietf.org; Thu, 10 Jan 2002 22:19:41 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id VAA24808;
	Thu, 10 Jan 2002 21:57:50 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id VAA24777
	for <tsvwg@optimus.ietf.org>; Thu, 10 Jan 2002 21:57:48 -0500 (EST)
Received: from hotmail.com (f52.law3.hotmail.com [209.185.241.52])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id VAA29188
	for <tsvwg@ietf.org>; Thu, 10 Jan 2002 21:57:45 -0500 (EST)
Received: from mail pickup service by hotmail.com with Microsoft SMTPSVC;
	 Thu, 10 Jan 2002 18:57:11 -0800
Received: from 207.46.137.9 by lw3fd.law3.hotmail.msn.com with HTTP;
	Fri, 11 Jan 2002 02:57:10 GMT
X-Originating-IP: [207.46.137.9]
From: "sanjay kaniyar" <skaniyar@hotmail.com>
To: kcpoon@eng.sun.com
Cc: tsvwg@ietf.org, Reiner.Ludwig@ericsson.com
Subject: Re: [Tsvwg] Eifel draft and usage of time-stamps.
Date: Thu, 10 Jan 2002 18:57:10 -0800
Mime-Version: 1.0
Content-Type: text/plain; format=flowed
Message-ID: <F52S01qCcWNYgEj4v5i0001e919@hotmail.com>
X-OriginalArrivalTime: 11 Jan 2002 02:57:11.0968 (UTC) FILETIME=[AAAB9200:01C19A4B]
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

1. Timestamps are bidirectional in the sense that in a segment going from 
A->B, there are 2 of them: 1:generated by A, 2:echoed value, sent by B 
originally. If the connection is doing unidirectional data transfer 
entirely, it is sufficient to have timestamps generated and sent by the 
sender in the data segments and acknowledgements echo them in the other 
direction. This is what RFC 1323 referred to as 'unidirectional use of 
timestamps'.

They chose to 'always use timestamps bidirectionally' instead because of 
reasons like simplicity etc. There is nothing new I am proposing here. I 
only quoted this to rebut your claim - 'always refers sending timestamps in 
each segment'.

3 & 4.
(a) I did answer your question. I never said it is always possible to figure 
out whether it is possible to figure out whether timestamps are necessary at 
the connection set up time (nor is it the problem I am trying to solve). All 
I am trying to do is to timestamp (and echo) segments on a connection which 
has timestamps negotiated on it only when it is necessary and useful, I said 
nothing about figuruing out whenther they should be negotiated or not.

(b) I know that to do PAWS check, every segment needs to be timstamped. 
However, to figure out whether PAWS check needs to be done or not, can be 
implementation specific because there can be many reasons within the perview 
of TCP or outside of it which could contribute to this. Note that, for a 
connection to require PAWS check it has to (i) have a transfer rate of 
greater than a threshold and (ii) have a life time greater than a threhold. 
If it is possible to determine one of these is not going to be met, there is 
no need to do PAWS on that connection. The maybe's were actually examples of 
such instances:

- A bandwidth limit is imposed on each connection {(i) fails}
- Connections are always secured with IPSEC, which has anti-replay 
protection.
- A timestamp negotiated connection starts with the assumption that PAWS is 
necessary at the peer and timestamps every segment to begin with. After it 
has an estimate of the RTT, it figures out (i) will fail because the maximum 
transfer rate is limited by the bandwidth-delay product.
- The sending application did a graceful close after submitting a buffer 
which is smaller than what is takes to cover the sequence space.
- A mail server admin knows each client has a mailbox size maximum of 1 
gigabytes and there is no danger of wrap around when a mail client tries to 
download it.

My point is that, this is really something that can't be put in a spec, but 
chosen by many factors.

(c) Again, determining whehter timestamping every segment for RTTM is an 
overkill or not should be an implemenation choice. Examples being:

- The timer on the sender TCP has a granularity of 100 ms and the first 10 
RTT measurements are all below 50 seconds.
- The last 10 RTT measurements did not vary by more than 5%.

5. My suggestion was, if a sender TCP is able to figure this out, it can 
selectively use timestamps. If the receiver also adheres to some well known 
rule, it would save sending some extra bytes in either direction. The rule 
is what I thought could be part of Eifel draft, as it was using timestamps 
in a way other than RTTM and PAWS, and could be achieved wihtout 
timestamping every segment. You might already have seen the following:
Sender-TCP would put a timestamp in a segment if:
* PAWS protection is necessary for the connection.
* An RTT measurement using timestamps is desired, and one such measurement 
is in progress. We consider a measurement to be in progress if sender has 
sent a timestamp after it started a measurement and it is not yet echoed 
back.
* Retransmission of a segment upon timeout.

Receiver-TCP would put a timestamp (or echo what it got) in an ack if:
* no more than 3 acks have been sent after any segment with a timestamp was 
received.

Sanjay.

>From: Kacheong Poon <kcpoon@eng.sun.com>
>Reply-To: Kacheong Poon <kcpoon@eng.sun.com>
>To: sanjay kaniyar <skaniyar@hotmail.com>
>CC: tsvwg@ietf.org, Reiner.Ludwig@ericsson.com
>Subject: Re: [Tsvwg] Eifel draft and usage of time-stamps.
>Date: Thu, 10 Jan 2002 17:01:51 -0800 (PST)
>
> > 1. Data transfer could be unidirectional, in which case timestamps would
> > only be necessary for the direction of data flow. Somewhere in the 1323,
> > they mention that they chose to 'always' use it bidirectionally for the
> > purpose of simplicity etc.
>
>Hmm, are you saying that one side sets the timestamp option and the
>other side does not echo it back?  Can you be more specific on exactly
>the format you are talking about?  Have you thought about the mechanism
>you are proposing?  Are you trying to come up with a new timestamp option
>for unidirectional transfer?  Please tell us more how it should work.
>
> > 2. I did not misinterpret your comment, I understood that Solaris does 
>not
> > reject segments without timestamps. My point is that, the behavior of
> > 'considering the peer to be not able to generate timestamps accurately 
>just
> > because some segments did not have timestamps' is what is Solaris 
>speicific.
>
>Point taken.
>
> > 3. It may not be necessary to time every segment in cases like, for 
>example,
> > where it does not change much at all? or maybe timing every 3rd segment 
>is
> > just as good? or maybe it is too small and the sender TCP is going to 
>stick
> > to its attainable minimum?
> >
> > 4. It is possible to say a connection does not need PAWS check based on 
>the
> > perceived RTT and the advertised receive window size, for example. Or if
> > there is a limit set on the maximum bandwidth a connection can use etc.
>
>I guess you didn't answer my questions at all.  How do you know at
>the connection setup time that timestamp should be used ?  TCP
>knows nothing about RTT (well, maybe temporal sharing of some info from
>a previous connection can help), the other side's advertised window
>(active opening), ...  And I believe you missed the reason for PAWS, it
>has to do with the maximum transfer bandwidth, not really RTT related, and
>is not easy to measure at the end host...  And I believe you agreed that
>for PAWS to work, every segments need to have the timestamp option.  So
>how should your proposed change to use timestamp option for retransmission
>indication work?
>
>You should think more about what you are proposing before you can say
>that it is easy to do (there are too many maybes in your statements...).
>Please remove those maybes and state exactly when RTTM is an overkill and
>show us how this can be determined at connection setup time.  Also note 
>that
>whether a TCP stack makes use of all RTT samples to calculate RTO and which
>samples to use can be an implementation choice.
>
> > 5. Whether 12 bytes in each segment and ack is an overhead or not (with 
>ROHC
> > or without) is quite subjective, I think. To me, a single byte that goes 
>on
> > the wire that does not have any purpose to it is an overhead and should 
>be
> > avoided if it is not too much of an engineering overhead (and I believe 
>with
> > timestamps it is not).
>
>In what way does this "engineering decision" support your comment about
>the main reason of limited use of timestamp option?  Honestly, I don't
>know if it is really of limited use or not.  Maybe you can show us the
>data.  And please show us what you meant by "not much of an engineering
>overhead."  That is the exact way of how your proposal works.
>
>							K. Poon.
>							kcpoon@eng.sun.com
>
>
>
>
>_______________________________________________
>tsvwg mailing list
>tsvwg@ietf.org
>http://www1.ietf.org/mailman/listinfo/tsvwg




_________________________________________________________________
Join the world뭩 largest e-mail service with MSN Hotmail. 
http://www.hotmail.com


_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Thu Jan 10 23:24:49 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id XAA00979
	for <tsvwg-archive@odin.ietf.org>; Thu, 10 Jan 2002 23:24:49 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id XAA05937
	for tsvwg-archive@odin.ietf.org; Thu, 10 Jan 2002 23:24:50 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id XAA05780;
	Thu, 10 Jan 2002 23:10:53 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id XAA05755
	for <tsvwg@optimus.ietf.org>; Thu, 10 Jan 2002 23:10:51 -0500 (EST)
Received: from lawyers.grc.nasa.gov (d134.as0.clev.oh.voyager.net [209.81.165.135])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id XAA00872
	for <tsvwg@ietf.org>; Thu, 10 Jan 2002 23:10:49 -0500 (EST)
Received: from lawyers (mallman@localhost)
	by lawyers.grc.nasa.gov (8.9.3/8.9.3) with ESMTP id XAA01325;
	Thu, 10 Jan 2002 23:08:26 -0500
Message-Id: <200201110408.XAA01325@lawyers.grc.nasa.gov>
To: Kacheong Poon <kcpoon@eng.sun.com>
From: Mark Allman <mallman@grc.nasa.gov>
Reply-To: mallman@grc.nasa.gov
cc: Reiner Ludwig <Reiner.Ludwig@ericsson.com>,
        sanjay kaniyar <skaniyar@hotmail.com>, gurtov@cs.Helsinki.FI,
        tsvwg@ietf.org
Subject: Re: [Tsvwg] Eifel draft and usage of time-stamps. 
Organization: BBN Technologies/NASA GRC
Song-of-the-Day: Lawyers, Guns and Money
Date: Thu, 10 Jan 2002 23:08:26 -0500
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org


> > Yes.  But, if you can detect a spurious retransmit (and go-back-n
> > misbehavior) you can avoid it in the future.  So, there is still
> > value in detection no matter when you do so.  I am not saying that
> > use of the DSACK method is *preferred*.  I am simply noting that if
> > someone does not want to incur the cost of using timestamps and
> > eifel there is still *something* that can be done.
> 
> But how does one spurious timeout correlate to the next timeout?
> I don't know how TCP can say the next timeout must be spurious
> given that the previous one is...

Oh, you can't.  (Or, at least it is not obvious to me that you can.)

What I meant was that if you know you took a spurious timeout (or
did a needless fast retransmit) you can take that as an indication
that your estimator is not working as well as you might like (i.e.,
the RTO is not waiting long enough for an ACK to return).  So, you
can tweak the estimator somehow (or alter the dup ack threshold
somehow) in an attempt to come up with a better trigger for
determining loss so that maybe you can prevent further spurious
retransmits (and the resulting negative behavior).

allman


---
Mark Allman -- NASA GRC/BBN -- http://roland.grc.nasa.gov/~mallman/

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Thu Jan 10 23:25:57 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id XAA01008
	for <tsvwg-archive@odin.ietf.org>; Thu, 10 Jan 2002 23:25:57 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id XAA05971
	for tsvwg-archive@odin.ietf.org; Thu, 10 Jan 2002 23:25:58 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id XAA05742;
	Thu, 10 Jan 2002 23:09:28 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id XAA05709
	for <tsvwg@optimus.ietf.org>; Thu, 10 Jan 2002 23:09:26 -0500 (EST)
Received: from patan.sun.com (patan.Sun.COM [192.18.98.43])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id XAA00864
	for <tsvwg@ietf.org>; Thu, 10 Jan 2002 23:09:24 -0500 (EST)
Received: from engmail1.Eng.Sun.COM ([129.146.1.13])
	by patan.sun.com (8.9.3+Sun/8.9.3) with ESMTP id VAA15639;
	Thu, 10 Jan 2002 21:09:25 -0700 (MST)
Received: from shield (shield.EBay.Sun.COM [129.150.113.84])
	by engmail1.Eng.Sun.COM (8.9.3+Sun/8.9.3/ENSMAIL,v2.1p1) with SMTP id UAA18665;
	Thu, 10 Jan 2002 20:09:24 -0800 (PST)
Date: Thu, 10 Jan 2002 20:08:44 -0800 (PST)
From: Kacheong Poon <kcpoon@eng.sun.com>
Reply-To: Kacheong Poon <kcpoon@eng.sun.com>
Subject: Re: [Tsvwg] Eifel draft and usage of time-stamps.
To: sanjay kaniyar <skaniyar@hotmail.com>
Cc: tsvwg@ietf.org, Reiner.Ludwig@ericsson.com
In-Reply-To: "Your message with ID" <F52S01qCcWNYgEj4v5i0001e919@hotmail.com>
Message-ID: <Roam.SIMC.2.0.6.1010722124.30197.kcpoon@eng.sun.com>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; CHARSET=US-ASCII
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

> 1. Timestamps are bidirectional in the sense that in a segment going from 
> A->B, there are 2 of them: 1:generated by A, 2:echoed value, sent by B 
> originally. If the connection is doing unidirectional data transfer 
> entirely, it is sufficient to have timestamps generated and sent by the 
> sender in the data segments and acknowledgements echo them in the other 
> direction. This is what RFC 1323 referred to as 'unidirectional use of 
> timestamps'.
> 
> They chose to 'always use timestamps bidirectionally' instead because of 
> reasons like simplicity etc. There is nothing new I am proposing here. I 
> only quoted this to rebut your claim - 'always refers sending timestamps in 
> each segment'.

Correct me if I am wrong, your problem with the current timestamp
option is that it takes up 12 bytes.  Now in your "unidirectional"
timestamp scenario, isn't it true that the 12 bytes timestamp option
is still in every segments in both directions even if only the data
sender chooses to use it?  Note that I never mentioned that "unidirectional
timestamp option" is not possible.  I am trying to say that the current
timestamp option, with both time and echo value in it, works in a
specific way.  It is designed to be used for both sides.  You can come up
with your own "unidirectional timestamp option."  But it is just not the 
same as RFC 1323....  Afterall, you said that in a previous email,
"I interpret it as 'don't use timestmaps in a unidirectional way'."  So
are you now saying that RFC 1323 allows you to use timestamps in
"undirectional way?"

And think more about interoperability (you can ignore what I said about
Solaris for this purpose).  The fact is that all stacks nowadays send 
timestamp option is every segments once it is negotiated.  And the fact 
that a TCP stack will never know how an app is going to use it.  And ...
There are a lot of issues to think about.

> 3 & 4.
> (a) I did answer your question. I never said it is always possible to figure 
> out whether it is possible to figure out whether timestamps are necessary at 
> the connection set up time (nor is it the problem I am trying to solve). All 
> I am trying to do is to timestamp (and echo) segments on a connection which 
> has timestamps negotiated on it only when it is necessary and useful, I said 
> nothing about figuruing out whenther they should be negotiated or not.

So are you suggesting something that you don't even know how to make
it work?

> (b) I know that to do PAWS check, every segment needs to be timstamped. 
> However, to figure out whether PAWS check needs to be done or not, can be 
> implementation specific because there can be many reasons within the perview 
> of TCP or outside of it which could contribute to this. Note that, for a 
> connection to require PAWS check it has to (i) have a transfer rate of 
> greater than a threshold and (ii) have a life time greater than a threhold. 
> If it is possible to determine one of these is not going to be met, there is 
> no need to do PAWS on that connection. The maybe's were actually examples of 
> such instances:
> 
> - A bandwidth limit is imposed on each connection {(i) fails}
> - Connections are always secured with IPSEC, which has anti-replay 
> protection.
> - A timestamp negotiated connection starts with the assumption that PAWS is 
> necessary at the peer and timestamps every segment to begin with. After it 
> has an estimate of the RTT, it figures out (i) will fail because the maximum 
> transfer rate is limited by the bandwidth-delay product. - The sending
> application did a graceful close after submitting a buffer  which is smaller
> than what is takes to cover the sequence space. - A mail server admin knows
> each client has a mailbox size maximum of 1  gigabytes and there is no
> danger of wrap around when a mail client tries to  download it.

I am not trying to discourage you.  But please think about how to
write the above code and make it works every time reliably.  Then
test it and give us some results first.  Write up a draft and we
can have more fruitful discussions.

> My point is that, this is really something that can't be put in a spec, but 
> chosen by many factors.

Oops, without a spec, how are you going to tell people to employ it?

> (c) Again, determining whehter timestamping every segment for RTTM is an 
> overkill or not should be an implemenation choice. Examples being:
> 
> - The timer on the sender TCP has a granularity of 100 ms and the first 10 
> RTT measurements are all below 50 seconds.
> - The last 10 RTT measurements did not vary by more than 5%.
>
> 5. My suggestion was, if a sender TCP is able to figure this out, it can 
> selectively use timestamps. If the receiver also adheres to some well known 
> rule, it would save sending some extra bytes in either direction. The rule 
> is what I thought could be part of Eifel draft, as it was using timestamps 
> in a way other than RTTM and PAWS, and could be achieved wihtout 
> timestamping every segment. You might already have seen the following:
> Sender-TCP would put a timestamp in a segment if:
> * PAWS protection is necessary for the connection.
> * An RTT measurement using timestamps is desired, and one such measurement 
> is in progress. We consider a measurement to be in progress if sender has 
> sent a timestamp after it started a measurement and it is not yet echoed 
> back.
> * Retransmission of a segment upon timeout.
> 
> Receiver-TCP would put a timestamp (or echo what it got) in an ack if:
> * no more than 3 acks have been sent after any segment with a timestamp was 
> received.

							K. Poon.
							kcpoon@eng.sun.com



_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Thu Jan 10 23:35:25 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id XAA01444
	for <tsvwg-archive@odin.ietf.org>; Thu, 10 Jan 2002 23:35:25 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id XAA06402
	for tsvwg-archive@odin.ietf.org; Thu, 10 Jan 2002 23:35:27 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id XAA05894;
	Thu, 10 Jan 2002 23:21:21 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id XAA05864
	for <tsvwg@optimus.ietf.org>; Thu, 10 Jan 2002 23:21:10 -0500 (EST)
Received: from starburst.demon.co.uk (starburst.demon.co.uk [194.222.114.233])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id XAA00964
	for <tsvwg@ietf.org>; Thu, 10 Jan 2002 23:21:07 -0500 (EST)
Received: (from richard@localhost)
	by starburst.demon.co.uk (8.8.7/8.8.7) id EAA08940;
	Fri, 11 Jan 2002 04:18:50 GMT
From: Richard Wendland <richard@starburst.demon.co.uk>
Message-Id: <200201110418.EAA08940@starburst.demon.co.uk>
Subject: Re: [Tsvwg] Eifel draft and usage of time-stamps.
To: kcpoon@eng.sun.com
Date: Fri, 11 Jan 2002 04:18:49 +0000 (GMT)
Cc: skaniyar@hotmail.com (sanjay kaniyar), tsvwg@ietf.org,
        Reiner.Ludwig@ericsson.com
In-Reply-To: <Roam.SIMC.2.0.6.1010632369.25582.kcpoon@eng.sun.com> from "Kacheong Poon" at Jan 09, 2002 07:12:49 PM
Reply-to: richard@codeburst.co.uk
X-Mailer: ELM [version 2.5 PL2]
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Transfer-Encoding: 7bit
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit

> I don't know of
> any implementation which does not send timestamp option in every
> segment (besides, for example, RST) once it is negotiated at the
> beginning.

I do see some such behaviour, but it's not that common: timestamp on
SYN-ACK but not subsequently.  It's probably caused by devices like HTTP
load-balancers (possibly old or mis-configured), I don't see enough of
such behaviour to suggest it's due to a major TCP implementation.

Try HTTP connections to: www.us.army.mil, www.fairchildsemi.com,
www.goretex.com, www.westin.com, www.lclark.edu.

I also see the opposite behaviour, no timestamp on SYN-ACK but timestamps
on subsequent segments.  To see this try HTTP to: www.arl.army.mil,
www.bns.rutgers.edu, www.coolbean.com, www4.whirlpool.com.

But I doubt it is worth specially accommodating this strange behaviour.

NB I see this from a FreeBSD system with net.inet.tcp.rfc1323=1.

> And for the sake of the RTTM and PAWS, I believe it
> is safe to assume that if one side suddenly stops sending timestamp
> option, there must be something wrong and RTTM and PAWS will not
> be reliable.

That seems a fair judgment to me at present.

	Richard
-- 
Richard Wendland				richard@codeburst.co.uk

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Thu Jan 10 23:56:56 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id XAA01750
	for <tsvwg-archive@odin.ietf.org>; Thu, 10 Jan 2002 23:56:56 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id XAA06611
	for tsvwg-archive@odin.ietf.org; Thu, 10 Jan 2002 23:56:58 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id XAA06466;
	Thu, 10 Jan 2002 23:39:29 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id XAA06435
	for <tsvwg@optimus.ietf.org>; Thu, 10 Jan 2002 23:39:23 -0500 (EST)
Received: from patan.sun.com (patan.Sun.COM [192.18.98.43])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id XAA01483
	for <tsvwg@ietf.org>; Thu, 10 Jan 2002 23:39:21 -0500 (EST)
Received: from engmail1.Eng.Sun.COM ([129.146.1.13])
	by patan.sun.com (8.9.3+Sun/8.9.3) with ESMTP id VAA24420;
	Thu, 10 Jan 2002 21:39:23 -0700 (MST)
Received: from shield (shield.EBay.Sun.COM [129.150.113.84])
	by engmail1.Eng.Sun.COM (8.9.3+Sun/8.9.3/ENSMAIL,v2.1p1) with SMTP id UAA23478;
	Thu, 10 Jan 2002 20:39:22 -0800 (PST)
Date: Thu, 10 Jan 2002 20:38:41 -0800 (PST)
From: Kacheong Poon <kcpoon@eng.sun.com>
Reply-To: Kacheong Poon <kcpoon@eng.sun.com>
Subject: Re: [Tsvwg] Eifel draft and usage of time-stamps. 
To: mallman@grc.nasa.gov
Cc: Kacheong Poon <kcpoon@eng.sun.com>,
        Reiner Ludwig <Reiner.Ludwig@ericsson.com>,
        sanjay kaniyar <skaniyar@hotmail.com>, gurtov@cs.Helsinki.FI,
        tsvwg@ietf.org
In-Reply-To: "Your message with ID" <200201110408.XAA01325@lawyers.grc.nasa.gov>
Message-ID: <Roam.SIMC.2.0.6.1010723921.29558.kcpoon@eng.sun.com>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; CHARSET=US-ASCII
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

> What I meant was that if you know you took a spurious timeout (or
> did a needless fast retransmit) you can take that as an indication
> that your estimator is not working as well as you might like (i.e.,
> the RTO is not waiting long enough for an ACK to return).  So, you
> can tweak the estimator somehow (or alter the dup ack threshold
> somehow) in an attempt to come up with a better trigger for
> determining loss so that maybe you can prevent further spurious
> retransmits (and the resulting negative behavior).

This makes more sense.  Let's forget about fast retransmit for
the moment and concentrate on spurious timeout.  It seems to me
that one RTO algorithm cannot really handle many different network
characteristics.  For example, the MAX method suggested by Andrei
does not work well, IMHO, in a conventional network.  But it may
work well in his GPRS network.  As far as I know, all TCP 
implementations nowadays have only 1 RTO calculation algorithm builtin.
Does it make sense for TCP to completely switch to another RTO 
calculation algorithm given certain conditions, or given some hints from
outside, say from apps?

Reiner's Eifel algorithm (ignoring his Eifel RTO calculation for
the moment because I don't think it also works well for the RTT
spikes people are talking about here) chooses to stop the damage 
of a spurious timeout immediately.  To some people, it may make TCP 
very agrressive if the detection turns out to be false.  Looking at 
it more conservatively, it is possibly better for TCP to try to
avoid it, as Mark suggested.  But given the current RTO algorithm, I'm
afraid it may not be possible.   And if the alternate algorithm is
actually hinted by an app (or human) which knows exactly the network 
condition, it can avoid the spurious timeout altogether (is this
possible?).

Do people in the wireless world have something in their mind about
an appropriate RTO calculation algorithm which TCP can switch to
when it knows that it is OK to do that?

							K. Poon.
							kcpoon@eng.sun.com




_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Fri Jan 11 04:07:40 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id EAA12192
	for <tsvwg-archive@odin.ietf.org>; Fri, 11 Jan 2002 04:07:40 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id EAA22314
	for tsvwg-archive@odin.ietf.org; Fri, 11 Jan 2002 04:07:42 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id DAA21698;
	Fri, 11 Jan 2002 03:54:29 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id DAA21667
	for <tsvwg@optimus.ietf.org>; Fri, 11 Jan 2002 03:54:27 -0500 (EST)
Received: from mgw-x2.nokia.com (mgw-x2.nokia.com [131.228.20.22])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id DAA12101
	for <tsvwg@ietf.org>; Fri, 11 Jan 2002 03:54:24 -0500 (EST)
From: dajiang.zhang@nokia.com
Received: from esvir03nok.nokia.com (esvir03nokt.ntc.nokia.com [172.21.143.35])
	by mgw-x2.nokia.com (Switch-2.1.0/Switch-2.1.0) with ESMTP id g0B8sQ907156
	for <tsvwg@ietf.org>; Fri, 11 Jan 2002 10:54:26 +0200 (EET)
Received: from esebh01nok.ntc.nokia.com (unverified) by esvir03nok.nokia.com
 (Content Technologies SMTPRS 4.2.5) with ESMTP id <T586197de0bac158f23077@esvir03nok.nokia.com> for <tsvwg@ietf.org>;
 Fri, 11 Jan 2002 10:54:24 +0200
Received: by esebh01nok with Internet Mail Service (5.5.2652.78)
	id <CHN4BKTT>; Fri, 11 Jan 2002 10:54:24 +0200
Message-ID: <E373C6E4507012439877385BCEBEDD8724B42B@beebe001.NOE.Nokia.com>
To: tsvwg@ietf.org
Date: Fri, 11 Jan 2002 10:53:12 +0200
MIME-Version: 1.0
X-Mailer: Internet Mail Service (5.5.2652.78)
content-class: urn:content-classes:message
Content-Type: text/plain;
	charset="gb2312"
Subject: [Tsvwg] question about Ipv4 mapped addresses
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

Hi

I have a questions related to Ipv4 mapped addresses.

In "Sockets API Extensions for SCTP"
on Page 5, socket()
    The first form creates an endpoint which can use only IPv4
    addresses, while, the second form creates an endpoint which can use
    both IPv6 and IPv4 mapped addresses.
			^^^^^^^^^^^^^^^^^^^^^^
but on page 6, bind() 
    If sd is an IPv4 socket, the address passed must be an IPv4 address.
    If the sd is an IPv6 socket, the address passed can either be an
    IPv4 or an IPv6 address.
    ^^^^^

Why is there such a difference? Why use IPv4 mapped addresses instead of
Ipv4 address in socket()?

Would anyone give me an answer?
Thanks!

Dajiang Zhang

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Fri Jan 11 07:37:11 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id HAA13695
	for <tsvwg-archive@odin.ietf.org>; Fri, 11 Jan 2002 07:37:11 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id HAA28018
	for tsvwg-archive@odin.ietf.org; Fri, 11 Jan 2002 07:37:13 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id HAA27296;
	Fri, 11 Jan 2002 07:21:48 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id HAA27220
	for <tsvwg@optimus.ietf.org>; Fri, 11 Jan 2002 07:21:44 -0500 (EST)
Received: from lawyers.grc.nasa.gov (d141.as2.clev.oh.voyager.net [209.81.167.142])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id HAA13551
	for <tsvwg@ietf.org>; Fri, 11 Jan 2002 07:21:41 -0500 (EST)
Received: from lawyers (mallman@localhost)
	by lawyers.grc.nasa.gov (8.9.3/8.9.3) with ESMTP id HAA01958;
	Fri, 11 Jan 2002 07:20:53 -0500
Message-Id: <200201111220.HAA01958@lawyers.grc.nasa.gov>
To: Kacheong Poon <kcpoon@eng.sun.com>
From: Mark Allman <mallman@grc.nasa.gov>
Reply-To: mallman@grc.nasa.gov
cc: Reiner Ludwig <Reiner.Ludwig@ericsson.com>,
        sanjay kaniyar <skaniyar@hotmail.com>, gurtov@cs.Helsinki.FI,
        tsvwg@ietf.org
Subject: Re: [Tsvwg] Eifel draft and usage of time-stamps. 
Organization: BBN Technologies/NASA GRC
Song-of-the-Day: Lawyers, Guns and Money
Date: Fri, 11 Jan 2002 07:20:53 -0500
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org


> As far as I know, all TCP implementations nowadays have only 1 RTO
> calculation algorithm builtin.  Does it make sense for TCP to
> completely switch to another RTO calculation algorithm given
> certain conditions, or given some hints from outside, say from
> apps?

You wouldn't necessarily need to choose a different algorithm on the
fly.  You could do any number of "tweaks" to the current algorithm
to try to avoid future spurious timeouts.  One simple thing could be
to raise the minimum RTO a little each time you needlessly
retransmitted.  I haven't thought about the RTO case in great
detail.  (However, Ethan Blanton and I have been thinking about the
fast retransmit case quite a bit and I should be able to post a
paper that will appear in the Jan/2002 issue of CCR here in a day or
so.)

The current situation is that if you send a needless retransmit (and
hence add a packet to the network that accomplishes no new work) you
pay a price -- i.e., you slow down.  Now if we make TCP realize this
and undo things the price has been substantially reduced.  So, there
is really no incentive to try to be conservative and easy on the
network.  Therefore, it seems to me (and others I have chatted with)
that in addition to reverting to old CC state and generally being
more aggressive after you detect a spurious retransmit you also need
to take some sort of action to try to prevent future spurious
retransmits (whether that be to use a different RTO algorithm or
increase the dup ACK threshold or increase RTOmin or whatever).

allman

---
Mark Allman -- NASA GRC/BBN -- http://roland.grc.nasa.gov/~mallman/

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Fri Jan 11 08:32:34 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id IAA14286
	for <tsvwg-archive@odin.ietf.org>; Fri, 11 Jan 2002 08:32:34 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id IAA29625
	for tsvwg-archive@odin.ietf.org; Fri, 11 Jan 2002 08:32:37 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id IAA29025;
	Fri, 11 Jan 2002 08:17:11 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id IAA28994
	for <tsvwg@optimus.ietf.org>; Fri, 11 Jan 2002 08:17:09 -0500 (EST)
Received: from sj-msg-core-4.cisco.com (sj-msg-core-4.cisco.com [171.71.163.10])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id IAA14132
	for <tsvwg@ietf.org>; Fri, 11 Jan 2002 08:17:06 -0500 (EST)
Received: from mira-sjc5-2.cisco.com (mira-sjc5-2.cisco.com [171.71.163.16])
	by sj-msg-core-4.cisco.com (8.11.3/8.9.1) with ESMTP id g0BDGbJ26605;
	Fri, 11 Jan 2002 05:16:37 -0800 (PST)
Received: from stewart.chicago.il.us (ssh-sj1.cisco.com [171.68.225.134])
	by mira-sjc5-2.cisco.com (Mirapoint)
	with ESMTP id AAS07404;
	Fri, 11 Jan 2002 05:16:36 -0800 (PST)
Message-ID: <3C3EE5B4.D9F6F577@stewart.chicago.il.us>
Date: Fri, 11 Jan 2002 07:16:36 -0600
From: Randall Stewart <randall@stewart.chicago.il.us>
X-Mailer: Mozilla 4.76 [en] (X11; U; Linux 2.2.12 i386)
X-Accept-Language: en
MIME-Version: 1.0
To: dajiang.zhang@nokia.com
CC: tsvwg@ietf.org
Subject: Re: [Tsvwg] question about Ipv4 mapped addresses
References: <E373C6E4507012439877385BCEBEDD8724B42B@beebe001.NOE.Nokia.com>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Transfer-Encoding: 7bit
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit

Dajiang:

I don't know if I will have an answer :> And I am
sure some of my peers co-authors may disagree :-0

But I can comment on how Peter and I did ours :>>

dajiang.zhang@nokia.com wrote:
> 
> Hi
> 
> I have a questions related to Ipv4 mapped addresses.
> 
> In "Sockets API Extensions for SCTP"
> on Page 5, socket()
>     The first form creates an endpoint which can use only IPv4
>     addresses, while, the second form creates an endpoint which can use
>     both IPv6 and IPv4 mapped addresses.

When a IPv4 socket is opened IPv6 addresses are NOT allowed.
When a IPv6 socket is opened we DO allow IPv4 addresses to
be used... say on a sendmsg() call.

Now I have noticed that some of the KAME code will look
for a mapped address coming down and convert it to a
IPv4 address and then call the IPv4 routines. 

For our case we did not feel we needed to do this since all
of our routines were completely dual stacked. Now this may
be a arguable point and maybe we should have at least checked
all V6 address coming down the socket and if they were mapped
convert them to v4.. ... but for now we are not doing that :>

So in my view maybe the mapped wording can be deleted :>>

>                         ^^^^^^^^^^^^^^^^^^^^^^
> but on page 6, bind()
>     If sd is an IPv4 socket, the address passed must be an IPv4 address.
>     If the sd is an IPv6 socket, the address passed can either be an
>     IPv4 or an IPv6 address.
>     ^^^^^

Well for our case this is true of course since it takes V4 and V6
addresses... our only problem is we need to add code to reject
V6 addresses on V4 sockets.. since we don't really care :-0

We probably should think about doing that :>>

> 
> Why is there such a difference? Why use IPv4 mapped addresses instead of
> Ipv4 address in socket()?

I think the idea is so a given socket type only gets one type
of address on it.. but as far as I am concerned it should not
matter.. since under the covers you would just convert a
mapped address into a true V4 address why not allow the V4 address
to begin with... it makes sense to me :>

I think I more agree with you rather than answer you :)

R
-- 
Randall R. Stewart
randall@stewart.chicago.il.us 815-342-5222 (cell phone)

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Fri Jan 11 08:57:37 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id IAA14899
	for <tsvwg-archive@odin.ietf.org>; Fri, 11 Jan 2002 08:57:37 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id IAA00014
	for tsvwg-archive@odin.ietf.org; Fri, 11 Jan 2002 08:57:37 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id IAA29800;
	Fri, 11 Jan 2002 08:41:57 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id IAA29775
	for <tsvwg@optimus.ietf.org>; Fri, 11 Jan 2002 08:41:55 -0500 (EST)
Received: from hindon.hss.co.in ([202.54.26.202])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id IAA14379
	for <tsvwg@ietf.org>; Fri, 11 Jan 2002 08:41:51 -0500 (EST)
From: ygahlaut@hss.hns.com
Received: from sandesh.hss.hns.com (sandesh [139.85.242.35])
	by hindon.hss.co.in (8.10.0/8.10.0) with SMTP id g0BDgWE11664
	for <tsvwg@ietf.org>; Fri, 11 Jan 2002 19:12:32 +0530 (IST)
Received: by sandesh.hss.hns.com(Lotus SMTP MTA v4.6.3  (733.2 10-16-1998))  id 65256B3E.004B2D26 ; Fri, 11 Jan 2002 19:11:07 +0530
X-Lotus-FromDomain: HSS
To: tsvwg@ietf.org
Message-ID: <65256B3E.004B2C4D.00@sandesh.hss.hns.com>
Date: Fri, 11 Jan 2002 19:12:26 +0530
Mime-Version: 1.0
Content-type: text/plain; charset=us-ascii
Content-Disposition: inline
Subject: [Tsvwg] Section 7.1.3 sctpsocket-02.txt
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org



hi all,

  in section 7.1.3   the following structure is being used for returning the
SCTP_INITMSG option value

struct sctp_initmsg {
       uint16_t sinit_num_ostreams;
       uint16_t sinit_max_instreams;
       uint16_t sinit_max_attempts;
       uint16_t sinit_max_init_timeo;
    };

 Since it does not have any input parameter to differentiate in case of more
than
one association in UDP style socket .

can anybody explain how to handle this case ?

thanks
Yogesh





_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Fri Jan 11 10:31:43 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id KAA17852
	for <tsvwg-archive@odin.ietf.org>; Fri, 11 Jan 2002 10:31:42 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id KAA03002
	for tsvwg-archive@odin.ietf.org; Fri, 11 Jan 2002 10:31:44 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id KAA02287;
	Fri, 11 Jan 2002 10:13:08 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id KAA02260
	for <tsvwg@optimus.ietf.org>; Fri, 11 Jan 2002 10:13:06 -0500 (EST)
Received: from sj-msg-core-3.cisco.com (sj-msg-core-3.cisco.com [171.70.157.152])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id KAA17039
	for <tsvwg@ietf.org>; Fri, 11 Jan 2002 10:13:04 -0500 (EST)
Received: from mira-sjc5-2.cisco.com (mira-sjc5-2.cisco.com [171.71.163.16])
	by sj-msg-core-3.cisco.com (8.11.3/8.9.1) with ESMTP id g0BFC9k28506;
	Fri, 11 Jan 2002 07:12:11 -0800 (PST)
Received: from stewart.chicago.il.us (ssh-sj1.cisco.com [171.68.225.134])
	by mira-sjc5-2.cisco.com (Mirapoint)
	with ESMTP id AAS08605;
	Fri, 11 Jan 2002 07:12:32 -0800 (PST)
Message-ID: <3C3F00DF.2D531C5A@stewart.chicago.il.us>
Date: Fri, 11 Jan 2002 09:12:32 -0600
From: Randall Stewart <randall@stewart.chicago.il.us>
X-Mailer: Mozilla 4.76 [en] (X11; U; Linux 2.2.12 i386)
X-Accept-Language: en
MIME-Version: 1.0
To: ygahlaut@hss.hns.com
CC: tsvwg@ietf.org
Subject: Re: [Tsvwg] Section 7.1.3 sctpsocket-02.txt
References: <65256B3E.004B2C4D.00@sandesh.hss.hns.com>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Transfer-Encoding: 7bit
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit


Yogesh:

Two things:

First:

The sctp_initmsg is used in the UDP model to start
a new association with a peer (i.e. you are sending the
INIT message out) and over-ride the default INIT parameters
with the value found in the sctp_initmsg structure.

So we have an app sending a message with a
sendmsg() and putting the sctp_initmsg structure
in a CMSG header to go with the first message you
are sending. The sendmsg() call also includes the
destination address in the msg_name field ... i.e. the
sockaddr * of who you are sending (and INIT'ing) to.

Second:

If you are asking instead about notifications, i.e.
the association just came up and how do I tell what
parameters were negotiated in the streams counts
etc.. you get this in a MSG_NOTIFICATION in the
form of a 

struct sctp_assoc_change x;

And it too is received via a recvmsg() call where
the msg_flags field gets set to MSG_NOTIFICATION and the
msg_name will have the peer that the association just came
up with .. the data read will be the  sctp_assoc_change..

Of course you must enable this event...

Hope that helps...

R

ygahlaut@hss.hns.com wrote:
> 
> hi all,
> 
>   in section 7.1.3   the following structure is being used for returning the
> SCTP_INITMSG option value
> 
> struct sctp_initmsg {
>        uint16_t sinit_num_ostreams;
>        uint16_t sinit_max_instreams;
>        uint16_t sinit_max_attempts;
>        uint16_t sinit_max_init_timeo;
>     };
> 
>  Since it does not have any input parameter to differentiate in case of more
> than
> one association in UDP style socket .
> 
> can anybody explain how to handle this case ?
> 
> thanks
> Yogesh
> 
> _______________________________________________
> tsvwg mailing list
> tsvwg@ietf.org
> http://www1.ietf.org/mailman/listinfo/tsvwg

-- 
Randall R. Stewart
randall@stewart.chicago.il.us 815-342-5222 (cell phone)

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Fri Jan 11 14:49:15 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id OAA27242
	for <tsvwg-archive@odin.ietf.org>; Fri, 11 Jan 2002 14:49:15 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id OAA13417
	for tsvwg-archive@odin.ietf.org; Fri, 11 Jan 2002 14:49:05 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id OAA12814;
	Fri, 11 Jan 2002 14:27:47 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id OAA12783
	for <tsvwg@optimus.ietf.org>; Fri, 11 Jan 2002 14:27:45 -0500 (EST)
Received: from mgw-x2.nokia.com (mgw-x2.nokia.com [131.228.20.22])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id OAA26750
	for <tsvwg@ietf.org>; Fri, 11 Jan 2002 14:27:41 -0500 (EST)
From: Ivan.Arias-Rodriguez@nokia.com
Received: from esvir02nok.ntc.nokia.com (esvir02nokt.ntc.nokia.com [172.21.143.34])
	by mgw-x2.nokia.com (Switch-2.1.0/Switch-2.1.0) with ESMTP id g0BJRh912623
	for <tsvwg@ietf.org>; Fri, 11 Jan 2002 21:27:43 +0200 (EET)
Received: from esebh03nok.ntc.nokia.com (unverified) by esvir02nok.ntc.nokia.com
 (Content Technologies SMTPRS 4.2.5) with ESMTP id <T5863dbaa4cac158f22141@esvir02nok.ntc.nokia.com> for <tsvwg@ietf.org>;
 Fri, 11 Jan 2002 21:27:42 +0200
Received: by esebh03nok with Internet Mail Service (5.5.2652.78)
	id <C13HM529>; Fri, 11 Jan 2002 21:27:42 +0200
Message-ID: <58AEDE27D48F884285007ABF827FF26308EE1E@esebe017.NOE.Nokia.com>
To: tsvwg@ietf.org
Date: Fri, 11 Jan 2002 21:27:35 +0200
MIME-Version: 1.0
X-Mailer: Internet Mail Service (5.5.2652.78)
content-class: urn:content-classes:message
Content-Type: text/plain;
	charset="iso-8859-1"
Subject: [Tsvwg] Lots of comments about sctpsocket-02...
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

	Hi all!

	I have been checking the draft and I have many
comments/doubts/typos caught/proposed changes... As I'm not an expert on
these issues, some of them might be completely wrong... Excuse me also
if some have been already commented in the list... Here they are:

	Section 3.1.2, last paragraph: It says that when binding, "One
of those addresses will be the primary address for the association"...
Uhm... Well, it will be the *peers* primary address , not ours... Maybe
it could say "The peer will chose one of those addresses as its primary
address for the association", or maybe nothing at all... This same
comment applies to Section 4.1.2, first paragraph of page 12, and to
Section 4.1.4, third paragraph of page 13.

	- Section 3.2, first paragraph: I don't understand how an
application can use recvmsg() to implicitly open a connection... It does
not carry any peer's address information... so? :-/
	Also, a client application does not usually use the bind() call,
and thus it is not necessary that "all bind() calls are complete".
	When I read it first time I got the impression that the UDP
style sockets could also use connect()... Maybe a "as is the case for
TCP Style sockets" could be added right after the "(i.e., no connect()
calls required".

	- Section 3.3, second paragraph: A typo, "Whenever the user
which want to" should be "Whenever the user which want*s* to".

	- Section 4.1.2, first paragraph of page 12: Another typo,
"equivalant" should be "equivalent".

	- Section 5.1, last paragraph of page 16: "Sections 3.1.3"
should be "Sections 3.1.4".

	- Section 5.2.1: Wouldn't it be interesting to add a new field
to the struct sctp_initmsg containing the cookie life? (maybe two of
them, the "standard" cookie life and the "extended" cookie life, in case
we receive a "Cookie Preservative" parameter).
	Other fields I can think about are: Max Path/Total
retransmissions, initial/Min/Alpha/Beta RTO, Heartbeats allowed,
heartbeat interval, or even if bundling is allowed or not. These values
must be fixed to some default values, and maybe it would be nice to be
able to modify those values since the very beginning. Some of them can
be modified later by calling setsockopt() or sctp_opt_info. However, the
cookie life, the Alpha/Beta values of RTO (would we like to modify
them?), enable/disable bundling, cannot be modified (and to
enable/disable heartbeats we have to go address by address)... Note also
that, while the initial RTO can be modified later, we cannot control its
value during the INIT/COOKIE ECHO retransmissions...
	And what about being able to set the value of the
inbound/outbound buffers?
	Another little issue. I guess that sinit_max_attempsts is
measured in milliseconds, but that's not stated.

	- Section 5.2.2, description of sinfo_stream: "an error
indication is returned", which one?
	The description of sinfo_flags appears out of order.
	In the first paragraph of page 20, "multple" should be
"multiple".
	In the sinfo_timetolive description: "if the message as not been
sent" should be "if the message *has* not been sent".

	- Section 5.3.1.1, beginning of page 23: "provided by this
notificaion" should be "provided by this notification".
	In the description of the sac_state parameter, it should say "16
bits (unsigned integer)".
	In the description of sac_error, is it set to the "Cause Code"
of the Error Cause received? (this also applies to the description of
ssf_error in Section 5.3.1.4)
	Also, it says there that it is a signed 32-bit number, but it is
defined as a uint16_t value. In Section 5.3.1.4 it happens the same with
ssf_error, but just the opposite.
	By the way, the descriptions of the sac/spc/sre/ssf/sse/sai_type
don't include the statement "16 bits (unsigned integer)".
	In that same paragraph "a error condition" should be "an error
condition".

	- Section 5.3.1.2: In the definition of the struct
sctp_paddr_change, I guess that the spc_aaddr field should be spc_addr
instead. At least it is what appears in the example of the appendix.
	In the description of spc_error, how is the relevant error
information expressed?

	- Section 5.3.1.5, in the definition of the struct
sctp_shutdown_event. The sse_flags field appears twice, both in the
definition and in the explanation of the fields.

	- Section 5.3.1.6. Why "ADAPTION" instead of "ADAPTATION"?...
I'm not sure of this. There are many other "adaption"s in the draft.
	In the first paragraph, "sends a Adaption" should be "sends an
Adaption". In any case, I don't understand the paragraph itself. Could
somebody explain me what is the use of this notification? What is an
Adaption Layer Indication parameter? Are them simply INIT/INIT ACK
parameters? In that case, when should they be passed to the application?

	- Section 6.1, page 31, it says "flags  - (described below)",
where?

	- Section 6.2, first line of page 32, "descript" should be
"descriptor".

	- Section 7, second paragraph, "the the".

	- Section 7.1.1, in the definition of the struct sctp_rtoinfo:
Could it be interesting including here srto_alpha and srto_beta?

	- Section 7.1.2, second paragraph starting from the end of the
page 33: "To access or modify these parameters," I guess is a cut n
paste mistake, there is only one parameter, so it would be better "To
access or modify this parameter,".
	In the last paragraph of page 33, "seperate" should be
"separate".

	- Section 7.1.3, second paragraph: "effected" should be
"affected".

	- Section 7.1.5: Then, if the integer is set to 1, the Nagle
algorithm is disabled, right?

	- Section 7.1.8: "An association being idle is defined an
association that has NOT sent or received user data." :-/... Wouldn't it
be better something like "An association is said to be idle for a lapse
of time if it has NOT sent nor received user data during that lapse of
time."? Or something like this...

	- Section 7.1.11: Apart from the Adaption vs. Adaptation issue,
I fail to understand what an Adaption Layer Indication parameter is...
:-0. If it is a parameter, why it is restricted to be 32-bits long?

	- Section 7.2.1, page 37: In the description of sstat_rwnd, is
that value the total peer's receiver window size, or just the unused
space of its receiver buffer? The word "current" leads me to the second
interpretation, but the length of the receiver buffer itself can be
modified, so...

	- Section 8.1: Isn't there any way to specify "all the
addresses"? I mean, a wildcard such as INADDR_ANY, but meaning all of
them, not just one...
	About SCTP_BINDX_ADD_ADDR and SCTP_BINDX_REM_ADDR... It says
"The flags parameter is formed from the bitwise OR of zero or more of
the following...", but later on it says "The two flags are mutually
exclusive"... What am I missing?
	In that same parameter, it says that we can add/remove addresses
from the association... But, if we are binding, which association is
going to be established? Moreover, there isn't any parameter in the
sctp_bindx() call indicating the association... so...

	- Section 8.2: This is not so clear for me... if we branch off
an UDP Style association, does it become a TCP Style one?
	In the description of addrlen it makes a reference to the addr
parameter, but that parameter does not exists...

	- Section 8.3: To keep the same naming conventions thorough the
whole document, I think that the "id" parameter of sctp_getpaddrs could
be "assoc_id".
	For the same reason, in sctp_getladdrs (section 8.5), "sock"
could be modified to be "sd", "id" to be "assoc_id" and "ss" to be
"addrs" (note that the first and last changes are necessary to keep the
consistence with the later description of the parameters).
	Again, in sctp_opt_info, "id" could be changed to be "assoc_id".
	Some other changes that I have just seen. In sendmsg and recvmsg
(Section 3.1.4), "socket" could be changed to be "sd". The same in
shutdown() (Section 4.1.7), getsockname() (Section 4.1.9). In
getsockname(), the definition "socklen_t *len" could be changed to "int
*addrlen (not sure of this). Same for getpeername() (Section 4.1.10)
	Of course, all these changes must be done both in the definition
of the function and in the description of the parameters.

	- Section 8.7: I fail to see why we need another function, and
we can't use getsockopt()... I have the idea that getsockopt is for the
whole association, and sctp_opt_info refers to a single address, and
thus, to keep getsockopt compatible with TCP implementations, you don't
want to modify it so it can carry an address parameter. However, I don't
understand the explanation given in the first paragraph of the section.
	It says later "See 8.5 subsections...", I guess it should be
"See 8.7 subsections..."

	- Section 8.7.1: Uhm... the Heartbeat Interval is a parameter
that affects all the addresses (see page 108 of RFC 2960). So I think
that it shouldn't appear here. I think it should appear as a parameter
in the struct sctp_assocparams defined in Section 7.1.2.
	In the description of spp_assoc_id, "This is filled in the
application", shouldn't it be "This is filled in *by* the application"?
This same sentence appears in Section 8.7.2, in the description of
spinfo_assoc_id...

	- Section 8.7.2: The word "addresses's" appears several times...
Not completely sure of this, but, shouldn't it be "addresses'"?

	- Section 9, first paragraph of page 43: "unprivelged" should be
"unprivileged".

	- Section 12: Well, the [STEVENS] reference is wrong... There is
already a 03 version, published on November 2001...

	- Appendix A, page 45: In the "case SCTP_REMOTE_ERROR", the
field sre_len of the sre variable (struct sctp_remote_error) does not
exist...
	

	As a final comment, the indentation in the struct definitions is
sometimes done with tabulations, and some other with spaces, so here in
my windows machine they don't coincide... If after such a long list of
changes you still feel like modifying this, its up to you :-)

	Thanks for your time!

	BR Iv�n Arias Rodr�guez

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Fri Jan 11 16:50:08 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id QAA00797
	for <tsvwg-archive@odin.ietf.org>; Fri, 11 Jan 2002 16:50:08 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id QAA17594
	for tsvwg-archive@odin.ietf.org; Fri, 11 Jan 2002 16:50:10 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id QAA16668;
	Fri, 11 Jan 2002 16:26:59 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id QAA16637
	for <tsvwg@optimus.ietf.org>; Fri, 11 Jan 2002 16:26:57 -0500 (EST)
Received: from mgw-dax2.ext.nokia.com (mgw-dax2.ext.nokia.com [63.78.179.217])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id QAA00152
	for <tsvwg@ietf.org>; Fri, 11 Jan 2002 16:26:50 -0500 (EST)
Received: from davir02nok.americas.nokia.com (davir02nok.americas.nokia.com [172.18.242.85])
	by mgw-dax2.ext.nokia.com (Switch-2.1.0/Switch-2.1.0) with ESMTP id g0BLSJQ13796
	for <tsvwg@ietf.org>; Fri, 11 Jan 2002 15:28:19 -0600 (CST)
Received: from daebh002.NOE.Nokia.com (unverified) by davir02nok.americas.nokia.com
 (Content Technologies SMTPRS 4.2.5) with ESMTP id <T58629146efac12f2550ef@davir02nok.americas.nokia.com> for <tsvwg@ietf.org>;
 Fri, 11 Jan 2002 15:26:50 -0600
Received: from mvebe001.NOE.Nokia.com ([172.18.140.37]) by daebh002.NOE.Nokia.com with Microsoft SMTPSVC(5.0.2195.2966);
	 Fri, 11 Jan 2002 15:26:49 -0600
X-MimeOLE: Produced By Microsoft Exchange V6.0.4712.0
content-class: urn:content-classes:message
MIME-Version: 1.0
Content-Type: text/plain;
	charset="iso-8859-1"
Date: Fri, 11 Jan 2002 13:26:48 -0800
Message-ID: <4D7B558499107545BB45044C63822DDE0F813B@mvebe001.NOE.Nokia.com>
Thread-Topic: section 5.3 draft-ietf-tsvwg-sctpsocket-02.txt
Thread-Index: AcGa5phX/itI7AZ4EdaaKwAAhkzCIw==
From: "Sud Shivani (NET/MtView)" <Shivani.Sud@nokia.com>
To: <tsvwg@ietf.org>
X-OriginalArrivalTime: 11 Jan 2002 21:26:50.0010 (UTC) FILETIME=[AE468FA0:01C19AE6]
Content-Transfer-Encoding: 8bit
X-MIME-Autoconverted: from quoted-printable to 8bit by optimus.ietf.org id QAA16638
Subject: [Tsvwg] section 5.3 draft-ietf-tsvwg-sctpsocket-02.txt
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 8bit

In the draft-ietf-tsvwg-sctpsocket-02.txt :
In Section 5.3cit says:
> 5.3 SCTP Events and Notifications 
> .........
> Multiple notifications may be returned to a single recvmsg() call. 
> ....
> A recvmsg() call will return only one notification at a time. 

From the rest of the text it is inferred that multiple
notifications/events can be received in any order - in a single recvmsg.
So probably the second line should go.

Also in section 6.1 :
6.1 send(), recv(), sendto(), recvfrom() 
.....
flags - (described below). 
....
The flags are not described below - and since no new flags are defined I
assume it is should read - no new flags are defined for SCTP for these
operations.


_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Fri Jan 11 18:15:41 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id SAA02740
	for <tsvwg-archive@odin.ietf.org>; Fri, 11 Jan 2002 18:15:41 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id SAA20613
	for tsvwg-archive@odin.ietf.org; Fri, 11 Jan 2002 18:15:44 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id RAA20035;
	Fri, 11 Jan 2002 17:55:34 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id RAA19991
	for <tsvwg@optimus.ietf.org>; Fri, 11 Jan 2002 17:55:32 -0500 (EST)
Received: from patan.sun.com (patan.Sun.COM [192.18.98.43])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id RAA02516
	for <tsvwg@ietf.org>; Fri, 11 Jan 2002 17:55:28 -0500 (EST)
Received: from engmail1.Eng.Sun.COM ([129.146.1.13])
	by patan.sun.com (8.9.3+Sun/8.9.3) with ESMTP id PAA02035;
	Fri, 11 Jan 2002 15:55:31 -0700 (MST)
Received: from shield (shield.EBay.Sun.COM [129.150.113.84])
	by engmail1.Eng.Sun.COM (8.9.3+Sun/8.9.3/ENSMAIL,v2.1p1) with SMTP id OAA06784;
	Fri, 11 Jan 2002 14:55:31 -0800 (PST)
Date: Fri, 11 Jan 2002 14:54:50 -0800 (PST)
From: Kacheong Poon <kcpoon@eng.sun.com>
Reply-To: Kacheong Poon <kcpoon@eng.sun.com>
Subject: Re: [Tsvwg] question about Ipv4 mapped addresses
To: dajiang.zhang@nokia.com
cc: tsvwg@ietf.org
In-Reply-To: "Your message with ID" <E373C6E4507012439877385BCEBEDD8724B42B@beebe001.NOE.Nokia.com>
Message-ID: <Roam.SIMC.2.0.6.1010789690.5503.kcpoon@eng.sun.com>
Content-Type: text
X-Sun-Text-Type: ascii
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

> but on page 6, bind() 
>     If sd is an IPv4 socket, the address passed must be an IPv4 address.
>     If the sd is an IPv6 socket, the address passed can either be an
>     IPv4 or an IPv6 address.
>     ^^^^^
> 
> Why is there such a difference? Why use IPv4 mapped addresses instead of
> Ipv4 address in socket()?

I guess we should correct the above to use mapped addresses also.
The goal is to make things clean so that for v6 socket, one should
not pass in a v4 address structure.

Thanks for pointing this out.

							K. Poon.
							kcpoon@eng.sun.com


_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Fri Jan 11 19:33:51 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id TAA03739
	for <tsvwg-archive@odin.ietf.org>; Fri, 11 Jan 2002 19:33:51 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id TAA22535
	for tsvwg-archive@odin.ietf.org; Fri, 11 Jan 2002 19:33:52 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id TAA21954;
	Fri, 11 Jan 2002 19:16:45 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id TAA21926
	for <tsvwg@optimus.ietf.org>; Fri, 11 Jan 2002 19:16:43 -0500 (EST)
Received: from mercury.Sun.COM (mercury.Sun.COM [192.9.25.1])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id TAA03436
	for <tsvwg@ietf.org>; Fri, 11 Jan 2002 19:16:40 -0500 (EST)
Received: from engmail1.Eng.Sun.COM ([129.146.1.13])
	by mercury.Sun.COM (8.9.3+Sun/8.9.3) with ESMTP id QAA08395;
	Fri, 11 Jan 2002 16:16:41 -0800 (PST)
Received: from shield (shield.EBay.Sun.COM [129.150.113.84])
	by engmail1.Eng.Sun.COM (8.9.3+Sun/8.9.3/ENSMAIL,v2.1p1) with SMTP id QAA02689;
	Fri, 11 Jan 2002 16:16:40 -0800 (PST)
Date: Fri, 11 Jan 2002 16:15:59 -0800 (PST)
From: Kacheong Poon <kcpoon@eng.sun.com>
Reply-To: Kacheong Poon <kcpoon@eng.sun.com>
Subject: Re: [Tsvwg] Eifel draft and usage of time-stamps. 
To: mallman@grc.nasa.gov
Cc: Reiner Ludwig <Reiner.Ludwig@ericsson.com>,
        sanjay kaniyar <skaniyar@hotmail.com>, gurtov@cs.Helsinki.FI,
        tsvwg@ietf.org
In-Reply-To: "Your message with ID" <200201111220.HAA01958@lawyers.grc.nasa.gov>
Message-ID: <Roam.SIMC.2.0.6.1010794559.13123.kcpoon@eng.sun.com>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; CHARSET=US-ASCII
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

> You wouldn't necessarily need to choose a different algorithm on the
> fly.  You could do any number of "tweaks" to the current algorithm
> to try to avoid future spurious timeouts.  One simple thing could be
> to raise the minimum RTO a little each time you needlessly
> retransmitted.  I haven't thought about the RTO case in great

Raise RTOmin to 5s then (-:  At least to me, the usefulness of
RTT sampling and RTO calculation as described in RFC 2988 decreases
as the RTOmin keeps increasing as it just masks everything!  The 
point I am trying to raise is that the RTO algorithm in RFC 2988 
should not be the end of all (Reiner should agree (-:).  And I feel,
just me (-:, that other algorithms are probably needed in different
situations.  For example, the basic assumption that the round trip 
route stays pretty much the same does not universally hold true 
nowadays.  This is why I hope researchers in this list can give
some advice.

> detail.  (However, Ethan Blanton and I have been thinking about the
> fast retransmit case quite a bit and I should be able to post a
> paper that will appear in the Jan/2002 issue of CCR here in a day or
> so.)

Do you have evidence that the problem related to fast retransmit
is significant?  The problem with timeout is very clear and people 
are seeing it very often in the wireless world now.  I guess it 
needs more attention and should be addressed asap...  But please do
update the list on your findings.

> The current situation is that if you send a needless retransmit (and
> hence add a packet to the network that accomplishes no new work) you
> pay a price -- i.e., you slow down.  Now if we make TCP realize this
> and undo things the price has been substantially reduced.  So, there
> is really no incentive to try to be conservative and easy on the
> network.  Therefore, it seems to me (and others I have chatted with)

I don't understand the comments above.  If TCP can know for sure
that it just has done a spurious retransmission, it should undo
it.  I agree totally.  But it does not mean that there is no 
incentive to try to be conservative.  At least to some people, the 
issue with it is that what if TCP makes mistakes too often and does
not react to real network problems as it does today.  If all routers
now "punish" overly aggressive TCP traffic, the incentive to be
nice is ample (-:

> that in addition to reverting to old CC state and generally being
> more aggressive after you detect a spurious retransmit you also need
> to take some sort of action to try to prevent future spurious
> retransmits (whether that be to use a different RTO algorithm or
> increase the dup ACK threshold or increase RTOmin or whatever).

I agree with the above.  But do they really need to go together?
That's my question.  They can be deployed separately.  My issue
with new algorithms is that we need time to really see what they
are doing.  And if there are multiple things going on, some
effects may be masked and we cannot correctly determine some
underlying problems.

							K. Poon.
							kcpoon@eng.sun.com



_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Fri Jan 11 19:35:19 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id TAA03758
	for <tsvwg-archive@odin.ietf.org>; Fri, 11 Jan 2002 19:35:19 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id TAA22558
	for tsvwg-archive@odin.ietf.org; Fri, 11 Jan 2002 19:35:20 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id TAA21909;
	Fri, 11 Jan 2002 19:14:32 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id TAA21878
	for <tsvwg@optimus.ietf.org>; Fri, 11 Jan 2002 19:14:30 -0500 (EST)
Received: from hotmail.com (f85.law3.hotmail.com [209.185.241.85])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id TAA03428
	for <tsvwg@ietf.org>; Fri, 11 Jan 2002 19:14:28 -0500 (EST)
Received: from mail pickup service by hotmail.com with Microsoft SMTPSVC;
	 Fri, 11 Jan 2002 16:13:59 -0800
Received: from 131.107.3.92 by lw3fd.law3.hotmail.msn.com with HTTP;
	Sat, 12 Jan 2002 00:13:59 GMT
X-Originating-IP: [131.107.3.92]
From: "sanjay kaniyar" <skaniyar@hotmail.com>
To: kcpoon@eng.sun.com
Cc: tsvwg@ietf.org, Reiner.Ludwig@ericsson.com
Subject: Re: [Tsvwg] Eifel draft and usage of time-stamps.
Date: Fri, 11 Jan 2002 16:13:59 -0800
Mime-Version: 1.0
Content-Type: text/plain; format=flowed
Message-ID: <F85GQvnbEG10Gai4omR00000f9a@hotmail.com>
X-OriginalArrivalTime: 12 Jan 2002 00:13:59.0799 (UTC) FILETIME=[087EE870:01C19AFE]
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

1. You seem to be responding to something that is not even remotely implied 
by anything I said. Going over the mail thread again would help.(I am not 
trying to use unidirectional timestamps, wonder what made you think so!!).

I agree with your point in the second pragraph, and there could very well be 
interop issues.

3 & 4.
(a) Same as 1 above. What made you think I suggested a way to figure out 
timestmaps are needed or not at connection setup time?

Sanjay.


>From: Kacheong Poon <kcpoon@eng.sun.com>
>Reply-To: Kacheong Poon <kcpoon@eng.sun.com>
>To: sanjay kaniyar <skaniyar@hotmail.com>
>CC: tsvwg@ietf.org, Reiner.Ludwig@ericsson.com
>Subject: Re: [Tsvwg] Eifel draft and usage of time-stamps.
>Date: Thu, 10 Jan 2002 20:08:44 -0800 (PST)
>
> > 1. Timestamps are bidirectional in the sense that in a segment going 
>from
> > A->B, there are 2 of them: 1:generated by A, 2:echoed value, sent by B
> > originally. If the connection is doing unidirectional data transfer
> > entirely, it is sufficient to have timestamps generated and sent by the
> > sender in the data segments and acknowledgements echo them in the other
> > direction. This is what RFC 1323 referred to as 'unidirectional use of
> > timestamps'.
> >
> > They chose to 'always use timestamps bidirectionally' instead because of
> > reasons like simplicity etc. There is nothing new I am proposing here. I
> > only quoted this to rebut your claim - 'always refers sending timestamps 
>in
> > each segment'.
>
>Correct me if I am wrong, your problem with the current timestamp
>option is that it takes up 12 bytes.  Now in your "unidirectional"
>timestamp scenario, isn't it true that the 12 bytes timestamp option
>is still in every segments in both directions even if only the data
>sender chooses to use it?  Note that I never mentioned that "unidirectional
>timestamp option" is not possible.  I am trying to say that the current
>timestamp option, with both time and echo value in it, works in a
>specific way.  It is designed to be used for both sides.  You can come up
>with your own "unidirectional timestamp option."  But it is just not the
>same as RFC 1323....  Afterall, you said that in a previous email,
>"I interpret it as 'don't use timestmaps in a unidirectional way'."  So
>are you now saying that RFC 1323 allows you to use timestamps in
>"undirectional way?"
>
>And think more about interoperability (you can ignore what I said about
>Solaris for this purpose).  The fact is that all stacks nowadays send
>timestamp option is every segments once it is negotiated.  And the fact
>that a TCP stack will never know how an app is going to use it.  And ...
>There are a lot of issues to think about.
>
> > 3 & 4.
> > (a) I did answer your question. I never said it is always possible to 
>figure
> > out whether it is possible to figure out whether timestamps are 
>necessary at
> > the connection set up time (nor is it the problem I am trying to solve). 
>All
> > I am trying to do is to timestamp (and echo) segments on a connection 
>which
> > has timestamps negotiated on it only when it is necessary and useful, I 
>said
> > nothing about figuruing out whenther they should be negotiated or not.
>
>So are you suggesting something that you don't even know how to make
>it work?
>
> > (b) I know that to do PAWS check, every segment needs to be timstamped.
> > However, to figure out whether PAWS check needs to be done or not, can 
>be
> > implementation specific because there can be many reasons within the 
>perview
> > of TCP or outside of it which could contribute to this. Note that, for a
> > connection to require PAWS check it has to (i) have a transfer rate of
> > greater than a threshold and (ii) have a life time greater than a 
>threhold.
> > If it is possible to determine one of these is not going to be met, 
>there is
> > no need to do PAWS on that connection. The maybe's were actually 
>examples of
> > such instances:
> >
> > - A bandwidth limit is imposed on each connection {(i) fails}
> > - Connections are always secured with IPSEC, which has anti-replay
> > protection.
> > - A timestamp negotiated connection starts with the assumption that PAWS 
>is
> > necessary at the peer and timestamps every segment to begin with. After 
>it
> > has an estimate of the RTT, it figures out (i) will fail because the 
>maximum
> > transfer rate is limited by the bandwidth-delay product. - The sending
> > application did a graceful close after submitting a buffer  which is 
>smaller
> > than what is takes to cover the sequence space. - A mail server admin 
>knows
> > each client has a mailbox size maximum of 1  gigabytes and there is no
> > danger of wrap around when a mail client tries to  download it.
>
>I am not trying to discourage you.  But please think about how to
>write the above code and make it works every time reliably.  Then
>test it and give us some results first.  Write up a draft and we
>can have more fruitful discussions.
>
> > My point is that, this is really something that can't be put in a spec, 
>but
> > chosen by many factors.
>
>Oops, without a spec, how are you going to tell people to employ it?
>
> > (c) Again, determining whehter timestamping every segment for RTTM is an
> > overkill or not should be an implemenation choice. Examples being:
> >
> > - The timer on the sender TCP has a granularity of 100 ms and the first 
>10
> > RTT measurements are all below 50 seconds.
> > - The last 10 RTT measurements did not vary by more than 5%.
> >
> > 5. My suggestion was, if a sender TCP is able to figure this out, it can
> > selectively use timestamps. If the receiver also adheres to some well 
>known
> > rule, it would save sending some extra bytes in either direction. The 
>rule
> > is what I thought could be part of Eifel draft, as it was using 
>timestamps
> > in a way other than RTTM and PAWS, and could be achieved wihtout
> > timestamping every segment. You might already have seen the following:
> > Sender-TCP would put a timestamp in a segment if:
> > * PAWS protection is necessary for the connection.
> > * An RTT measurement using timestamps is desired, and one such 
>measurement
> > is in progress. We consider a measurement to be in progress if sender 
>has
> > sent a timestamp after it started a measurement and it is not yet echoed
> > back.
> > * Retransmission of a segment upon timeout.
> >
> > Receiver-TCP would put a timestamp (or echo what it got) in an ack if:
> > * no more than 3 acks have been sent after any segment with a timestamp 
>was
> > received.
>
>							K. Poon.
>							kcpoon@eng.sun.com
>
>
>
>_______________________________________________
>tsvwg mailing list
>tsvwg@ietf.org
>http://www1.ietf.org/mailman/listinfo/tsvwg




_________________________________________________________________
Chat with friends online, try MSN Messenger: http://messenger.msn.com


_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Sat Jan 12 00:23:05 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id AAA08632
	for <tsvwg-archive@odin.ietf.org>; Sat, 12 Jan 2002 00:23:05 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id AAA28719
	for tsvwg-archive@odin.ietf.org; Sat, 12 Jan 2002 00:23:05 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id AAA28182;
	Sat, 12 Jan 2002 00:00:24 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id AAA28142
	for <tsvwg@optimus.ietf.org>; Sat, 12 Jan 2002 00:00:21 -0500 (EST)
Received: from lawyers.grc.nasa.gov (d16.as2.clev.oh.voyager.net [209.81.167.17])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id AAA08470
	for <tsvwg@ietf.org>; Sat, 12 Jan 2002 00:00:15 -0500 (EST)
Received: from lawyers (mallman@localhost)
	by lawyers.grc.nasa.gov (8.9.3/8.9.3) with ESMTP id XAA03634;
	Fri, 11 Jan 2002 23:46:51 -0500
Message-Id: <200201120446.XAA03634@lawyers.grc.nasa.gov>
To: Kacheong Poon <kcpoon@eng.sun.com>
From: Mark Allman <mallman@grc.nasa.gov>
Reply-To: mallman@grc.nasa.gov
cc: Reiner Ludwig <Reiner.Ludwig@ericsson.com>,
        sanjay kaniyar <skaniyar@hotmail.com>, gurtov@cs.helsinki.fi,
        tsvwg@ietf.org
Subject: Re: [Tsvwg] Eifel draft and usage of time-stamps. 
Organization: BBN Technologies/NASA GRC
Song-of-the-Day: Hammer to Fall
Date: Fri, 11 Jan 2002 23:46:51 -0500
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org


> Raise RTOmin to 5s then (-: At least to me, the usefulness of RTT
> sampling and RTO calculation as described in RFC 2988 decreases as
> the RTOmin keeps increasing as it just masks everything!  The
> point I am trying to raise is that the RTO algorithm in RFC 2988
> should not be the end of all (Reiner should agree (-:).  And I
> feel, just me (-:, that other algorithms are probably needed in
> different situations.

The comment about raising RTOmin was meant as an example.  I am not
advocating that approach.  I have not thought about what to do with
the RTO much at all and so that came off the top of my head as a
quick example.  The point was that you do *something* in an attempt
to prevent future spurious retransmits.

More generally, I have two lines of thinking on RTO estimation...

- Yes, different RTO algorithms may help things a bit in some
  cases, or even in general.  That is, I do not necessarily believe
  the current algorithm is the end-all of RTO estimators.

- The RTO should be fairly conservative.  The normal case should be
  that we repair loss using some other mechanism and the RTO should
  truely be the method of last resort for repairing loss and we
  should use it only rarely.

(I mostly think in terms of the second point -- that RTO
optimizations are less useful than scheme to try to prevent RTOs).

> Do you have evidence that the problem related to fast retransmit
> is significant?

There are studies in the literature that show reordering is a fairly
significant problem on some links (and reordering can cause
duplicate ACKs that trigger fast retransmit unnecessarily).

> > The current situation is that if you send a needless retransmit
> > (and hence add a packet to the network that accomplishes no new
> > work) you pay a price -- i.e., you slow down.  Now if we make
> > TCP realize this and undo things the price has been
> > substantially reduced.  So, there is really no incentive to try
> > to be conservative and easy on the network.  Therefore, it seems
> > to me (and others I have chatted with)
> 
> I don't understand the comments above.  If TCP can know for sure
> that it just has done a spurious retransmission, it should undo
> it.  I agree totally.  But it does not mean that there is no
> incentive to try to be conservative.  At least to some people, the
> issue with it is that what if TCP makes mistakes too often and
> does not react to real network problems as it does today.  

Sure, you should be able to recover from mistakes that you know you
made.  I agree completely.  Think about the extreme for a minute...
An example might be a connection in which every third packet sent is
a bogus retransmit.  So, a third of the packets we are spewing into
the network are worthless and are chewing up resources along the
path (bandwidth, processing, queue space, etc.).  Is this
acceptable?  I don't think so.  Currently, the TCP flow has to slow
down when it finds a loss.  So, there is an incentive to get the
decision correct -- i.e., if it decides something is a loss and it
really isn't then it needlessly slows down.  If we come up with a
good scheme that lets the sender continue to send at the same rate
even when the retransmit is determined to be bogus the incentive to
make good decisions is gone -- i.e., if we retransmit prematurely it
doesn't matter, we'll figure it out and recover without much of a
performance hit.

> If all routers now "punish" overly aggressive TCP traffic, the
> incentive to be nice is ample (-:

**if**

And, even if they did, I cannot imagine a scheme that a router could
reasonably implement that would be able to figure out how many
useless segments a particular flow is sending.  That would be a very
complicated and resource intensive algorithm to run on the fly in
the middle of the network, I think.

So, I think it is unreasonable to hope that the network is going to
protect itself from this particular phenomenon.

> I agree with the above.  But do they really need to go together?
> That's my question.  They can be deployed separately.

Yes, in my opinion.  They *do* need to go together.  If you are
making TCP more aggressive on one hand (reverting CC state) you also
need to make an effort to try to avoid such situations in the
future.

> My issue with new algorithms is that we need time to really see
> what they are doing.  And if there are multiple things going on,
> some effects may be masked and we cannot correctly determine some
> underlying problems.

Things do get tricky when they are mingled together.  And, to be
clear, I think there could be some good and useful research that
considers these things seperately.  But, from the point of view of
writing an RFC (we are using an IETF list here) I think we need to
couple these two things pretty closely -- at least until someone
shows that one without the other is OK.

allman


---
Mark Allman -- NASA GRC/BBN -- http://roland.grc.nasa.gov/~mallman/

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Sat Jan 12 03:36:03 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id DAA18579
	for <tsvwg-archive@odin.ietf.org>; Sat, 12 Jan 2002 03:35:58 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id DAA12085
	for tsvwg-archive@odin.ietf.org; Sat, 12 Jan 2002 03:36:01 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id DAA11552;
	Sat, 12 Jan 2002 03:11:46 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id DAA11473
	for <tsvwg@optimus.ietf.org>; Sat, 12 Jan 2002 03:11:42 -0500 (EST)
Received: from sj-msg-core-4.cisco.com (sj-msg-core-4.cisco.com [171.71.163.10])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id DAA18380
	for <tsvwg@ietf.org>; Sat, 12 Jan 2002 03:11:39 -0500 (EST)
Received: from mira-sjc5-2.cisco.com (mira-sjc5-2.cisco.com [171.71.163.16])
	by sj-msg-core-4.cisco.com (8.11.3/8.9.1) with ESMTP id g0C8BBJ06136;
	Sat, 12 Jan 2002 00:11:11 -0800 (PST)
Received: from stewart.chicago.il.us (ssh-sj1.cisco.com [171.68.225.134])
	by mira-sjc5-2.cisco.com (Mirapoint)
	with ESMTP id AAT03643;
	Sat, 12 Jan 2002 00:11:09 -0800 (PST)
Message-ID: <3C3FEF9D.1E1277AF@stewart.chicago.il.us>
Date: Sat, 12 Jan 2002 02:11:09 -0600
From: Randall Stewart <randall@stewart.chicago.il.us>
X-Mailer: Mozilla 4.76 [en] (X11; U; Linux 2.2.12 i386)
X-Accept-Language: en
MIME-Version: 1.0
To: Kacheong Poon <kcpoon@eng.sun.com>
CC: dajiang.zhang@nokia.com, tsvwg@ietf.org,
        Erik Nordmark <Erik.Nordmark@eng.sun.com>
Subject: Re: [Tsvwg] question about Ipv4 mapped addresses
References: <Roam.SIMC.2.0.6.1010789690.5503.kcpoon@eng.sun.com>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Transfer-Encoding: 7bit
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit

Kacheong:

Somehow I knew we would dis-agree on this :)

From reading the advanced socket BIS Peter and I get
the impression the IPV6_ONLY flag applies to the 
interface... which sort of makes sense since it
says that a mapped address is allowed on a IPv6 socket
that is v6 only....

And that leaves us with the thought that if you don't
have the V6 only flag on then you can pass IPv4 or IPv6
addresses up the socket.

Otherwise you end up doing something really silly in 
both the kernel and app.. i.e:

---on send--
App: Mapps IPv4 address to IPv6
Kernel: Gets address/ checks to see if mapped
Kernel: Unmapps address and calls routine to handle send

OR

---on recv--
Kernel: Mapps IPv4 address to IPv6 and sends up stack
App: gets/address checks to see if its mapped
App: unmapps address and then processes..


This is very very awkward and wastes a lot of cycles
for NO GOOD point.

Now I can see doing the mapping if the IPV6ONLY flag is
applied.. since the app is saying I only deal with V6 addresses.

We have asked Erik what he thinks on this (offline).. since I think
either the advanced socket draft needs more clearification 
and we are really doing something silly besides... or we
have it interpreted right...

I can see if you had a bunch of code that only handled
AF_INET and you needed to preserve as much of it as
possible this mapping thing may be worthwhile.. but in
our case with a new stack it would be awful wasteful and
silly IMHO...


R




Kacheong Poon wrote:
> 
> > but on page 6, bind()
> >     If sd is an IPv4 socket, the address passed must be an IPv4 address.
> >     If the sd is an IPv6 socket, the address passed can either be an
> >     IPv4 or an IPv6 address.
> >     ^^^^^
> >
> > Why is there such a difference? Why use IPv4 mapped addresses instead of
> > Ipv4 address in socket()?
> 
> I guess we should correct the above to use mapped addresses also.
> The goal is to make things clean so that for v6 socket, one should
> not pass in a v4 address structure.
> 
> Thanks for pointing this out.
> 
>                                                         K. Poon.
>                                                         kcpoon@eng.sun.com
> 
> _______________________________________________
> tsvwg mailing list
> tsvwg@ietf.org
> http://www1.ietf.org/mailman/listinfo/tsvwg

-- 
Randall R. Stewart
randall@stewart.chicago.il.us 815-342-5222 (cell phone)

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Sat Jan 12 17:03:18 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id RAA24348
	for <tsvwg-archive@odin.ietf.org>; Sat, 12 Jan 2002 17:03:18 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id RAA28582
	for tsvwg-archive@odin.ietf.org; Sat, 12 Jan 2002 17:03:20 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id QAA27832;
	Sat, 12 Jan 2002 16:42:52 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id QAA27804
	for <tsvwg@optimus.ietf.org>; Sat, 12 Jan 2002 16:42:43 -0500 (EST)
Received: from mercury.Sun.COM (mercury.Sun.COM [192.9.25.1])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id QAA24239
	for <tsvwg@ietf.org>; Sat, 12 Jan 2002 16:42:40 -0500 (EST)
Received: from engmail1.Eng.Sun.COM ([129.146.1.13])
	by mercury.Sun.COM (8.9.3+Sun/8.9.3) with ESMTP id NAA17291;
	Sat, 12 Jan 2002 13:42:43 -0800 (PST)
Received: from shield (shield.EBay.Sun.COM [129.150.113.84])
	by engmail1.Eng.Sun.COM (8.9.3+Sun/8.9.3/ENSMAIL,v2.1p1) with SMTP id NAA07831;
	Sat, 12 Jan 2002 13:42:41 -0800 (PST)
Date: Sat, 12 Jan 2002 13:42:00 -0800 (PST)
From: Kacheong Poon <kcpoon@eng.sun.com>
Reply-To: Kacheong Poon <kcpoon@eng.sun.com>
Subject: Re: [Tsvwg] Eifel draft and usage of time-stamps. 
To: mallman@grc.nasa.gov
Cc: Reiner Ludwig <Reiner.Ludwig@ericsson.com>,
        sanjay kaniyar <skaniyar@hotmail.com>, gurtov@cs.helsinki.fi,
        tsvwg@ietf.org
In-Reply-To: "Your message with ID" <200201120446.XAA03634@lawyers.grc.nasa.gov>
Message-ID: <Roam.SIMC.2.0.6.1010871720.22510.kcpoon@eng.sun.com>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; CHARSET=US-ASCII
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

> Sure, you should be able to recover from mistakes that you know you
> made.  I agree completely.  Think about the extreme for a minute...
> An example might be a connection in which every third packet sent is
> a bogus retransmit.  So, a third of the packets we are spewing into
> the network are worthless and are chewing up resources along the
> path (bandwidth, processing, queue space, etc.).  Is this
> acceptable?  I don't think so.  Currently, the TCP flow has to slow
> down when it finds a loss.  So, there is an incentive to get the
> decision correct -- i.e., if it decides something is a loss and it

I guess now I see why I didn't understand what you said...  It is not
an "incentive" to slow down after a loss, as far as I can tell.  There
is "incentive" for TCP stacks not to slow down.  TCP slows down now
because if there is a stack not doing it, you know what people will
say about it.  If a stack really wants to get the most from the network,
slowing down is properly the last thing to do.  

> really isn't then it needlessly slows down.  If we come up with a
> good scheme that lets the sender continue to send at the same rate
> even when the retransmit is determined to be bogus the incentive to
> make good decisions is gone -- i.e., if we retransmit prematurely it
> doesn't matter, we'll figure it out and recover without much of a
> performance hit.

If you look at it another way, there is no decision to be made now,
it is the "standard" practise.  If TCP has a timeout, it does slow
start and changes its rate of sending (ss threshold changes...).  This
is supposed to be good to the network, but not an incentive for a stack
to do because in some situations, this does not help anything at all 
(simple example is loss due to errors).  IMHO, I don't think you have to
worry about that TCP will never be slowed down.

Look at the example we are discussing now, spurious timeout.  If a
stack detects a spurious timeout for sure, it keeps the states prior
to the timeout and stops immediately sending duplicate packets out.
If it is not sure, the standard practise of slow start will take place.
So if the detection is very good, you don't need to worry about that.

Looking at it from your point, I believe you are more worried about 
things like making a correct decision to trigger a fast retransmit, as 
what you are working on.  But IMHO, regardless of whether there is 
a machanism to revert back to the previous state, there is an incentive
(a real incentive this time (-:) not to unnecessarily fast retransmit.  
In the extreme case you mentioned, it means that the real throughput, 
even if TCP never slows down, can be at best 2/3 of the max.  This is 
probably not acceptable to most people...

> > If all routers now "punish" overly aggressive TCP traffic, the
> > incentive to be nice is ample (-:
> 
> **if**
> 
> And, even if they did, I cannot imagine a scheme that a router could
> reasonably implement that would be able to figure out how many
> useless segments a particular flow is sending.  That would be a very
> complicated and resource intensive algorithm to run on the fly in
> the middle of the network, I think.
> 
> So, I think it is unreasonable to hope that the network is going to
> protect itself from this particular phenomenon.

First, if routers can "punish" overly aggressive traffic, there is
an incentive for stacks not to do that.  Second, given that, the
worst thing a stack can do is to unnecessarily send useless (say
duplicate) packets, but without gaining unfair (the fairness 
determined by the routers) advantage.  Then what it hurts is itself,
as I mentioned above.  I don't believe routers need to spend time 
determining whether a packet is sent unnecessarily at all...

> Yes, in my opinion.  They *do* need to go together.  If you are
> making TCP more aggressive on one hand (reverting CC state) you also
> need to make an effort to try to avoid such situations in the
> future.

To me, the key is if we are really making TCP more aggressive if,
for example, the spurious timeout detection mechanism is "extremely
(need to quantify it) good."  I think it is not.

> Things do get tricky when they are mingled together.  And, to be
> clear, I think there could be some good and useful research that
> considers these things seperately.  But, from the point of view of
> writing an RFC (we are using an IETF list here) I think we need to
> couple these two things pretty closely -- at least until someone
> shows that one without the other is OK.

I guess I cannot convince you )-:  But please don't get me wrong,
I am not saying that we don't need to find a better, say, fast
retransmit triggering mechanism.  For the topic we are discussing,
I think it is safe to revert things back after a timeout if TCP
can know for sure the timeout is spurious.  And as I asked previously,
I think we need another RTO algorithm.  I hope other researchers
can jump in and give their opinions.

							K. Poon.
							kcpoon@eng.sun.com



_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Sat Jan 12 17:08:08 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id RAA24370
	for <tsvwg-archive@odin.ietf.org>; Sat, 12 Jan 2002 17:08:08 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id RAA28614
	for tsvwg-archive@odin.ietf.org; Sat, 12 Jan 2002 17:08:11 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id QAA27928;
	Sat, 12 Jan 2002 16:54:57 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id QAA27900
	for <tsvwg@optimus.ietf.org>; Sat, 12 Jan 2002 16:54:55 -0500 (EST)
Received: from patan.sun.com (patan.Sun.COM [192.18.98.43])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id QAA24291
	for <tsvwg@ietf.org>; Sat, 12 Jan 2002 16:54:52 -0500 (EST)
Received: from engmail1.Eng.Sun.COM ([129.146.1.13])
	by patan.sun.com (8.9.3+Sun/8.9.3) with ESMTP id OAA01031;
	Sat, 12 Jan 2002 14:54:55 -0700 (MST)
Received: from shield (shield.EBay.Sun.COM [129.150.113.84])
	by engmail1.Eng.Sun.COM (8.9.3+Sun/8.9.3/ENSMAIL,v2.1p1) with SMTP id NAA09220;
	Sat, 12 Jan 2002 13:54:54 -0800 (PST)
Date: Sat, 12 Jan 2002 13:54:13 -0800 (PST)
From: Kacheong Poon <kcpoon@eng.sun.com>
Reply-To: Kacheong Poon <kcpoon@eng.sun.com>
Subject: Re: [Tsvwg] question about Ipv4 mapped addresses
To: Randall Stewart <randall@stewart.chicago.il.us>
Cc: dajiang.zhang@nokia.com, tsvwg@ietf.org,
        Erik Nordmark <Erik.Nordmark@eng.sun.com>
In-Reply-To: "Your message with ID" <3C3FEF9D.1E1277AF@stewart.chicago.il.us>
Message-ID: <Roam.SIMC.2.0.6.1010872453.20112.kcpoon@eng.sun.com>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; CHARSET=US-ASCII
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

> From reading the advanced socket BIS Peter and I get
> the impression the IPV6_ONLY flag applies to the 
> interface... which sort of makes sense since it
> says that a mapped address is allowed on a IPv6 socket
> that is v6 only....

Do you mean the IPV6_V6ONLY option as described in the
basic socket draft (draft-ietf-ipngwg-rfc2553bis-04.txt)?
If it is, then I believe it is actually the other way
round.  For example, a v6 TCP socket, for portability reason, 
always works with TCP/IPv4 connections.  That option is
to tell the stack that the socket does not want anything
to do with TCP/IPv4 connections.

> And that leaves us with the thought that if you don't
> have the V6 only flag on then you can pass IPv4 or IPv6
> addresses up the socket.

If I am not mistaken, the IPv6 socket draft says that the
address structure passed in to a v6 socket is the v6
address structure.

> Otherwise you end up doing something really silly in 
> both the kernel and app.. i.e:
> 
> ---on send--
> App: Mapps IPv4 address to IPv6
> Kernel: Gets address/ checks to see if mapped
> Kernel: Unmapps address and calls routine to handle send
> 
> OR
> 
> ---on recv--
> Kernel: Mapps IPv4 address to IPv6 and sends up stack
> App: gets/address checks to see if its mapped
> App: unmapps address and then processes..

Hmmm, I believe this is just one way of doing things....

							K. Poon.
							kcpoon@eng.sun.com



_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Sat Jan 12 23:57:26 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id XAA29276
	for <tsvwg-archive@odin.ietf.org>; Sat, 12 Jan 2002 23:57:26 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id XAA06272
	for tsvwg-archive@odin.ietf.org; Sat, 12 Jan 2002 23:57:29 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id XAA05523;
	Sat, 12 Jan 2002 23:30:17 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id XAA05468
	for <tsvwg@optimus.ietf.org>; Sat, 12 Jan 2002 23:30:13 -0500 (EST)
Received: from seraph3.grc.nasa.gov (firewall-user@seraph3.lerc.nasa.gov [128.156.10.12])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id XAA28743
	for <tsvwg@ietf.org>; Sat, 12 Jan 2002 23:30:09 -0500 (EST)
Received: by seraph3.grc.nasa.gov (Postfix, from userid 5)
	id C7AB9640FB; Sat, 12 Jan 2002 23:28:53 -0500 (EST)
Received: from lombok-fi.lerc.nasa.gov(139.88.112.33) by seraph3.grc.nasa.gov via csmap (V6.0)
	id srcAAAY3aiTE; Sat, 12 Jan 02 23:28:53 -0500
Received: from lawyers.grc.nasa.gov (ras110.lerc.nasa.gov [139.88.123.110])
	by lombok-fi.lerc.nasa.gov (NASA LeRC 8.9.1.1/8.9.1) with ESMTP id XAA01459
	for <tsvwg@ietf.org>; Sat, 12 Jan 2002 23:28:51 -0500 (EST)
Received: from lawyers (mallman@localhost)
	by lawyers.grc.nasa.gov (8.9.3/8.9.3) with ESMTP id RAA04693;
	Sat, 12 Jan 2002 17:37:59 -0500
Message-Id: <200201122237.RAA04693@lawyers.grc.nasa.gov>
To: Kacheong Poon <kcpoon@eng.sun.com>
From: Mark Allman <mallman@grc.nasa.gov>
Reply-To: mallman@grc.nasa.gov
Cc: Reiner Ludwig <Reiner.Ludwig@ericsson.com>,
        sanjay kaniyar <skaniyar@HOTMAIL.COM>, gurtov@cs.helsinki.fi,
        tsvwg@ietf.org
Subject: Re: [Tsvwg] Eifel draft and usage of time-stamps. 
Organization: BBN Technologies/NASA GRC
Song-of-the-Day: Hammer to Fall
Date: Sat, 12 Jan 2002 17:37:59 -0500
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org


> If you look at it another way, there is no decision to be made
> now, it is the "standard" practise.  

Right.

> Look at the example we are discussing now, spurious timeout.  If a
> stack detects a spurious timeout for sure, it keeps the states
> prior to the timeout and stops immediately sending duplicate
> packets out.  If it is not sure, the standard practise of slow
> start will take place.  So if the detection is very good, you
> don't need to worry about that.

I don't need to worry about what?  If we know we spuriously
retransmitted something we know we added a packet to the network
that accomplished no work (yet used resources).

> Looking at it from your point, I believe you are more worried
> about things like making a correct decision to trigger a fast
> retransmit, as what you are working on.

No, I am worried about both RTOs and fast retransmits, actually.  I
have *studied* only the fast retransmit case.  But, both cases need
solved. 

> But IMHO, regardless of whether there is a machanism to revert
> back to the previous state, there is an incentive (a real
> incentive this time (-:) not to unnecessarily fast retransmit.  In
> the extreme case you mentioned, it means that the real throughput,
> even if TCP never slows down, can be at best 2/3 of the max.  This
> is probably not acceptable to most people...

I'll get a copy of our paper posted early next week.  But, we have
an example in the paper that compares two TCPs:

1.  a standard SACK-based version of TCP
2.  a version of TCP that can detect spurios fast retransmits and
    undo the bogus CC state changes (but, takes no action to prevent
    further needless retransmits)

This is done in a network with varying amounts of reordering -- so
needless retransmits do happen.

What we see is that (2) gets throughput that is almost as good as
(1) does when there is no reordering (i.e., no needless
retransmits).  There is about a 1% performance hit if we can detect
spurious retransmits and undo the CC changes.  On the other hand, if
you look at the number of needless packets sent variant (2) sends
about 6 times as many as standard SACK (variant (1)).

So, while there are cases when I think you are right that there is
an incentive because needless retransmits throw away *your*
bandwidth and that can have a big impact, I also think there are
cases when this incentive is negligible.

And, I think this is a problem.  I like the performance gain from
detecting spurious retransmits.  But, I wouldn't like to see people
pile in a lot of unnecessary packets into the network.

> To me, the key is if we are really making TCP more aggressive if,
> for example, the spurious timeout detection mechanism is
> "extremely (need to quantify it) good."  I think it is not.

I do not understand this comment.  Eifel is pretty robust, it seems
to me.  But, regardless I haven't heard of any failure cases --
i.e., when it says there was a spurous timeout and there was not.
Are there some?

> For the topic we are discussing, I think it is safe to revert
> things back after a timeout if TCP can know for sure the timeout
> is spurious.

And, just so we're clear my own opinion is that reverting CC state
is not safe unless some action is taken such that the chance for
spurious retransmit (RTO or FR) in the future is reduced.

allman


---
Mark Allman -- NASA GRC/BBN -- http://roland.grc.nasa.gov/~mallman/

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Sun Jan 13 08:05:59 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id IAA10272
	for <tsvwg-archive@odin.ietf.org>; Sun, 13 Jan 2002 08:05:59 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id IAA25700
	for tsvwg-archive@odin.ietf.org; Sun, 13 Jan 2002 08:06:03 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id HAA25120;
	Sun, 13 Jan 2002 07:30:58 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id HAA25091
	for <tsvwg@optimus.ietf.org>; Sun, 13 Jan 2002 07:30:56 -0500 (EST)
Received: from fep02-app.kolumbus.fi (fep02-0.kolumbus.fi [193.229.0.44])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id HAA10096
	for <tsvwg@ietf.org>; Sun, 13 Jan 2002 07:30:52 -0500 (EST)
Received: from there ([212.54.4.36]) by fep02-app.kolumbus.fi
          (InterMail vM.5.01.03.08 201-253-122-118-108-20010628) with SMTP
          id <20020113123053.FUIE1908.fep02-app.kolumbus.fi@there>;
          Sun, 13 Jan 2002 14:30:53 +0200
Content-Type: text/plain;
  charset="iso-8859-1"
From: Pasi Sarolahti <sarolaht@cs.helsinki.fi>
To: mallman@grc.nasa.gov, Kacheong Poon <kcpoon@eng.sun.com>
Subject: Re: [Tsvwg] Eifel draft and usage of time-stamps.
Date: Sun, 13 Jan 2002 14:30:45 +0200
X-Mailer: KMail [version 1.3.1]
Cc: Reiner Ludwig <Reiner.Ludwig@ericsson.com>,
        sanjay kaniyar <skaniyar@hotmail.com>, gurtov@cs.helsinki.fi,
        tsvwg@ietf.org
References: <200201122237.RAA04693@lawyers.grc.nasa.gov>
In-Reply-To: <200201122237.RAA04693@lawyers.grc.nasa.gov>
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
Message-Id: <20020113123053.FUIE1908.fep02-app.kolumbus.fi@there>
Content-Transfer-Encoding: 8bit
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 8bit

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hello,

May I comment one point of your discussion?

On Sunday 13 January 2002 00:37, Mark Allman wrote:

> I do not understand this comment.  Eifel is pretty robust, it seems
> to me.  But, regardless I haven't heard of any failure cases --
> i.e., when it says there was a spurous timeout and there was not.
> Are there some?

I have been running into one suspicious case. If there is a "link outage" 
when both data and ack packets get dropped in the same window, and timestamps 
as specified in RFC 1323 are used as the retransmission indicator, Eifel can 
make a wrong conclusion. This is a simplified example of my problem I sent to 
Reiner Ludwig and Andrei Gurtov a week ago:

1. Sender sends a window of data segments 11-20 <timestamp 101-110> as
clocked by incoming acks
2. Data segment 15 is dropped
3. Receiver acknowledges 11-14 (tsecr 101-104) plus sends 5 dupacks (tsecr
104 in all of those), but ACK segment(s) for 13-14 and the dupacks are
dropped
4. RTO expires at the sender, sender retransmits data segment 13 <tstamp
120>
5. Receiver acknowledges segment 14. Since this is dupack, tsecr is still
104. However, at the sender this appears as a new ack. Because 104 < 120,
the sender declares the retransmission spurious.

The basic Eifel would undo the congestion control parameters and continue 
transmitting new data, right? Eifel detection is basically correct, since the 
retransmission triggered by RTO is unnecessarily made, but undoing the window 
and continuing with new, unsent data is a wrong action to take, since 
data segment was lost. In my tests the RTO that occurred during link outage 
was followed by a backed off RTO. First the sender did fast retransmit due to 
duplicate ACKs triggered by the new segments sent after Eifel detection. cwnd 
was reduced, and there were not enough packets in the pipe to keep the fast 
recovery running (due to the recent link outage, which actually dropped more 
than one data segments). SACK and limited transmit were in use.

As Andrei pointed out to me, many implementations use a different (fixed?) 
version of timestamps which avoid this problem. The issue about timestamp 
specification has been discussed under the pilc list lately, although in a 
slightly different context.

- - Pasi

- -- 
http://www.cs.helsinki.fi/u/sarolaht
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.6 (GNU/Linux)
Comment: For info see http://www.gnupg.org

iD8DBQE8QX37oNa7NH1G2csRAhaJAJ9mamg9jGmVMQdcNSy3/aesPy5gaQCeKmTR
iFNEV29YyWRAg7VOHnCVRJY=
=hK4Q
-----END PGP SIGNATURE-----


_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Sun Jan 13 11:26:50 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id LAA12073
	for <tsvwg-archive@odin.ietf.org>; Sun, 13 Jan 2002 11:26:50 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id LAA29127
	for tsvwg-archive@odin.ietf.org; Sun, 13 Jan 2002 11:26:53 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id KAA28595;
	Sun, 13 Jan 2002 10:55:03 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id KAA28566
	for <tsvwg@optimus.ietf.org>; Sun, 13 Jan 2002 10:54:57 -0500 (EST)
Received: from smtp.dave.sonera.fi (smtp.dave.sonera.fi [131.177.130.21])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id KAA11650
	for <tsvwg@ietf.org>; Sun, 13 Jan 2002 10:54:52 -0500 (EST)
Received: from gurtoan1-nb.etela.sonera.fi ([131.177.36.107]:3841 "HELO
        gurtoan1nb") by inside.dave.sonera.fi with SMTP id <S59027AbSAMPyd>;
        Sun, 13 Jan 2002 17:54:33 +0200
Message-ID: <01e901c19c4a$2c834d40$6b24b183@etela.sonera.fi>
From: "Andrei Gurtov" <gurtov@cs.helsinki.fi>
To: <mallman@grc.nasa.gov>, "Kacheong Poon" <kcpoon@eng.sun.com>
Cc: "Reiner Ludwig" <Reiner.Ludwig@ericsson.com>,
        "sanjay kaniyar" <skaniyar@hotmail.com>, <tsvwg@ietf.org>
References: <200201111220.HAA01958@lawyers.grc.nasa.gov>
Subject: Re: [Tsvwg] Eifel draft and usage of time-stamps. 
Date:   Sun, 13 Jan 2002 17:51:32 +0200
MIME-Version: 1.0
Content-Type: text/plain;
        charset="iso-8859-1"
Content-Transfer-Encoding: 7bit
X-Priority: 3
X-MSMail-Priority: Normal
X-Mailer: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Content-Transfer-Encoding: 7bit
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit

Mark Allman wrote:

> You wouldn't necessarily need to choose a different algorithm on the
> fly.  You could do any number of "tweaks" to the current algorithm
> to try to avoid future spurious timeouts.  One simple thing could be
> to raise the minimum RTO a little each time you needlessly
> retransmitted.

In case when delay spikes exceed the normal RTT by several times and occur
at random intervals I don't see how VJ timer could be adapted except making
it non-decreasing i.e. estimating the MAX possible RTT on the path.

The RTO value with VJ timer peaks shortly after a spurios timeout and then
it gradually decreases back to normal RTT. You may need to do something
competely opposite:  keep the RTO low immediately after the spurious timeout
to repaire lost segments if fast retransmit, etc. fail. Then, gradually
increase the timer as time goes and probability of a delay occurance rises.

>And, just so we're clear my own opinion is that reverting CC state
>is not safe unless some action is taken such that the chance for
>spurious retransmit (RTO or FR) in the future is reduced.

Reverting CC state does not necessary makes you more aggressive. Imagine a
connection in congestion avoidance that experinces a spurious RTO and goes
into go-back-N using slow start. This can easily lead to congestion losses
since the bottleneck is filled with old packets and retransmissions are send
at twice higher rate. When CC state is reverted, the connection continues in
CA that preserves packet conservation.

Because of that I think Eifel is *more conservative* than vanila TCP in
loading the network. If connection has say a 10-segment window, then Eifel
could detect 10 spurious timeouts paying the same price in terms of
unnecessary retransmissions as vanila TCP does for a single timeout.
Consequently, deploying Eifel with the rfc2988 timer should be safe (and
even good) for network. Of course, I'm not against searching for a better
algorithm for RTO estimation.

Andrei


_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Sun Jan 13 13:46:33 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id NAA13655
	for <tsvwg-archive@odin.ietf.org>; Sun, 13 Jan 2002 13:46:33 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id NAA02004
	for tsvwg-archive@odin.ietf.org; Sun, 13 Jan 2002 13:46:36 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id NAA01484;
	Sun, 13 Jan 2002 13:17:58 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id NAA01455
	for <tsvwg@optimus.ietf.org>; Sun, 13 Jan 2002 13:17:56 -0500 (EST)
Received: from lawyers.grc.nasa.gov (d193.as4.clev.oh.voyager.net [209.81.205.194])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id NAA13378
	for <tsvwg@ietf.org>; Sun, 13 Jan 2002 13:17:51 -0500 (EST)
Received: from lawyers (mallman@localhost)
	by lawyers.grc.nasa.gov (8.9.3/8.9.3) with ESMTP id NAA09683;
	Sun, 13 Jan 2002 13:15:32 -0500
Message-Id: <200201131815.NAA09683@lawyers.grc.nasa.gov>
To: "Andrei Gurtov" <gurtov@cs.helsinki.fi>
From: Mark Allman <mallman@grc.nasa.gov>
Reply-To: mallman@grc.nasa.gov
cc: "Kacheong Poon" <kcpoon@eng.sun.com>,
        "Reiner Ludwig" <Reiner.Ludwig@ericsson.com>,
        "sanjay kaniyar" <skaniyar@HOTMAIL.COM>, tsvwg@ietf.org
Subject: Re: [Tsvwg] Eifel draft and usage of time-stamps. 
Organization: BBN Technologies/NASA GRC
Song-of-the-Day: Hammer to Fall
Date: Sun, 13 Jan 2002 13:15:32 -0500
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org


> In case when delay spikes exceed the normal RTT by several times
> and occur at random intervals I don't see how VJ timer could be
> adapted except making it non-decreasing i.e. estimating the MAX
> possible RTT on the path.

Maybe you cannot.  I see a ton of work here.  Reiner's idea of
re-seeding the RTO algorithm with the RTT of the spike is
interesting.  As far as I can tell nobody has tried to characterize
these delay spikes all that well.  Are they a little bigger than the
RTO?  A lot bigger than the RTO?  I am sure someone could come up
with examples of both.

There could be cases when a TCP would determine that no matter what
is done to the RTO it will never be able to stop spurious
retransmits *and* provide anything resembling reasonable performance
(e.g., if the spike is 10 times the RTO).  On the other hand, I can
envision cases where the RTO was off by "a little" and we could
tweak the algorithm or re-seed it of something so that we prevent
future spurious retransmits.  I can see a decision on what to do
being based on the size of the RTT spike (and, maybe the past
history of these things).  Point is that the response does not have
to be a simple tweak -- it could be a more complicated decsision.

> Reverting CC state does not necessary makes you more
> aggressive. Imagine a connection in congestion avoidance that
> experinces a spurious RTO and goes into go-back-N using slow
> start. This can easily lead to congestion losses since the
> bottleneck is filled with old packets and retransmissions are send
> at twice higher rate. When CC state is reverted, the connection
> continues in CA that preserves packet conservation.

By "reverting the CC state" I assume we are talking setting ssthresh
to the value of cwnd when loss was detected (as that seems to be
what most folks are talking about).  So, we are always going to be
in slow start after a spurious retransmit (not CA).  The question is
whether we slow start to the previous value of cwnd or half that
value.

> Because of that I think Eifel is *more conservative* than vanila
> TCP in loading the network. If connection has say a 10-segment
> window, then Eifel could detect 10 spurious timeouts paying the
> same price in terms of unnecessary retransmissions as vanila TCP
> does for a single timeout.  Consequently, deploying Eifel with the
> rfc2988 timer should be safe (and even good) for network. Of
> course, I'm not against searching for a better algorithm for RTO
> estimation.

I think this confuses the two things we can do when we detect a
spurious retransmit.  We can use eifel to detect spurious retrasmits
and avoid the go-back-n behavior without reverting the CC state.
(Or, we could keep go-back-n and revert the CC state -- although
that would not seem to make sense.)  I would say that eifel for
avoiding go-back-n is reasonable and safe (as long as there are no
CC changes made).  (This is also modulo the other note on a failure
case for eifel that I have not yet had a chance to think about.)
And, I would agree with you that this is *better* for the network
because we possibly avoid a bunch of unnecessary retransmissions
that cost resources and do no useful work.  This scheme does not
increase TCP's agressiveness from a CC perspective.  This would
simply make TCP smarter about what it sends.

On the other hand, if you want to immediatly revert the CC state you
are being more aggressive than current TCP because you are letting
cwnd grow exponentially for a longer period of time (in the RTO
case).  And, you are reducing the incentive for TCP to make a good
decision about when to retransmit.  In that case I still think we
need some balancing logic that tries to prevent future occurances of
needless retransmits.

allman


---
Mark Allman -- NASA GRC/BBN -- http://roland.grc.nasa.gov/~mallman/


_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Sun Jan 13 19:58:25 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id TAA16397
	for <tsvwg-archive@odin.ietf.org>; Sun, 13 Jan 2002 19:58:25 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id TAA09547
	for tsvwg-archive@odin.ietf.org; Sun, 13 Jan 2002 19:58:28 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id TAA09425;
	Sun, 13 Jan 2002 19:40:46 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id TAA09393
	for <tsvwg@optimus.ietf.org>; Sun, 13 Jan 2002 19:40:43 -0500 (EST)
Received: from hotmail.com (f100.law3.hotmail.com [209.185.241.100])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id TAA16259
	for <tsvwg@ietf.org>; Sun, 13 Jan 2002 19:40:40 -0500 (EST)
Received: from mail pickup service by hotmail.com with Microsoft SMTPSVC;
	 Sun, 13 Jan 2002 16:40:12 -0800
Received: from 131.107.3.84 by lw3fd.law3.hotmail.msn.com with HTTP;
	Mon, 14 Jan 2002 00:40:12 GMT
X-Originating-IP: [131.107.3.84]
From: "sanjay kaniyar" <skaniyar@hotmail.com>
To: sarolaht@cs.helsinki.fi, mallman@grc.nasa.gov, kcpoon@eng.sun.com
Cc: Reiner.Ludwig@ericsson.com, gurtov@cs.helsinki.fi, tsvwg@ietf.org
Subject: Re: [Tsvwg] Eifel draft and usage of time-stamps.
Date: Sun, 13 Jan 2002 16:40:12 -0800
Mime-Version: 1.0
Content-Type: text/plain; format=flowed
Message-ID: <F100sSSINegNYTvl5Pn00021169@hotmail.com>
X-OriginalArrivalTime: 14 Jan 2002 00:40:12.0588 (UTC) FILETIME=[06C6F6C0:01C19C94]
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

Interesting example, Pasi.

>As Andrei pointed out to me, many implementations use a different (fixed?)
>version of timestamps which avoid this problem.

Can you please describe the rules for the fixed version?

Sanjay.

_________________________________________________________________
MSN Photos is the easiest way to share and print your photos: 
http://photos.msn.com/support/worldwide.aspx


_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Sun Jan 13 23:36:27 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id XAA20169
	for <tsvwg-archive@odin.ietf.org>; Sun, 13 Jan 2002 23:36:26 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id XAA14340
	for tsvwg-archive@odin.ietf.org; Sun, 13 Jan 2002 23:36:30 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id XAA13809;
	Sun, 13 Jan 2002 23:15:11 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id XAA13780
	for <tsvwg@optimus.ietf.org>; Sun, 13 Jan 2002 23:15:09 -0500 (EST)
Received: from hindon.hss.co.in ([202.54.26.202])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id XAA19678
	for <tsvwg@ietf.org>; Sun, 13 Jan 2002 23:15:02 -0500 (EST)
From: ygahlaut@hss.hns.com
Received: from sandesh.hss.hns.com (sandesh [139.85.242.35])
	by hindon.hss.co.in (8.10.0/8.10.0) with SMTP id g0E4FbE25532;
	Mon, 14 Jan 2002 09:45:38 +0530 (IST)
Received: by sandesh.hss.hns.com(Lotus SMTP MTA v4.6.3  (733.2 10-16-1998))  id 65256B41.001742FF ; Mon, 14 Jan 2002 09:44:04 +0530
X-Lotus-FromDomain: HSS
To: Randall Stewart <randall@stewart.chicago.il.us>
cc: tsvwg@ietf.org
Message-ID: <65256B41.0016E2E4.00@sandesh.hss.hns.com>
Date: Mon, 14 Jan 2002 09:41:19 +0530
Subject: Re: [Tsvwg] Section 7.1.3 sctpsocket-02.txt
Mime-Version: 1.0
Content-type: text/plain; charset=us-ascii
Content-Disposition: inline
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org



Randall :
     thanks for for answer. i have one more doubt ...
  lets consider this sequence of events

   open a udp socket .
   make association with peer endpoint A with streams setting X  and got
assoc_id 1
   set some diifernt no of in and out streams using setsockopt
  make another association with other peer endpoint B with streams setting Y and
got assoc_id 2
  and now if i use getsockopt with socket option SCTP_INITMSG
  what should i get stream info from assoc_id 1 or 2....
the stream setting X and Y  may be different ( after stream negotaition )
is this sequence valid or i am missing something
i hope you understand my doubt .

thanks
yogesh




Randall Stewart <randall@stewart.chicago.il.us> on 01/11/2002 08:42:32 PM

To:   Yogesh Gahlaut/HSS@HSS
cc:   tsvwg@ietf.org

Subject:  Re: [Tsvwg] Section 7.1.3 sctpsocket-02.txt





Yogesh:

Two things:

First:

The sctp_initmsg is used in the UDP model to start
a new association with a peer (i.e. you are sending the
INIT message out) and over-ride the default INIT parameters
with the value found in the sctp_initmsg structure.

So we have an app sending a message with a
sendmsg() and putting the sctp_initmsg structure
in a CMSG header to go with the first message you
are sending. The sendmsg() call also includes the
destination address in the msg_name field ... i.e. the
sockaddr * of who you are sending (and INIT'ing) to.

Second:

If you are asking instead about notifications, i.e.
the association just came up and how do I tell what
parameters were negotiated in the streams counts
etc.. you get this in a MSG_NOTIFICATION in the
form of a

struct sctp_assoc_change x;

And it too is received via a recvmsg() call where
the msg_flags field gets set to MSG_NOTIFICATION and the
msg_name will have the peer that the association just came
up with .. the data read will be the  sctp_assoc_change..

Of course you must enable this event...

Hope that helps...

R

ygahlaut@hss.hns.com wrote:
>
> hi all,
>
>   in section 7.1.3   the following structure is being used for returning the
> SCTP_INITMSG option value
>
> struct sctp_initmsg {
>        uint16_t sinit_num_ostreams;
>        uint16_t sinit_max_instreams;
>        uint16_t sinit_max_attempts;
>        uint16_t sinit_max_init_timeo;
>     };
>
>  Since it does not have any input parameter to differentiate in case of more
> than
> one association in UDP style socket .
>
> can anybody explain how to handle this case ?
>
> thanks
> Yogesh
>
> _______________________________________________
> tsvwg mailing list
> tsvwg@ietf.org
> http://www1.ietf.org/mailman/listinfo/tsvwg

--
Randall R. Stewart
randall@stewart.chicago.il.us 815-342-5222 (cell phone)

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg





_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Mon Jan 14 02:37:28 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id CAA29429
	for <tsvwg-archive@odin.ietf.org>; Mon, 14 Jan 2002 02:37:28 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id CAA26880
	for tsvwg-archive@odin.ietf.org; Mon, 14 Jan 2002 02:37:30 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id CAA26271;
	Mon, 14 Jan 2002 02:20:15 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id CAA26241
	for <tsvwg@optimus.ietf.org>; Mon, 14 Jan 2002 02:20:13 -0500 (EST)
Received: from porsta.cs.Helsinki.FI (root@porsta.cs.Helsinki.FI [128.214.48.124])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id CAA29280
	for <tsvwg@ietf.org>; Mon, 14 Jan 2002 02:20:10 -0500 (EST)
Received: from there (IDENT:sarolaht@kiviletto.cs.Helsinki.FI [128.214.9.223])
	by porsta.cs.Helsinki.FI (8.11.6/8.11.6) with SMTP id g0E7JY703966;
	Mon, 14 Jan 2002 09:19:35 +0200
Message-Id: <200201140719.g0E7JY703966@porsta.cs.Helsinki.FI>
Content-Type: text/plain;
  charset="iso-8859-1"
From: Pasi Sarolahti <sarolaht@cs.helsinki.fi>
To: "sanjay kaniyar" <skaniyar@hotmail.com>, mallman@grc.nasa.gov,
        kcpoon@eng.sun.com
Subject: Re: [Tsvwg] Eifel draft and usage of time-stamps.
Date: Mon, 14 Jan 2002 09:19:29 +0200
X-Mailer: KMail [version 1.3.1]
Cc: Reiner.Ludwig@ericsson.com, gurtov@cs.helsinki.fi, tsvwg@ietf.org
References: <F100sSSINegNYTvl5Pn00021169@hotmail.com>
In-Reply-To: <F100sSSINegNYTvl5Pn00021169@hotmail.com>
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
Content-Transfer-Encoding: 8bit
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 8bit

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On Monday 14 January 2002 02:40, sanjay kaniyar wrote:

> >As Andrei pointed out to me, many implementations use a different
> > (fixed?) version of timestamps which avoid this problem.
>
> Can you please describe the rules for the fixed version?

The original RFC 1323 has the following rule:

(2)  If Last.ACK.sent falls within the range of sequence numbers
     of an incoming segment:

     SEG.SEQ <= Last.ACK.sent < SEG.SEQ + SEG.LEN

     then the TSval from the segment is copied to TS.Recent;
     otherwise, the TSval is ignored.

Whereas an expired draft to update RFC 1323 available at
http://www.kohala.com/start/tcplw-extensions.txt changes the 
abovementioned condition to:

     SEG.TSval >= TSrecent and SEG.SEQ <= Last.ACK.sent

The latter condition would update TSrecent when incoming segment has 
already been received, for example because the sender has retransmitted 
the segment due to lost ACK. In my example, using the latter alternative 
the timestamp of the retransmission would have been echoed back and Eifel 
sender would not have considered the retransmission spurious, thus 
retransmitting the unacknowledged segments.

- - Pasi

- -- 
http://www.cs.helsinki.fi/u/sarolaht/
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.6 (GNU/Linux)
Comment: For info see http://www.gnupg.org

iD8DBQE8QoaFoNa7NH1G2csRAk6MAKDapDMDmuWcx81tnpEbFDPN954ZJgCg5Ow/
6kVE71uipDVuPem4lHhrqxA=
=b8xU
-----END PGP SIGNATURE-----

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Mon Jan 14 03:57:12 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id DAA00093
	for <tsvwg-archive@odin.ietf.org>; Mon, 14 Jan 2002 03:57:12 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id DAA28719
	for tsvwg-archive@odin.ietf.org; Mon, 14 Jan 2002 03:57:14 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id DAA28517;
	Mon, 14 Jan 2002 03:37:19 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id DAA28489
	for <tsvwg@optimus.ietf.org>; Mon, 14 Jan 2002 03:37:17 -0500 (EST)
Received: from smtp.dave.sonera.fi (smtp.dave.sonera.fi [131.177.130.21])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id DAA29908
	for <tsvwg@ietf.org>; Mon, 14 Jan 2002 03:37:14 -0500 (EST)
Received: from gurtoan1-nb.etela.sonera.fi ([131.177.36.107]:4264 "HELO
        gurtoan1nb") by inside.dave.sonera.fi with SMTP id <S57933AbSANIhE>;
        Mon, 14 Jan 2002 10:37:04 +0200
Message-ID: <029001c19cd6$2cc30fc0$6b24b183@etela.sonera.fi>
From: "Andrei Gurtov" <gurtov@cs.helsinki.fi>
To: <mallman@grc.nasa.gov>
Cc: "Kacheong Poon" <kcpoon@eng.sun.com>,
        "Reiner Ludwig" <Reiner.Ludwig@ericsson.com>,
        "sanjay kaniyar" <skaniyar@HOTMAIL.COM>, <tsvwg@ietf.org>
References: <200201131815.NAA09683@lawyers.grc.nasa.gov>
Subject: Re: [Tsvwg] Eifel draft and usage of time-stamps. 
Date:   Mon, 14 Jan 2002 10:33:41 +0200
MIME-Version: 1.0
Content-Type: text/plain;
        charset="iso-8859-1"
Content-Transfer-Encoding: 7bit
X-Priority: 3
X-MSMail-Priority: Normal
X-Mailer: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Content-Transfer-Encoding: 7bit
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit

Mark Allman wrote:

> Maybe you cannot.  I see a ton of work here.  Reiner's idea of
> re-seeding the RTO algorithm with the RTT of the spike is
> interesting.  As far as I can tell nobody has tried to characterize
> these delay spikes all that well.  Are they a little bigger than the
> RTO?  A lot bigger than the RTO?  I am sure someone could come up
> with examples of both.

Re-seeding the timer with samples from the spike may not be the solution
(recall my comment in the previous mail). The assumption made by the
conventional timer is that current RTT has some correlation to a coming
delay spike. In many cases it does not and thus a different approach
(statistical model?) is needed to predict a next delay spike as you
suggested. I totally agree that it would be nice to know magnitude and
sources of delay spikes in the Internet and are they correlated to RTT.
However, as we discussed before it may a challenging task. At the moment
there seems to be at least two real, understood and quite different
instances to work with: 1) delay spikes that several times exceed RTT (GPRS)
and 2) moderate  delay jumps due to "bandwidth oscillation" (found by Farid
for cdma2000). In both cases it seems that delays are independent of current
RTT...

> By "reverting the CC state" I assume we are talking setting ssthresh
> to the value of cwnd when loss was detected (as that seems to be
> what most folks are talking about).  So, we are always going to be
> in slow start after a spurious retransmit (not CA).  The question is
> whether we slow start to the previous value of cwnd or half that
> value.

I don't think setting ssthresh to the old value of cwnd is a truly general
solution. Consider e.g. TCP connection starting up on a 100-KB
delay-bandwidth network and experiencing a spurious timeout. Would you set
ssthresh to 10 KB or so rendering the connection to dwell in congestion
avoidance for the rest of its life and severely underutilize the link? The
issue here is a spurious timeout by its own a sound indication of a
congestion? Should we instead wait for a packet loss and then slow down?
Based on that, I was speaking of reverting both ssthresh and cwnd.
Independently on whichever we do,  we do not want to send packets at the
twice rate immediately after a spurious timeout if connection was in CA
since the bottleneck queue is filled with old packets. Continuing in CA or
skipping some ACKs until flight<cwnd (not transmitting anything on them)
rather than slow-starting to some ssthresh (transmitting a bunch of packets
at twice rate and then switching to CA) preserves packet conservation.

> I think this confuses the two things we can do when we detect a
> spurious retransmit.  We can use eifel to detect spurious retrasmits
> and avoid the go-back-n behavior without reverting the CC state.
> (Or, we could keep go-back-n and revert the CC state -- although
> that would not seem to make sense.)  I would say that eifel for
> avoiding go-back-n is reasonable and safe (as long as there are no
> CC changes made).  (This is also modulo the other note on a failure
> case for eifel that I have not yet had a chance to think about.)
> And, I would agree with you that this is *better* for the network
> because we possibly avoid a bunch of unnecessary retransmissions
> that cost resources and do no useful work.  This scheme does not
> increase TCP's agressiveness from a CC perspective.  This would
> simply make TCP smarter about what it sends.

I referred to Eifel as resuming with snd_max and fully reverting CC. It may
be that reducing number of unnecessary retransmissions buys you enough to
also revert the CC state? Still, talking about being more aggressive may not
apply here if we think of a lost packet (or ECN) as the only real indication
of congestion.
It seems that rfc2988 with patches from the tech report is quite
conservative and the sender is unlikely to go into chain of spurious
timeouts. It may well be that letting Eifel to handle occasional delay
spikes is a better solution than to invent some overly complicated timer to
predict them.

Andrei


_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Mon Jan 14 04:36:41 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id EAA00364
	for <tsvwg-archive@odin.ietf.org>; Mon, 14 Jan 2002 04:36:40 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id EAA00170
	for tsvwg-archive@odin.ietf.org; Mon, 14 Jan 2002 04:36:43 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id EAA29360;
	Mon, 14 Jan 2002 04:14:42 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id EAA29329
	for <tsvwg@optimus.ietf.org>; Mon, 14 Jan 2002 04:14:40 -0500 (EST)
Received: from smtp.dave.sonera.fi (smtp.dave.sonera.fi [131.177.130.21])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id EAA00238
	for <tsvwg@ietf.org>; Mon, 14 Jan 2002 04:14:37 -0500 (EST)
Received: from gurtoan1-nb.etela.sonera.fi ([131.177.36.107]:4270 "HELO
        gurtoan1nb") by inside.dave.sonera.fi with SMTP id <S57872AbSANJOQ>;
        Mon, 14 Jan 2002 11:14:16 +0200
Message-ID: <02a601c19cdb$5befd620$6b24b183@etela.sonera.fi>
From: "Andrei Gurtov" <gurtov@cs.helsinki.fi>
To: <mallman@grc.nasa.gov>
Cc: "Kacheong Poon" <kcpoon@eng.sun.com>,
        "Reiner Ludwig" <Reiner.Ludwig@ericsson.com>,
        "sanjay kaniyar" <skaniyar@HOTMAIL.COM>, <tsvwg@ietf.org>
References: <200201131815.NAA09683@lawyers.grc.nasa.gov>
Subject: Re: [Tsvwg] Eifel draft and usage of time-stamps. 
Date:   Mon, 14 Jan 2002 11:10:49 +0200
MIME-Version: 1.0
Content-Type: text/plain;
        charset="iso-8859-1"
Content-Transfer-Encoding: 7bit
X-Priority: 3
X-MSMail-Priority: Normal
X-Mailer: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Content-Transfer-Encoding: 7bit
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit

Unfortunately, I forgot an important comment.

Reverting CC state is in a way tightly coupled with preventing go-back-n,
because setting snd_nxt to snd_max reverts the estimate of the flight size
as well! This is a crucial thing to do since vanilla TCP makes an assumption
that all segments outstanding before the timeout have left the network and
it's not the case after a spurious timeout. If flight size is reverted,
fully restoring CC (both cwnd and ssthresh) does not lead to unwanted
bursts. If CC is not reverted, then a part of ACKs returning after the delay
spike ends are ignored until the flight size reaches the reduced value of
cwnd. If cwnd is a half of the value before the spurious timeout, you are
idle for half of RTT. If CC is not reverted (cwnd is one), you end up being
idle until all packets drain from the network.

Andrei


_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Mon Jan 14 06:05:10 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id GAA01121
	for <tsvwg-archive@odin.ietf.org>; Mon, 14 Jan 2002 06:05:05 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id GAA02718
	for tsvwg-archive@odin.ietf.org; Mon, 14 Jan 2002 06:05:08 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id FAA02142;
	Mon, 14 Jan 2002 05:49:52 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id FAA02111
	for <tsvwg@optimus.ietf.org>; Mon, 14 Jan 2002 05:49:50 -0500 (EST)
Received: from sj-msg-core-4.cisco.com (sj-msg-core-4.cisco.com [171.71.163.10])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id FAA01002
	for <tsvwg@ietf.org>; Mon, 14 Jan 2002 05:49:46 -0500 (EST)
Received: from mira-sjc5-2.cisco.com (mira-sjc5-2.cisco.com [171.71.163.16])
	by sj-msg-core-4.cisco.com (8.11.3/8.9.1) with ESMTP id g0EAnJJ04623;
	Mon, 14 Jan 2002 02:49:19 -0800 (PST)
Received: from stewart.chicago.il.us (ssh-sj1.cisco.com [171.68.225.134])
	by mira-sjc5-2.cisco.com (Mirapoint)
	with ESMTP id AAT14263;
	Mon, 14 Jan 2002 02:49:17 -0800 (PST)
Message-ID: <3C42B7AC.CEF70A0D@stewart.chicago.il.us>
Date: Mon, 14 Jan 2002 04:49:17 -0600
From: Randall Stewart <randall@stewart.chicago.il.us>
X-Mailer: Mozilla 4.76 [en] (X11; U; Linux 2.2.12 i386)
X-Accept-Language: en
MIME-Version: 1.0
To: ygahlaut@hss.hns.com
CC: tsvwg@ietf.org
Subject: Re: [Tsvwg] Section 7.1.3 sctpsocket-02.txt
References: <65256B41.0016E2E4.00@sandesh.hss.hns.com>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Transfer-Encoding: 7bit
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit

Yogesh:

To quote that section:

"
    Setting initialization parameters is effective only on an
    unconnected socket (for UDP-style sockets only future associations
    are effected by the change). With TCP-style sockets, this option is
    inherited by sockets derived from a listener socket.
"

You do not GET anything back from this request. If you do a
getsocketopt() with this parameter you will get back what the 
current default is. For your example 2... This has nothing
to do with current associations but what FUTURE associations
will be setup with.

Now if you want the specific details on an association you
turn on the flag to receive association change events..


Which will cause you to receive a SCTP_ASOC_CHANGE event telling you how
many 
streams you have in/out.


Now you do bring up a good point. If I don't have this on
and want to know how many streams there are there is no
way to tell..


Hmm.. maybe we should add two parameters to the SCTP_STATUS
structure to tell you the in/out streams..


R


ygahlaut@hss.hns.com wrote:
> 
> Randall :
>      thanks for for answer. i have one more doubt ...
>   lets consider this sequence of events
> 
>    open a udp socket .
>    make association with peer endpoint A with streams setting X  and got
> assoc_id 1
>    set some diifernt no of in and out streams using setsockopt
>   make another association with other peer endpoint B with streams setting Y and
> got assoc_id 2
>   and now if i use getsockopt with socket option SCTP_INITMSG
>   what should i get stream info from assoc_id 1 or 2....
> the stream setting X and Y  may be different ( after stream negotaition )
> is this sequence valid or i am missing something
> i hope you understand my doubt .
> 
> thanks
> yogesh
> 
> Randall Stewart <randall@stewart.chicago.il.us> on 01/11/2002 08:42:32 PM
> 
> To:   Yogesh Gahlaut/HSS@HSS
> cc:   tsvwg@ietf.org
> 
> Subject:  Re: [Tsvwg] Section 7.1.3 sctpsocket-02.txt
> 
> Yogesh:
> 
> Two things:
> 
> First:
> 
> The sctp_initmsg is used in the UDP model to start
> a new association with a peer (i.e. you are sending the
> INIT message out) and over-ride the default INIT parameters
> with the value found in the sctp_initmsg structure.
> 
> So we have an app sending a message with a
> sendmsg() and putting the sctp_initmsg structure
> in a CMSG header to go with the first message you
> are sending. The sendmsg() call also includes the
> destination address in the msg_name field ... i.e. the
> sockaddr * of who you are sending (and INIT'ing) to.
> 
> Second:
> 
> If you are asking instead about notifications, i.e.
> the association just came up and how do I tell what
> parameters were negotiated in the streams counts
> etc.. you get this in a MSG_NOTIFICATION in the
> form of a
> 
> struct sctp_assoc_change x;
> 
> And it too is received via a recvmsg() call where
> the msg_flags field gets set to MSG_NOTIFICATION and the
> msg_name will have the peer that the association just came
> up with .. the data read will be the  sctp_assoc_change..
> 
> Of course you must enable this event...
> 
> Hope that helps...
> 
> R
> 
> ygahlaut@hss.hns.com wrote:
> >
> > hi all,
> >
> >   in section 7.1.3   the following structure is being used for returning the
> > SCTP_INITMSG option value
> >
> > struct sctp_initmsg {
> >        uint16_t sinit_num_ostreams;
> >        uint16_t sinit_max_instreams;
> >        uint16_t sinit_max_attempts;
> >        uint16_t sinit_max_init_timeo;
> >     };
> >
> >  Since it does not have any input parameter to differentiate in case of more
> > than
> > one association in UDP style socket .
> >
> > can anybody explain how to handle this case ?
> >
> > thanks
> > Yogesh
> >
> > _______________________________________________
> > tsvwg mailing list
> > tsvwg@ietf.org
> > http://www1.ietf.org/mailman/listinfo/tsvwg
> 
> --
> Randall R. Stewart
> randall@stewart.chicago.il.us 815-342-5222 (cell phone)
> 
> _______________________________________________
> tsvwg mailing list
> tsvwg@ietf.org
> http://www1.ietf.org/mailman/listinfo/tsvwg
> 
> _______________________________________________
> tsvwg mailing list
> tsvwg@ietf.org
> http://www1.ietf.org/mailman/listinfo/tsvwg

-- 
Randall R. Stewart
randall@stewart.chicago.il.us 815-342-5222 (cell phone)

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Mon Jan 14 08:06:47 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id IAA02099
	for <tsvwg-archive@odin.ietf.org>; Mon, 14 Jan 2002 08:06:46 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id IAA05956
	for tsvwg-archive@odin.ietf.org; Mon, 14 Jan 2002 08:06:49 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id HAA05199;
	Mon, 14 Jan 2002 07:45:08 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id HAA05168
	for <tsvwg@optimus.ietf.org>; Mon, 14 Jan 2002 07:45:06 -0500 (EST)
Received: from lawyers.grc.nasa.gov (d118.as1.clev.oh.voyager.net [209.81.166.119])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id HAA01886
	for <tsvwg@ietf.org>; Mon, 14 Jan 2002 07:45:03 -0500 (EST)
Received: from lawyers (mallman@localhost)
	by lawyers.grc.nasa.gov (8.9.3/8.9.3) with ESMTP id HAA11345;
	Mon, 14 Jan 2002 07:24:36 -0500
Message-Id: <200201141224.HAA11345@lawyers.grc.nasa.gov>
To: "Andrei Gurtov" <gurtov@cs.helsinki.fi>
From: Mark Allman <mallman@grc.nasa.gov>
Reply-To: mallman@grc.nasa.gov
cc: "Kacheong Poon" <kcpoon@eng.sun.com>,
        "Reiner Ludwig" <Reiner.Ludwig@ericsson.com>,
        "sanjay kaniyar" <skaniyar@hotmail.com>, tsvwg@ietf.org
Subject: Re: [Tsvwg] Eifel draft and usage of time-stamps. 
Organization: BBN Technologies/NASA GRC
Song-of-the-Day: Hammer to Fall
Date: Mon, 14 Jan 2002 07:24:35 -0500
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org


While I don't agree with much of your note I have said everything I
want to say more than once in this thread.  So, just this...

> It seems that rfc2988 with patches from the tech report is quite
> conservative and the sender is unlikely to go into chain of
> spurious timeouts. It may well be that letting Eifel to handle
> occasional delay spikes is a better solution than to invent some
> overly complicated timer to predict them.

Occasional spurious retransmits do not bother me at all.  In that
case, we should be able to revert the CC state (whatever we decide
that means) and continue on.  The case that concerns me more is that
of persistent conditions that cause spurious retransmits in the
network to occur regularly.  In that case, I think we need some sort
of throttle to make sure that we are not being over-aggressive in
piling unnecessay packets into the network.  That throttle could be
any number of things: adjusting what RTO algorithm we use, tweaking
the current algorithm on the fly a bit, adjusting the number of dup
ACKs needed to fast retransmit, turning the CC state undo-ing off if
a certain percentage of our transmissions are spurious, etc., etc.,
etc.  There are any number of a ton of things that one could
envision doing.  I am not advocating for any one particular scheme,
just argueing that there needs to be something that keeps us from
the case of agressively transmitting needless junk into the
network. 

allman


---
Mark Allman -- NASA GRC/BBN -- http://roland.grc.nasa.gov/~mallman/

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Mon Jan 14 08:10:37 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id IAA02146
	for <tsvwg-archive@odin.ietf.org>; Mon, 14 Jan 2002 08:10:37 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id IAA05995
	for tsvwg-archive@odin.ietf.org; Mon, 14 Jan 2002 08:10:40 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id HAA05316;
	Mon, 14 Jan 2002 07:49:38 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id HAA05285
	for <tsvwg@optimus.ietf.org>; Mon, 14 Jan 2002 07:49:36 -0500 (EST)
Received: from hindon.hss.co.in ([202.54.26.202])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id HAA01919
	for <tsvwg@ietf.org>; Mon, 14 Jan 2002 07:49:32 -0500 (EST)
From: ygahlaut@hss.hns.com
Received: from sandesh.hss.hns.com (sandesh [139.85.242.35])
	by hindon.hss.co.in (8.10.0/8.10.0) with SMTP id g0ECo5E04696;
	Mon, 14 Jan 2002 18:20:05 +0530 (IST)
Received: by sandesh.hss.hns.com(Lotus SMTP MTA v4.6.3  (733.2 10-16-1998))  id 65256B41.00465FA8 ; Mon, 14 Jan 2002 18:18:40 +0530
X-Lotus-FromDomain: HSS
To: Randall Stewart <randall@stewart.chicago.il.us>
cc: tsvwg@ietf.org
Message-ID: <65256B41.004514FD.00@sandesh.hss.hns.com>
Date: Mon, 14 Jan 2002 18:05:53 +0530
Subject: Re: [Tsvwg] Section 7.1.3 sctpsocket-02.txt
Mime-Version: 1.0
Content-type: text/plain; charset=us-ascii
Content-Disposition: inline
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org



Randall,
thanks for clarification
    i think there should be more clear explanation in this para that
 getsockopt  call with SCTP_INITMSG will return only the values set by user or
dafault
may be inherited from server socket ( in case of TCP style) and these values may
be different
from the actual values ( after stream negotaition ) . ofcouse these will be used
for future
associations.  and to get actual values use SCTP_STATUS ( if you add the new
fields)

i have one one doubt ragarding assoc_change notification
in TCP style interface
the scenerio is  a server socket (say A) is listening for connection
a connection request comes and connection is made successfully
now i should get assoc_change notification (assume it is enabled) on server
socket before accept      OR
assoc_change notification on new socket descriptor after accept call

my understanding is assoc_change should come on new socket after accept call

if it is implementation dependent issue still it should be same for ULP
i hope you understand my doubt
thank
yogesh





Randall Stewart <randall@stewart.chicago.il.us> on 01/14/2002 04:19:17 PM

To:   Yogesh Gahlaut/HSS@HSS
cc:   tsvwg@ietf.org

Subject:  Re: [Tsvwg] Section 7.1.3 sctpsocket-02.txt




Yogesh:

To quote that section:

"
    Setting initialization parameters is effective only on an
    unconnected socket (for UDP-style sockets only future associations
    are effected by the change). With TCP-style sockets, this option is
    inherited by sockets derived from a listener socket.
"

You do not GET anything back from this request. If you do a
getsocketopt() with this parameter you will get back what the
current default is. For your example 2... This has nothing
to do with current associations but what FUTURE associations
will be setup with.

Now if you want the specific details on an association you
turn on the flag to receive association change events..


Which will cause you to receive a SCTP_ASOC_CHANGE event telling you how
many
streams you have in/out.


Now you do bring up a good point. If I don't have this on
and want to know how many streams there are there is no
way to tell..


Hmm.. maybe we should add two parameters to the SCTP_STATUS
structure to tell you the in/out streams..


R


ygahlaut@hss.hns.com wrote:
>
> Randall :
>      thanks for for answer. i have one more doubt ...
>   lets consider this sequence of events
>
>    open a udp socket .
>    make association with peer endpoint A with streams setting X  and got
> assoc_id 1
>    set some diifernt no of in and out streams using setsockopt
>   make another association with other peer endpoint B with streams setting Y
and
> got assoc_id 2
>   and now if i use getsockopt with socket option SCTP_INITMSG
>   what should i get stream info from assoc_id 1 or 2....
> the stream setting X and Y  may be different ( after stream negotaition )
> is this sequence valid or i am missing something
> i hope you understand my doubt .
>
> thanks
> yogesh
>
> Randall Stewart <randall@stewart.chicago.il.us> on 01/11/2002 08:42:32 PM
>
> To:   Yogesh Gahlaut/HSS@HSS
> cc:   tsvwg@ietf.org
>
> Subject:  Re: [Tsvwg] Section 7.1.3 sctpsocket-02.txt
>
> Yogesh:
>
> Two things:
>
> First:
>
> The sctp_initmsg is used in the UDP model to start
> a new association with a peer (i.e. you are sending the
> INIT message out) and over-ride the default INIT parameters
> with the value found in the sctp_initmsg structure.
>
> So we have an app sending a message with a
> sendmsg() and putting the sctp_initmsg structure
> in a CMSG header to go with the first message you
> are sending. The sendmsg() call also includes the
> destination address in the msg_name field ... i.e. the
> sockaddr * of who you are sending (and INIT'ing) to.
>
> Second:
>
> If you are asking instead about notifications, i.e.
> the association just came up and how do I tell what
> parameters were negotiated in the streams counts
> etc.. you get this in a MSG_NOTIFICATION in the
> form of a
>
> struct sctp_assoc_change x;
>
> And it too is received via a recvmsg() call where
> the msg_flags field gets set to MSG_NOTIFICATION and the
> msg_name will have the peer that the association just came
> up with .. the data read will be the  sctp_assoc_change..
>
> Of course you must enable this event...
>
> Hope that helps...
>
> R
>
> ygahlaut@hss.hns.com wrote:
> >
> > hi all,
> >
> >   in section 7.1.3   the following structure is being used for returning the
> > SCTP_INITMSG option value
> >
> > struct sctp_initmsg {
> >        uint16_t sinit_num_ostreams;
> >        uint16_t sinit_max_instreams;
> >        uint16_t sinit_max_attempts;
> >        uint16_t sinit_max_init_timeo;
> >     };
> >
> >  Since it does not have any input parameter to differentiate in case of more
> > than
> > one association in UDP style socket .
> >
> > can anybody explain how to handle this case ?
> >
> > thanks
> > Yogesh
> >
> > _______________________________________________
> > tsvwg mailing list
> > tsvwg@ietf.org
> > http://www1.ietf.org/mailman/listinfo/tsvwg
>
> --
> Randall R. Stewart
> randall@stewart.chicago.il.us 815-342-5222 (cell phone)
>
> _______________________________________________
> tsvwg mailing list
> tsvwg@ietf.org
> http://www1.ietf.org/mailman/listinfo/tsvwg
>
> _______________________________________________
> tsvwg mailing list
> tsvwg@ietf.org
> http://www1.ietf.org/mailman/listinfo/tsvwg

--
Randall R. Stewart
randall@stewart.chicago.il.us 815-342-5222 (cell phone)

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg





_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Mon Jan 14 08:49:24 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id IAA02496
	for <tsvwg-archive@odin.ietf.org>; Mon, 14 Jan 2002 08:49:23 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id IAA07010
	for tsvwg-archive@odin.ietf.org; Mon, 14 Jan 2002 08:49:26 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id IAA06352;
	Mon, 14 Jan 2002 08:29:20 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id IAA06276
	for <tsvwg@optimus.ietf.org>; Mon, 14 Jan 2002 08:29:16 -0500 (EST)
Received: from sj-msg-core-1.cisco.com (sj-msg-core-1.cisco.com [171.71.163.11])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id IAA02319
	for <tsvwg@ietf.org>; Mon, 14 Jan 2002 08:29:12 -0500 (EST)
Received: from mira-sjc5-2.cisco.com (mira-sjc5-2.cisco.com [171.71.163.16])
	by sj-msg-core-1.cisco.com (8.11.3/8.9.1) with ESMTP id g0EDSiF04173;
	Mon, 14 Jan 2002 05:28:44 -0800 (PST)
Received: from stewart.chicago.il.us (ssh-sj1.cisco.com [171.68.225.134])
	by mira-sjc5-2.cisco.com (Mirapoint)
	with ESMTP id AAT14893;
	Mon, 14 Jan 2002 05:28:41 -0800 (PST)
Message-ID: <3C42DD09.670184AF@stewart.chicago.il.us>
Date: Mon, 14 Jan 2002 07:28:41 -0600
From: Randall Stewart <randall@stewart.chicago.il.us>
X-Mailer: Mozilla 4.76 [en] (X11; U; Linux 2.2.12 i386)
X-Accept-Language: en
MIME-Version: 1.0
To: ygahlaut@hss.hns.com
CC: tsvwg@ietf.org
Subject: Re: [Tsvwg] Section 7.1.3 sctpsocket-02.txt
References: <65256B41.004514FD.00@sandesh.hss.hns.com>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Transfer-Encoding: 7bit
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit

ygahlaut@hss.hns.com wrote:
> 
> Randall,
> thanks for clarification
>     i think there should be more clear explanation in this para that
>  getsockopt  call with SCTP_INITMSG will return only the values set by user or
> dafault
> may be inherited from server socket ( in case of TCP style) and these values may
> be different
> from the actual values ( after stream negotaition ) . ofcouse these will be used
> for future
> associations.  and to get actual values use SCTP_STATUS ( if you add the new
> fields)
> 

We probably should :>

> i have one one doubt ragarding assoc_change notification
> in TCP style interface
> the scenerio is  a server socket (say A) is listening for connection
> a connection request comes and connection is made successfully
> now i should get assoc_change notification (assume it is enabled) on server
> socket before accept      OR
> assoc_change notification on new socket descriptor after accept call
> 
> my understanding is assoc_change should come on new socket after accept call
> 
> if it is implementation dependent issue still it should be same for ULP

I don't understand this last line.. "should be same for ULP" what do
you mean... that even if it is implementation dependant the ULP should
see the same thing? Or something else?



> i hope you understand my doubt
> thank


No I think you need to do it AFTER the accept call. This is because
you really don't want the assoc-change event to show up 
on the listen() socket.. but the new one. Thus the socket does
not exist until after the accept() which creates the fd.

But maybe it should be implementation dependant... I would defintly
welcome other comments/opinions on this..

R


> 
> Randall Stewart <randall@stewart.chicago.il.us> on 01/14/2002 04:19:17 PM
> 
> To:   Yogesh Gahlaut/HSS@HSS
> cc:   tsvwg@ietf.org
> 
> Subject:  Re: [Tsvwg] Section 7.1.3 sctpsocket-02.txt
> 
> Yogesh:
> 
> To quote that section:
> 
> "
>     Setting initialization parameters is effective only on an
>     unconnected socket (for UDP-style sockets only future associations
>     are effected by the change). With TCP-style sockets, this option is
>     inherited by sockets derived from a listener socket.
> "
> 
> You do not GET anything back from this request. If you do a
> getsocketopt() with this parameter you will get back what the
> current default is. For your example 2... This has nothing
> to do with current associations but what FUTURE associations
> will be setup with.
> 
> Now if you want the specific details on an association you
> turn on the flag to receive association change events..
> 
> Which will cause you to receive a SCTP_ASOC_CHANGE event telling you how
> many
> streams you have in/out.
> 
> Now you do bring up a good point. If I don't have this on
> and want to know how many streams there are there is no
> way to tell..
> 
> Hmm.. maybe we should add two parameters to the SCTP_STATUS
> structure to tell you the in/out streams..
> 
> R
> 
> ygahlaut@hss.hns.com wrote:
> >
> > Randall :
> >      thanks for for answer. i have one more doubt ...
> >   lets consider this sequence of events
> >
> >    open a udp socket .
> >    make association with peer endpoint A with streams setting X  and got
> > assoc_id 1
> >    set some diifernt no of in and out streams using setsockopt
> >   make another association with other peer endpoint B with streams setting Y
> and
> > got assoc_id 2
> >   and now if i use getsockopt with socket option SCTP_INITMSG
> >   what should i get stream info from assoc_id 1 or 2....
> > the stream setting X and Y  may be different ( after stream negotaition )
> > is this sequence valid or i am missing something
> > i hope you understand my doubt .
> >
> > thanks
> > yogesh
> >
> > Randall Stewart <randall@stewart.chicago.il.us> on 01/11/2002 08:42:32 PM
> >
> > To:   Yogesh Gahlaut/HSS@HSS
> > cc:   tsvwg@ietf.org
> >
> > Subject:  Re: [Tsvwg] Section 7.1.3 sctpsocket-02.txt
> >
> > Yogesh:
> >
> > Two things:
> >
> > First:
> >
> > The sctp_initmsg is used in the UDP model to start
> > a new association with a peer (i.e. you are sending the
> > INIT message out) and over-ride the default INIT parameters
> > with the value found in the sctp_initmsg structure.
> >
> > So we have an app sending a message with a
> > sendmsg() and putting the sctp_initmsg structure
> > in a CMSG header to go with the first message you
> > are sending. The sendmsg() call also includes the
> > destination address in the msg_name field ... i.e. the
> > sockaddr * of who you are sending (and INIT'ing) to.
> >
> > Second:
> >
> > If you are asking instead about notifications, i.e.
> > the association just came up and how do I tell what
> > parameters were negotiated in the streams counts
> > etc.. you get this in a MSG_NOTIFICATION in the
> > form of a
> >
> > struct sctp_assoc_change x;
> >
> > And it too is received via a recvmsg() call where
> > the msg_flags field gets set to MSG_NOTIFICATION and the
> > msg_name will have the peer that the association just came
> > up with .. the data read will be the  sctp_assoc_change..
> >
> > Of course you must enable this event...
> >
> > Hope that helps...
> >
> > R
> >
> > ygahlaut@hss.hns.com wrote:
> > >
> > > hi all,
> > >
> > >   in section 7.1.3   the following structure is being used for returning the
> > > SCTP_INITMSG option value
> > >
> > > struct sctp_initmsg {
> > >        uint16_t sinit_num_ostreams;
> > >        uint16_t sinit_max_instreams;
> > >        uint16_t sinit_max_attempts;
> > >        uint16_t sinit_max_init_timeo;
> > >     };
> > >
> > >  Since it does not have any input parameter to differentiate in case of more
> > > than
> > > one association in UDP style socket .
> > >
> > > can anybody explain how to handle this case ?
> > >
> > > thanks
> > > Yogesh
> > >
> > > _______________________________________________
> > > tsvwg mailing list
> > > tsvwg@ietf.org
> > > http://www1.ietf.org/mailman/listinfo/tsvwg
> >
> > --
> > Randall R. Stewart
> > randall@stewart.chicago.il.us 815-342-5222 (cell phone)
> >
> > _______________________________________________
> > tsvwg mailing list
> > tsvwg@ietf.org
> > http://www1.ietf.org/mailman/listinfo/tsvwg
> >
> > _______________________________________________
> > tsvwg mailing list
> > tsvwg@ietf.org
> > http://www1.ietf.org/mailman/listinfo/tsvwg
> 
> --
> Randall R. Stewart
> randall@stewart.chicago.il.us 815-342-5222 (cell phone)
> 
> _______________________________________________
> tsvwg mailing list
> tsvwg@ietf.org
> http://www1.ietf.org/mailman/listinfo/tsvwg

-- 
Randall R. Stewart
randall@stewart.chicago.il.us 815-342-5222 (cell phone)

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Mon Jan 14 11:14:33 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id LAA05455
	for <tsvwg-archive@odin.ietf.org>; Mon, 14 Jan 2002 11:14:33 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id LAA12296
	for tsvwg-archive@odin.ietf.org; Mon, 14 Jan 2002 11:14:35 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id KAA11235;
	Mon, 14 Jan 2002 10:44:11 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id KAA11204
	for <tsvwg@optimus.ietf.org>; Mon, 14 Jan 2002 10:44:09 -0500 (EST)
Received: from gorilla.mchh.siemens.de (gorilla.mchh.siemens.de [194.138.158.18])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id KAA04733
	for <tsvwg@ietf.org>; Mon, 14 Jan 2002 10:44:05 -0500 (EST)
Received: from blues.mchh.siemens.de (mail2.mchh.siemens.de [194.138.158.227])
	by gorilla.mchh.siemens.de (8.9.3/8.9.3) with ESMTP id QAA09723;
	Mon, 14 Jan 2002 16:43:56 +0100 (MET)
Received: from mchh246e.demchh201e.icn.siemens.de ([139.21.200.56])
	by blues.mchh.siemens.de (8.9.1/8.9.1) with ESMTP id QAA03158;
	Mon, 14 Jan 2002 16:43:58 +0100 (MET)
Received: by MCHH246E with Internet Mail Service (5.5.2653.19)
	id <ZWXSL3CF>; Mon, 14 Jan 2002 16:43:58 +0100
Message-ID: <F92978223B01D311ACFF0008C71EE19D015DF6ED@MCHH225E>
From: Tuexen Michael <Michael.Tuexen@icn.siemens.de>
To: "'tsvwg@ietf.org'" <tsvwg@ietf.org>
Cc: "'rrs@cisco.com'" <rrs@cisco.com>
Date: Mon, 14 Jan 2002 16:43:53 +0100
MIME-Version: 1.0
X-Mailer: Internet Mail Service (5.5.2653.19)
Content-Type: text/plain
Subject: [Tsvwg] SCTP checksum ID comment
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

Dear all,

looking at the code on page 9 there are constants
NMIN and NMAX used. I think they describe the message
length limits between the given algorithm is valid.

But there are no values given for NMIN and NMAX. Could
you state in the ID for what message length the algorthm is
valid. Or simply remove the lines of code which do the check.

Best regards
Michael

PS.: One editorial comment: my name should be Tuexen, not Tuxen in section 3.
Michael Tuexen    Tel.:   +49 89 722 47210
Siemens AG        Fax:    +49 89 722 48212
ICN WN CS SE 5    E-mail: Michael.Tuexen@icn.siemens.de


_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Mon Jan 14 11:23:53 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id LAA05665
	for <tsvwg-archive@odin.ietf.org>; Mon, 14 Jan 2002 11:23:48 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id LAA12483
	for tsvwg-archive@odin.ietf.org; Mon, 14 Jan 2002 11:23:50 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id LAA12108;
	Mon, 14 Jan 2002 11:01:55 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id LAA12077
	for <tsvwg@optimus.ietf.org>; Mon, 14 Jan 2002 11:01:53 -0500 (EST)
Received: from sj-msg-core-1.cisco.com (sj-msg-core-1.cisco.com [171.71.163.11])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id LAA05040
	for <tsvwg@ietf.org>; Mon, 14 Jan 2002 11:01:50 -0500 (EST)
Received: from mira-sjc5-2.cisco.com (mira-sjc5-2.cisco.com [171.71.163.16])
	by sj-msg-core-1.cisco.com (8.11.3/8.9.1) with ESMTP id g0EG1MF26445;
	Mon, 14 Jan 2002 08:01:22 -0800 (PST)
Received: from cisco.com (ssh-sj1.cisco.com [171.68.225.134])
	by mira-sjc5-2.cisco.com (Mirapoint)
	with ESMTP id AAT16453;
	Mon, 14 Jan 2002 08:01:14 -0800 (PST)
Message-ID: <3C4300CA.74C96629@cisco.com>
Date: Mon, 14 Jan 2002 10:01:14 -0600
From: Randall Stewart at work <rrs@cisco.com>
X-Mailer: Mozilla 4.76 [en] (X11; U; Linux 2.2.12 i386)
X-Accept-Language: en
MIME-Version: 1.0
To: Tuexen Michael <Michael.Tuexen@icn.siemens.de>
CC: "'tsvwg@ietf.org'" <tsvwg@ietf.org>
References: <F92978223B01D311ACFF0008C71EE19D015DF6ED@MCHH225E>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Transfer-Encoding: 7bit
Subject: [Tsvwg] Re: SCTP checksum ID comment
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit

Tuexen Michael wrote:
> 
> Dear all,
> 
> looking at the code on page 9 there are constants
> NMIN and NMAX used. I think they describe the message
> length limits between the given algorithm is valid.
> 
> But there are no values given for NMIN and NMAX. Could
> you state in the ID for what message length the algorthm is
> valid. Or simply remove the lines of code which do the check.
> 
> Best regards
> Michael
> 
> PS.: One editorial comment: my name should be Tuexen, not Tuxen in section 3.
> Michael Tuexen    Tel.:   +49 89 722 47210
> Siemens AG        Fax:    +49 89 722 48212
> ICN WN CS SE 5    E-mail: Michael.Tuexen@icn.siemens.de
Yes,

I think we should remove these lines.. since this is
what I did when I pulled it into our
FreeBSD implementation :-)

R
-- 
Randall R. Stewart
ITD
Cisco Systems Inc.
rrs@cisco.com 815-342-5222 or 815-477-2127

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Mon Jan 14 12:23:28 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id MAA08035
	for <tsvwg-archive@odin.ietf.org>; Mon, 14 Jan 2002 12:23:28 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id MAA15359
	for tsvwg-archive@odin.ietf.org; Mon, 14 Jan 2002 12:23:31 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id MAA14867;
	Mon, 14 Jan 2002 12:05:44 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id MAA14836
	for <tsvwg@optimus.ietf.org>; Mon, 14 Jan 2002 12:05:41 -0500 (EST)
Received: from seraph3.grc.nasa.gov (firewall-user@seraph3.lerc.nasa.gov [128.156.10.12])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id MAA07143
	for <tsvwg@ietf.org>; Mon, 14 Jan 2002 12:05:37 -0500 (EST)
Received: by seraph3.grc.nasa.gov (Postfix, from userid 5)
	id AC12C640A5; Mon, 14 Jan 2002 12:05:39 -0500 (EST)
Received: from lombok-fi.lerc.nasa.gov(139.88.112.33) by seraph3.grc.nasa.gov via csmap (V6.0)
	id srcAAAYcaibJ; Mon, 14 Jan 02 12:05:39 -0500
Received: from guns.lerc.nasa.gov (guns.lerc.nasa.gov [139.88.87.35])
	by lombok-fi.lerc.nasa.gov (NASA LeRC 8.9.1.1/8.9.1) with ESMTP id MAA24102;
	Mon, 14 Jan 2002 12:05:38 -0500 (EST)
Received: from guns.lerc.nasa.gov (localhost.lerc.nasa.gov [127.0.0.1]) by guns.lerc.nasa.gov with ESMTP (NASA LeRC 8.7.4.1/2.01-local)
        id MAA77924; Mon, 14 Jan 2002 12:05:39 -0500 (EST)
Message-Id: <200201141705.MAA77924@guns.lerc.nasa.gov>
To: tsvwg@ietf.org
Cc: "Ethan Blanton" <eblanton@prime.cs.ohiou.edu>
From: Mark Allman <mallman@grc.nasa.gov>
Reply-To: mallman@grc.nasa.gov
Organization: BBN Technologies/NASA GRC
Song-of-the-Day: Honky Tonk Woman
Date: Mon, 14 Jan 2002 12:05:39 -0500
Subject: [Tsvwg] paper on TCP and packet reordering
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

 
[Appologies if you see this announcement more than once.]
 
Folks-

As I mentioned late last week Ethan Blanton and I have just finished
up a paper on TCP's performance in the face of packet reordering and
what might be done about the performance degradation.  The paper is
now available if anyone is interested.

    Ethan Blanton, Mark Allman. On Making TCP More Robust to Packet 
    Reordering. ACM Computer Communication Review, 32(1), January
    2002.
    http://roland.grc.nasa.gov/~mallman/papers/tcp-reorder-ccr.ps

Abstract:

    Previous research indicates that packet reordering is not a rare
    event on some Internet paths.  Reordering can cause performance
    problems for TCP's fast retransmission algorithm, which uses the
    arrival of duplicate acknowledgments to detect segment loss.
    Duplicate acknowledgments can be caused by the loss of a segment
    or by the reordering of segments by the network.  In this paper we
    illustrate the impact of reordering on TCP performance.  In
    addition, we show the performance of a conservative approach to
    ``undo'' the congestion control state changes made in conjunction
    with spurious retransmissions.  Finally, we propose several
    alternatives to dynamically make the fast retransmission algorithm
    more tolerant of the reordering observed in the network and assess
    these algorithms.

We'd be happy to hear any comments you might have.

allman


---
Mark Allman -- BBN/NASA GRC -- http://roland.grc.nasa.gov/~mallman/

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Mon Jan 14 16:23:49 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id QAA19964
	for <tsvwg-archive@odin.ietf.org>; Mon, 14 Jan 2002 16:23:45 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id QAA26031
	for tsvwg-archive@odin.ietf.org; Mon, 14 Jan 2002 16:23:48 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id QAA24811;
	Mon, 14 Jan 2002 16:00:51 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id QAA24782
	for <tsvwg@optimus.ietf.org>; Mon, 14 Jan 2002 16:00:49 -0500 (EST)
Received: from sj-msg-core-2.cisco.com (sj-msg-core-2.cisco.com [171.69.24.11])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id QAA18929
	for <tsvwg@ietf.org>; Mon, 14 Jan 2002 16:00:44 -0500 (EST)
Received: from mira-sjc5-2.cisco.com (mira-sjc5-2.cisco.com [171.71.163.16])
	by sj-msg-core-2.cisco.com (8.11.3/8.9.1) with ESMTP id g0EL0E414137
	for <tsvwg@ietf.org>; Mon, 14 Jan 2002 13:00:15 -0800 (PST)
Received: from stewart.chicago.il.us (ssh-sj1.cisco.com [171.68.225.134])
	by mira-sjc5-2.cisco.com (Mirapoint)
	with ESMTP id AAT26129;
	Mon, 14 Jan 2002 13:00:13 -0800 (PST)
Message-ID: <3C4346DC.D5F3301B@stewart.chicago.il.us>
Date: Mon, 14 Jan 2002 15:00:12 -0600
From: Randall Stewart <randall@stewart.chicago.il.us>
X-Mailer: Mozilla 4.76 [en] (X11; U; Linux 2.2.12 i386)
X-Accept-Language: en
MIME-Version: 1.0
To: TSV WG list <tsvwg@ietf.org>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Transfer-Encoding: 7bit
Subject: [Tsvwg] fragmentation and PMTU
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit

Dear All:

It has come to my attention that the following
is currently in RFC2960:

"
6.9 Fragmentation and Reassembly

   An endpoint MAY support fragmentation when sending DATA chunks, but
   MUST support reassembly when receiving DATA chunks.  If an endpoint
   supports fragmentation, it MUST fragment a user message if the size
   of the user message to be sent causes the outbound SCTP packet size
   to exceed the current MTU.  If an implementation does not support
   fragmentation of outbound user messages, the endpoint must return an
   error to its upper layer and not attempt to send the user message.
"

This all seems fine until one looks carefully at one little aspect:

- If I an implementation supports fragmentation it MUST fragment.

The issue I have with this is what if an application layer says:

"hey I know you can fragment, but please don't, I want you
 to reject sends that exceed the P-MTU."

This may be a perfectly reasonable behavior but the wording above
dis-allows that... I know the intent was to prevent folks from
just setting the DF bit and NOT doing P-MTU discovery... but what
we probably need here is a little clearification for the implementors
guide... something like.

"
 Note: If an implementation that supports fragmentation makes
 available to its upper layer a mechanism to turn off fragmentation
 it may do so. However in so doing, it MUST react just like an
 implementation that does NOT support fragmentation i.e. it MUST
 reject sends that exceed the current P-MTU.
"

I think this note (+wordsmitthing :>) would allow the behavior 
I mention above without harming the intent of the P-MTU/Fragmentation.

(Word smithing suggestions would be welcome here :>)

What do you'all think?

Thanks

R

-- 
Randall R. Stewart
randall@stewart.chicago.il.us 815-342-5222 (cell phone)

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Mon Jan 14 19:32:15 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id TAA27191
	for <tsvwg-archive@odin.ietf.org>; Mon, 14 Jan 2002 19:32:14 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id TAA02658
	for tsvwg-archive@odin.ietf.org; Mon, 14 Jan 2002 19:32:16 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id TAA01899;
	Mon, 14 Jan 2002 19:07:40 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id TAA01875
	for <tsvwg@optimus.ietf.org>; Mon, 14 Jan 2002 19:07:38 -0500 (EST)
Received: from c003.snv.cp.net (c003-h015.c003.snv.cp.net [209.228.32.229])
	by ietf.org (8.9.1a/8.9.1a) with SMTP id TAA26641
	for <tsvwg@ietf.org>; Mon, 14 Jan 2002 19:07:36 -0500 (EST)
Received: (cpmta 11567 invoked from network); 14 Jan 2002 16:07:07 -0800
Received: from 64.130.130.105 (HELO dougrmt)
  by smtp.telocity.com (209.228.32.229) with SMTP; 14 Jan 2002 16:07:07 -0800
X-Sent: 15 Jan 2002 00:07:07 GMT
From: "Douglas Otis" <dotis@sanlight.net>
To: "Randall Stewart at work" <rrs@cisco.com>,
        "Tuexen Michael" <Michael.Tuexen@icn.siemens.de>
Cc: <tsvwg@ietf.org>
Subject: RE: [Tsvwg] Re: SCTP checksum ID comment
Date: Mon, 14 Jan 2002 16:07:38 -0800
Message-ID: <NEBBJGDMMLHHCIKHGBEJEENECOAA.dotis@sanlight.net>
MIME-Version: 1.0
Content-Type: text/plain;
	charset="US-ASCII"
Content-Transfer-Encoding: 7bit
X-Priority: 3 (Normal)
X-MSMail-Priority: Normal
X-Mailer: Microsoft Outlook IMO, Build 9.0.2416 (9.0.2910.0)
In-Reply-To: <3C4300CA.74C96629@cisco.com>
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4807.1700
Importance: Normal
Content-Transfer-Encoding: 7bit
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit

Michael,

These size limits are not directly related to the CRC algorithm and can be
safely removed provided there is a  check somewhere in the protocol ensuring
a proper length packet both with respect to a minimum such that the header
checksum manipulation is not beyond the end of a truncated packet or too
long to fit within system allocations.  Indeed this check can be removed,
but was included to ensure no false assumptions result from packets of
incorrect length.

Doug

> Tuexen Michael wrote:
> >
> > Dear all,
> >
> > looking at the code on page 9 there are constants
> > NMIN and NMAX used. I think they describe the message
> > length limits between the given algorithm is valid.
> >
> > But there are no values given for NMIN and NMAX. Could
> > you state in the ID for what message length the algorthm is
> > valid. Or simply remove the lines of code which do the check.
> >
> > Best regards
> > Michael
> >
> > PS.: One editorial comment: my name should be Tuexen, not Tuxen
> in section 3.
> > Michael Tuexen    Tel.:   +49 89 722 47210
> > Siemens AG        Fax:    +49 89 722 48212
> > ICN WN CS SE 5    E-mail: Michael.Tuexen@icn.siemens.de
> Yes,
>
> I think we should remove these lines.. since this is
> what I did when I pulled it into our
> FreeBSD implementation :-)
>
> R
> --
> Randall R. Stewart
> ITD
> Cisco Systems Inc.
> rrs@cisco.com 815-342-5222 or 815-477-2127
>
> _______________________________________________
> tsvwg mailing list
> tsvwg@ietf.org
> http://www1.ietf.org/mailman/listinfo/tsvwg
>


_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Mon Jan 14 21:41:21 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id VAA00917
	for <tsvwg-archive@odin.ietf.org>; Mon, 14 Jan 2002 21:41:20 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id VAA06522
	for tsvwg-archive@odin.ietf.org; Mon, 14 Jan 2002 21:41:23 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id VAA05712;
	Mon, 14 Jan 2002 21:06:04 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id VAA05681
	for <tsvwg@optimus.ietf.org>; Mon, 14 Jan 2002 21:06:02 -0500 (EST)
Received: from patan.sun.com (patan.Sun.COM [192.18.98.43])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id VAA29302
	for <tsvwg@ietf.org>; Mon, 14 Jan 2002 21:05:59 -0500 (EST)
Received: from engmail1.Eng.Sun.COM ([129.146.1.13])
	by patan.sun.com (8.9.3+Sun/8.9.3) with ESMTP id TAA06792;
	Mon, 14 Jan 2002 19:06:01 -0700 (MST)
Received: from shield (shield.EBay.Sun.COM [129.150.113.84])
	by engmail1.Eng.Sun.COM (8.9.3+Sun/8.9.3/ENSMAIL,v2.1p1) with SMTP id SAA02892;
	Mon, 14 Jan 2002 18:06:00 -0800 (PST)
Date: Mon, 14 Jan 2002 18:05:17 -0800 (PST)
From: Kacheong Poon <kcpoon@eng.sun.com>
Reply-To: Kacheong Poon <kcpoon@eng.sun.com>
Subject: Re: [Tsvwg] Eifel draft and usage of time-stamps. 
To: mallman@grc.nasa.gov
Cc: Kacheong Poon <kcpoon@eng.sun.com>,
        Reiner Ludwig <Reiner.Ludwig@ericsson.com>,
        sanjay kaniyar <skaniyar@HOTMAIL.COM>, gurtov@cs.helsinki.fi,
        tsvwg@ietf.org
In-Reply-To: "Your message with ID" <200201122237.RAA04693@lawyers.grc.nasa.gov>
Message-ID: <Roam.SIMC.2.0.6.1011060317.29073.kcpoon@eng.sun.com>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; CHARSET=US-ASCII
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

> I don't need to worry about what?  If we know we spuriously
> retransmitted something we know we added a packet to the network
> that accomplished no work (yet used resources).

You don't need to worry about TCP sending too many unnecessary
segments out because it is being too aggressive if the detection
mechanism is very good.

> I'll get a copy of our paper posted early next week.  But, we have
> an example in the paper that compares two TCPs:
> 
> 1.  a standard SACK-based version of TCP
> 2.  a version of TCP that can detect spurios fast retransmits and
>     undo the bogus CC state changes (but, takes no action to prevent
>     further needless retransmits)
> 
> This is done in a network with varying amounts of reordering -- so
> needless retransmits do happen.
> 
> What we see is that (2) gets throughput that is almost as good as
> (1) does when there is no reordering (i.e., no needless
> retransmits).  There is about a 1% performance hit if we can detect
> spurious retransmits and undo the CC changes.  On the other hand, if
> you look at the number of needless packets sent variant (2) sends
> about 6 times as many as standard SACK (variant (1)).

First, I guess the number 6 is only about the number of needless
segments.  Please correct me if I am wrong.  What is the overall ratio
of the needless segments to the number of segments in total?  I am not
suggesting that if the ratio is low, it is perfectly fine.  I just want
to point out that the 600% may seem a big number to people but the
effect of it may not be as big as it seems.  Again, if routers are
equipped to "punish" overly aggressive TCP, if the ratio of needless
segments to total segments is high, the aggressive TCP will back off.

Again, please don't get me wrong, I am not saying that we should not
look at the problems you gave above.  I just want to point out that
it may not be a problem as big as people seem to suggest.  And I
believe we do have a big problem in wireless world now, spurious
timeout is clear and present.  This is clearly my own opinion (-:

> So, while there are cases when I think you are right that there is
> an incentive because needless retransmits throw away *your*
> bandwidth and that can have a big impact, I also think there are
> cases when this incentive is negligible.
> 
> And, I think this is a problem.  I like the performance gain from
> detecting spurious retransmits.  But, I wouldn't like to see people
> pile in a lot of unnecessary packets into the network.

Point taken.  I guess we are only disagreeing on the impact.

> I do not understand this comment.  Eifel is pretty robust, it seems
> to me.  But, regardless I haven't heard of any failure cases --
> i.e., when it says there was a spurous timeout and there was not.
> Are there some?

I believe I came across a mail (reading mail after long vacation
is a pain...) pointing out that it is not.  I guess Reiner can
provide us with some data on his experience with Eifel.

							K. Poon.
							kcpoon@eng.sun.com



_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Mon Jan 14 22:51:49 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id WAA03450
	for <tsvwg-archive@odin.ietf.org>; Mon, 14 Jan 2002 22:51:49 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id WAA09149
	for tsvwg-archive@odin.ietf.org; Mon, 14 Jan 2002 22:51:53 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id WAA08351;
	Mon, 14 Jan 2002 22:31:27 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id WAA08322
	for <tsvwg@optimus.ietf.org>; Mon, 14 Jan 2002 22:31:25 -0500 (EST)
Received: from mercury.Sun.COM (mercury.Sun.COM [192.9.25.1])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id WAA02505
	for <tsvwg@ietf.org>; Mon, 14 Jan 2002 22:31:20 -0500 (EST)
Received: from engmail1.Eng.Sun.COM ([129.146.1.13])
	by mercury.Sun.COM (8.9.3+Sun/8.9.3) with ESMTP id TAA19833;
	Mon, 14 Jan 2002 19:31:23 -0800 (PST)
Received: from shield (shield.EBay.Sun.COM [129.150.113.84])
	by engmail1.Eng.Sun.COM (8.9.3+Sun/8.9.3/ENSMAIL,v2.1p1) with SMTP id TAA19918;
	Mon, 14 Jan 2002 19:31:22 -0800 (PST)
Date: Mon, 14 Jan 2002 19:30:39 -0800 (PST)
From: Kacheong Poon <kcpoon@eng.sun.com>
Reply-To: Kacheong Poon <kcpoon@eng.sun.com>
Subject: Re: [Tsvwg] Section 7.1.3 sctpsocket-02.txt
To: Randall Stewart <randall@stewart.chicago.il.us>
Cc: ygahlaut@hss.hns.com, tsvwg@ietf.org
In-Reply-To: "Your message with ID" <3C42DD09.670184AF@stewart.chicago.il.us>
Message-ID: <Roam.SIMC.2.0.6.1011065439.109.kcpoon@eng.sun.com>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; CHARSET=US-ASCII
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

> No I think you need to do it AFTER the accept call. This is because
> you really don't want the assoc-change event to show up 
> on the listen() socket.. but the new one. Thus the socket does
> not exist until after the accept() which creates the fd.
> 
> But maybe it should be implementation dependant... I would defintly
> welcome other comments/opinions on this..

First, I think apps using TCP-style socket will probably not
enable this notification.  What Randy suggested make sense.  But
then if we want to make it similar to UDP-style socket, the
notification should be sent to the listening socket.  After 
getting it, the app can do an accept().  I don't have preference
of one to the other.  It is probably safe to leave it
as implementation dependent for the moment, until we see a strong
reason to do it one way or the other.

							K. Poon.
							kcpoon@eng.sun.com



_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Tue Jan 15 07:04:05 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id HAA19146
	for <tsvwg-archive@odin.ietf.org>; Tue, 15 Jan 2002 07:04:05 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id HAA02473
	for tsvwg-archive@odin.ietf.org; Tue, 15 Jan 2002 07:04:06 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id GAA01573;
	Tue, 15 Jan 2002 06:34:22 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id GAA01545
	for <tsvwg@optimus.ietf.org>; Tue, 15 Jan 2002 06:34:20 -0500 (EST)
Received: from ms1.uibk.ac.at (ms1.uibk.ac.at [138.232.1.201])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id GAA18523
	for <tsvwg@ietf.org>; Tue, 15 Jan 2002 06:34:18 -0500 (EST)
Received: from c703-pc55.uibk.ac.at (c703-pc55.uibk.ac.at [138.232.65.57] michael.welzl@uibk.ac.at)
  by ms1.uibk.ac.at (8.11.3/E2) with ESMTP id g0FBY5W40012501;
  Tue, 15 Jan 2002 12:34:06 +0100 (MET)
Subject: Re: [Tsvwg] fragmentation and PMTU
From: Michael Welzl <michael.welzl@uibk.ac.at>
To: Randall Stewart <randall@stewart.chicago.il.us>
Cc: TSV WG list <tsvwg@ietf.org>
In-Reply-To: <3C4346DC.D5F3301B@stewart.chicago.il.us>
References: <3C4346DC.D5F3301B@stewart.chicago.il.us>
Content-Type: text/plain
Content-Transfer-Encoding: 7bit
X-Mailer: Evolution/1.0 (Preview Release)
Date: 15 Jan 2002 12:34:10 +0100
Message-Id: <1011094456.2381.4.camel@c703-pc55>
Mime-Version: 1.0
Content-Transfer-Encoding: 7bit
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit

> 6.9 Fragmentation and Reassembly
> 
>    An endpoint MAY support fragmentation when sending DATA chunks, but
>    MUST support reassembly when receiving DATA chunks.  If an endpoint
>    supports fragmentation, it MUST fragment a user message if the size
>    of the user message to be sent causes the outbound SCTP packet size
>    to exceed the current MTU.  If an implementation does not support
>    fragmentation of outbound user messages, the endpoint must return an
>    error to its upper layer and not attempt to send the user message.
> "
> 
> This all seems fine until one looks carefully at one little aspect:
> 
> - If I an implementation supports fragmentation it MUST fragment.
> 
> The issue I have with this is what if an application layer says:
> 
> "hey I know you can fragment, but please don't, I want you
>  to reject sends that exceed the P-MTU."

Couldn't we just say " ... it MUST be able to fragment" or something
like that?


>  Note: If an implementation that supports fragmentation makes
>  available to its upper layer a mechanism to turn off fragmentation
>  it may do so. However in so doing, it MUST react just like an
>  implementation that does NOT support fragmentation i.e. it MUST
>  reject sends that exceed the current P-MTU.
> "
> 
> I think this note (+wordsmitthing :>) would allow the behavior 
> I mention above without harming the intent of the P-MTU/Fragmentation.

Sounds reasonable too. Maybe I just don't understand it.  :)
Those are the little necessities that make some RFC's hard to read ...


But this brings me to a different question about Path MTU Discovery (not
SCTP-related):

How is it implemented in real life?
RFC 1191 only makes suggestions about the layering, where PMTU
information is cached etc. - does anybody know the most common way to
implement it? Just layer 3? Initiated by TCP?

Cheers,
Michael


_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Tue Jan 15 07:34:21 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id HAA20057
	for <tsvwg-archive@odin.ietf.org>; Tue, 15 Jan 2002 07:34:21 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id HAA03458
	for tsvwg-archive@odin.ietf.org; Tue, 15 Jan 2002 07:34:22 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id HAA02796;
	Tue, 15 Jan 2002 07:20:35 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id HAA02767
	for <tsvwg@optimus.ietf.org>; Tue, 15 Jan 2002 07:20:33 -0500 (EST)
Received: from mgw-x3.nokia.com (mgw-x3.nokia.com [131.228.20.26])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id HAA19612
	for <tsvwg@ietf.org>; Tue, 15 Jan 2002 07:20:30 -0500 (EST)
Received: from esvir01nok.ntc.nokia.com (esvir01nokt.ntc.nokia.com [172.21.143.33])
	by mgw-x3.nokia.com (Switch-2.1.0/Switch-2.1.0) with ESMTP id g0FCKak23936
	for <tsvwg@ietf.org>; Tue, 15 Jan 2002 14:20:36 +0200 (EET)
Received: from esebh002.NOE.Nokia.com (unverified) by esvir01nok.ntc.nokia.com
 (Content Technologies SMTPRS 4.2.5) with ESMTP id <T5876edfcf9ac158f21083@esvir01nok.ntc.nokia.com>;
 Tue, 15 Jan 2002 14:20:30 +0200
Received: from esebe017.NOE.Nokia.com ([172.21.138.56]) by esebh002.NOE.Nokia.com with Microsoft SMTPSVC(5.0.2195.2966);
	 Tue, 15 Jan 2002 14:20:30 +0200
x-mimeole: Produced By Microsoft Exchange V6.0.4712.0
content-class: urn:content-classes:message
Subject: RE: [Tsvwg] fragmentation and PMTU
MIME-Version: 1.0
Content-Type: text/plain;
	charset="iso-8859-1"
Date: Tue, 15 Jan 2002 14:20:29 +0200
Message-ID: <58AEDE27D48F884285007ABF827FF26308EE2B@esebe017.NOE.Nokia.com>
Thread-Topic: [Tsvwg] fragmentation and PMTU
Thread-Index: AcGdPp91gk7ajwkvEdaJYAAIx6TWpQAfmnuw
From: "Arias-Rodriguez Ivan (NRC/Helsinki)" <Ivan.Arias-Rodriguez@nokia.com>
To: "'ext Randall Stewart'" <randall@stewart.chicago.il.us>
Cc: "TSV WG list" <tsvwg@ietf.org>
X-OriginalArrivalTime: 15 Jan 2002 12:20:30.0242 (UTC) FILETIME=[05A9E820:01C19DBF]
Content-Transfer-Encoding: 8bit
X-MIME-Autoconverted: from quoted-printable to 8bit by optimus.ietf.org id HAA02768
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 8bit

	Hi Randall!

	Some comments inline...

> Dear All:
> 
> It has come to my attention that the following
> is currently in RFC2960:
> 
> "
> 6.9 Fragmentation and Reassembly
> 
>    An endpoint MAY support fragmentation when sending DATA chunks, but
>    MUST support reassembly when receiving DATA chunks.  If an endpoint
>    supports fragmentation, it MUST fragment a user message if the size
>    of the user message to be sent causes the outbound SCTP packet size
>    to exceed the current MTU.  If an implementation does not support
>    fragmentation of outbound user messages, the endpoint must 
> return an
>    error to its upper layer and not attempt to send the user message.
> "
> 
> This all seems fine until one looks carefully at one little aspect:
> 
> - If I an implementation supports fragmentation it MUST fragment.
> 
> The issue I have with this is what if an application layer says:
> 
> "hey I know you can fragment, but please don't, I want you
>  to reject sends that exceed the P-MTU."

	Excuse my ignorance, but, why would you like such kind of
behavior? Couldn't the upper user simply discard those messages
beforehand? The upper user should be aware of the variations in the
P-MTU but... :-/

	Now that I think about it... About the socket extension...
Wouldn't it be a good idea to include a spinfo_pmtu parameter in the
struct sctp_paddrinfo (8.7.2)? Otherwise, AFAIK there is no way to know
about the MTU at all...

	Maybe the struct sctp_status (7.2.1) could also contain a
sstat_mtu field... This would be the lowest of all the P-MTUs... If you
are sending messages bigger than that value, you know they will be
fragmented (because you can never know to which address they will be
sent/retransmitted)...

> This may be a perfectly reasonable behavior but the wording above
> dis-allows that... I know the intent was to prevent folks from
> just setting the DF bit and NOT doing P-MTU discovery... but what
> we probably need here is a little clearification for the implementors
> guide... something like.
> 
> "
>  Note: If an implementation that supports fragmentation makes
>  available to its upper layer a mechanism to turn off fragmentation
>  it may do so. However in so doing, it MUST react just like an
>  implementation that does NOT support fragmentation i.e. it MUST
>  reject sends that exceed the current P-MTU.
> "
> 
> I think this note (+wordsmitthing :>) would allow the behavior 
> I mention above without harming the intent of the P-MTU/Fragmentation.
> 
> (Word smithing suggestions would be welcome here :>)
> 
> What do you'all think?

	Well, I think the Note doesn't damage to anybody. I'm just
curious why somebody would like to behave like that... (uhm, I think I
have made to myself this same question in some other situations... :->)

	BR Iv�n Arias Rodr�guez

> Thanks
> 
> R
> 
> -- 
> Randall R. Stewart
> randall@stewart.chicago.il.us 815-342-5222 (cell phone)
> 
> _______________________________________________
> tsvwg mailing list
> tsvwg@ietf.org
> http://www1.ietf.org/mailman/listinfo/tsvwg
> 

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Tue Jan 15 07:59:22 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id HAA20866
	for <tsvwg-archive@odin.ietf.org>; Tue, 15 Jan 2002 07:59:22 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id HAA03903
	for tsvwg-archive@odin.ietf.org; Tue, 15 Jan 2002 07:59:24 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id HAA03679;
	Tue, 15 Jan 2002 07:44:51 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id HAA03650
	for <tsvwg@optimus.ietf.org>; Tue, 15 Jan 2002 07:44:49 -0500 (EST)
Received: from sj-msg-core-3.cisco.com (sj-msg-core-3.cisco.com [171.70.157.152])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id HAA20432
	for <tsvwg@ietf.org>; Tue, 15 Jan 2002 07:44:47 -0500 (EST)
Received: from mira-sjc5-2.cisco.com (mira-sjc5-2.cisco.com [171.71.163.16])
	by sj-msg-core-3.cisco.com (8.11.3/8.9.1) with ESMTP id g0FChpk15811;
	Tue, 15 Jan 2002 04:43:51 -0800 (PST)
Received: from stewart.chicago.il.us (ssh-sj1.cisco.com [171.68.225.134])
	by mira-sjc5-2.cisco.com (Mirapoint)
	with ESMTP id AAT42152;
	Tue, 15 Jan 2002 04:44:16 -0800 (PST)
Message-ID: <3C44241F.2F410B53@stewart.chicago.il.us>
Date: Tue, 15 Jan 2002 06:44:15 -0600
From: Randall Stewart <randall@stewart.chicago.il.us>
X-Mailer: Mozilla 4.76 [en] (X11; U; Linux 2.2.12 i386)
X-Accept-Language: en
MIME-Version: 1.0
To: Michael Welzl <michael.welzl@uibk.ac.at>
CC: TSV WG list <tsvwg@ietf.org>
Subject: Re: [Tsvwg] fragmentation and PMTU
References: <3C4346DC.D5F3301B@stewart.chicago.il.us> <1011094456.2381.4.camel@c703-pc55>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Transfer-Encoding: 7bit
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit

Michael Welzl wrote:
> But this brings me to a different question about Path MTU Discovery (not
> SCTP-related):
> 
> How is it implemented in real life?
> RFC 1191 only makes suggestions about the layering, where PMTU
> information is cached etc. - does anybody know the most common way to
> implement it? Just layer 3? Initiated by TCP?

Well lets see.. For TCP it is simpler than SCTP but
here is a rough run down on what happens...

When the ICMP message arrives a callback is generated 
to the specific protocol that launched the message that 
got the MSG_SIZE error out on the net. 

On the TCP side (found in tcp_subr on FreeBSD) the callback
hits tcp_ctlinput() with a PRC_MSGSIZE. The connection 
is found (if it can be) and then the tcp_mtudisc() routine is
called. This routine updates the mss and then looks to see
if it needs to do any output i.e. finds the data that was
just dropped and trys to resend it.. and since it is
a byte stream that is much easier than what SCTP has
to do :-0 (You can get better details than my quick
pass through the code .. with little coffee I might add ..
from Stevens TCP Illustrated Vol 2).

On the SCTP side we find the association, validate that
the V-Tag is correct. Then if all is good we proceed and
update the fragmentation threshold (PMTU variable kept). Once
we finish this we then walk through all the outstanding 
data chunks. If one exceeds the P-MTU (which may or may
not be the case) we mark it for resend AND set a flag that
will allow the IP_DF bit to be cleared (only V4 here). Then
we call the output which hopefully sends off our way-ward
packet. Note I said we may not find anything larger than
the PMTU in our chunk search.. this is because it may
have been a bundle of multiple chunks that exceeded the
PMTU size. For V6 if we are really smart we can try to
walk through the packet that was rejected and find the
specific chunks to mark for resend (my impl is too dumb
yet to do this though :0) for V4 we can't because we
only get (for sure) 8 bytes.. just enough to give us
the V-Tag and ports... so we may end up doing a t-o on
the chunk :-0

That more or less covers it.. for the SCTP details I
do cover some of it in the SCTP book ( the documentation
of the user space implementation in chapter 14 and
overview in 6.5) but to get a better view of a kernel implementation
you will probably have to wait to look at the 
new kernel code Peter and I are cranking on :0

We are almost done by the way .. :))



R


-- 
Randall R. Stewart
randall@stewart.chicago.il.us 815-342-5222 (cell phone)

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Tue Jan 15 08:50:39 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id IAA23302
	for <tsvwg-archive@odin.ietf.org>; Tue, 15 Jan 2002 08:50:39 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id IAA05662
	for tsvwg-archive@odin.ietf.org; Tue, 15 Jan 2002 08:50:42 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id IAA05256;
	Tue, 15 Jan 2002 08:33:39 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id IAA05222
	for <tsvwg@optimus.ietf.org>; Tue, 15 Jan 2002 08:33:37 -0500 (EST)
Received: from hindon.hss.co.in ([202.54.26.202])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id IAA22247
	for <tsvwg@ietf.org>; Tue, 15 Jan 2002 08:33:33 -0500 (EST)
From: samahajan@hss.hns.com
Received: from sandesh.hss.hns.com (sandesh [139.85.242.35])
	by hindon.hss.co.in (8.10.0/8.10.0) with SMTP id g0FDWTE14939;
	Tue, 15 Jan 2002 19:02:29 +0530 (IST)
Received: by sandesh.hss.hns.com(Lotus SMTP MTA v4.6.3  (733.2 10-16-1998))  id 65256B42.004A3D7D ; Tue, 15 Jan 2002 19:00:54 +0530
X-Lotus-FromDomain: HSS
To: Kacheong Poon <kcpoon@eng.sun.com>
cc: Randall Stewart <randall@stewart.chicago.il.us>, ygahlaut@hss.hns.com,
        tsvwg@ietf.org
Message-ID: <65256B42.004A3B96.00@sandesh.hss.hns.com>
Date: Tue, 15 Jan 2002 19:00:48 +0530
Subject: Re: [Tsvwg] Section 7.1.3 sctpsocket-02.txt
Mime-Version: 1.0
Content-type: text/plain; charset=us-ascii
Content-Disposition: inline
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org




Hi,

Comments below ..

Best Regards,
-Sandeep





Kacheong Poon <kcpoon@eng.sun.com> on 01/15/2002 09:00:39 AM

Please respond to Kacheong Poon <kcpoon@eng.sun.com>

To:   Randall Stewart <randall@stewart.chicago.il.us>
cc:   Yogesh Gahlaut/HSS@HSS, tsvwg@ietf.org (bcc: Sandeep Mahajan/HSS)

Subject:  Re: [Tsvwg] Section 7.1.3 sctpsocket-02.txt




> No I think you need to do it AFTER the accept call. This is because
> you really don't want the assoc-change event to show up
> on the listen() socket.. but the new one. Thus the socket does
> not exist until after the accept() which creates the fd.
>
> But maybe it should be implementation dependant... I would defintly
> welcome other comments/opinions on this..

First, I think apps using TCP-style socket will probably not
enable this notification.  What Randy suggested make sense.  But
then if we want to make it similar to UDP-style socket, the
notification should be sent to the listening socket.  After
getting it, the app can do an accept(). I don't have preference
of one to the other.  It is probably safe to leave it
as implementation dependent for the moment, until we see a strong
reason to do it one way or the other.

Sandeep >> Third option would be to not send Endpoint UP notification at all at
Server side. Application waiting for events on server sockets ( either via
blocking accept OR via select) only needs to do an accept call to get the new
socket fd. To get the parameters of new association like peer address list and
number of negotiated streams (which would have returned to the application in
Endpoint UP notification), application can use API "sctp_getpaddrs" and
getsockopt with option "SCTP_STATUS". Provided numbe of negotiated streams
parameter is added in the sctp_status structure.



                                   K. Poon.
                                   kcpoon@eng.sun.com



_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg







_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Tue Jan 15 09:32:40 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id JAA24634
	for <tsvwg-archive@odin.ietf.org>; Tue, 15 Jan 2002 09:32:35 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id JAA07664
	for tsvwg-archive@odin.ietf.org; Tue, 15 Jan 2002 09:32:38 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id JAA06866;
	Tue, 15 Jan 2002 09:17:12 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id JAA06835
	for <tsvwg@optimus.ietf.org>; Tue, 15 Jan 2002 09:17:10 -0500 (EST)
Received: from sj-msg-core-3.cisco.com (sj-msg-core-3.cisco.com [171.70.157.152])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id JAA24117
	for <tsvwg@ietf.org>; Tue, 15 Jan 2002 09:17:06 -0500 (EST)
Received: from mira-sjc5-2.cisco.com (mira-sjc5-2.cisco.com [171.71.163.16])
	by sj-msg-core-3.cisco.com (8.11.3/8.9.1) with ESMTP id g0FEGBk16522;
	Tue, 15 Jan 2002 06:16:11 -0800 (PST)
Received: from stewart.chicago.il.us (ssh-sj1.cisco.com [171.68.225.134])
	by mira-sjc5-2.cisco.com (Mirapoint)
	with ESMTP id AAT42832;
	Tue, 15 Jan 2002 06:16:35 -0800 (PST)
Message-ID: <3C4439C2.5BA4291B@stewart.chicago.il.us>
Date: Tue, 15 Jan 2002 08:16:34 -0600
From: Randall Stewart <randall@stewart.chicago.il.us>
X-Mailer: Mozilla 4.76 [en] (X11; U; Linux 2.2.12 i386)
X-Accept-Language: en
MIME-Version: 1.0
To: "Arias-Rodriguez Ivan (NRC/Helsinki)" <Ivan.Arias-Rodriguez@nokia.com>
CC: TSV WG list <tsvwg@ietf.org>
Subject: Re: [Tsvwg] fragmentation and PMTU
References: <58AEDE27D48F884285007ABF827FF26308EE2B@esebe017.NOE.Nokia.com>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Transfer-Encoding: 7bit
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit

"Arias-Rodriguez Ivan (NRC/Helsinki)" wrote:
> 
>         Hi Randall!
> 
>         Some comments inline...
> 
> > Dear All:
> >
> > It has come to my attention that the following
> > is currently in RFC2960:
> >
> > "
> > 6.9 Fragmentation and Reassembly
> >
> >    An endpoint MAY support fragmentation when sending DATA chunks, but
> >    MUST support reassembly when receiving DATA chunks.  If an endpoint
> >    supports fragmentation, it MUST fragment a user message if the size
> >    of the user message to be sent causes the outbound SCTP packet size
> >    to exceed the current MTU.  If an implementation does not support
> >    fragmentation of outbound user messages, the endpoint must
> > return an
> >    error to its upper layer and not attempt to send the user message.
> > "
> >
> > This all seems fine until one looks carefully at one little aspect:
> >
> > - If I an implementation supports fragmentation it MUST fragment.
> >
> > The issue I have with this is what if an application layer says:
> >
> > "hey I know you can fragment, but please don't, I want you
> >  to reject sends that exceed the P-MTU."
> 
>         Excuse my ignorance, but, why would you like such kind of
> behavior? Couldn't the upper user simply discard those messages
> beforehand? The upper user should be aware of the variations in the
> P-MTU but... :-/


One place such a thing MIGHT be wanted is in something like
SCSI over IP... there are other applications as well that
may NOT want to ever send anything over a PMTU. 

We may not be able to come up with more than one or so
applications that do this now.. but why exclude a behavior
that some future app's may also want for no good reason...


> 
>         Now that I think about it... About the socket extension...
> Wouldn't it be a good idea to include a spinfo_pmtu parameter in the
> struct sctp_paddrinfo (8.7.2)? Otherwise, AFAIK there is no way to know
> about the MTU at all...

This is a GOOD suggestion. We could put this in the association
status field SCTP_STATUS... something like sctp_fragmentation_threshold

This may be differnt than the individual P-MTU's.. but we probably
should ALSO include this in the paddr_info as well...

If none of my co-authors have an objection I will add these in for
the next spin :>

> 
>         Maybe the struct sctp_status (7.2.1) could also contain a
> sstat_mtu field... This would be the lowest of all the P-MTUs... If you
> are sending messages bigger than that value, you know they will be
> fragmented (because you can never know to which address they will be
> sent/retransmitted)...

Yep, fragmentation threshold... most implementations use the
lowest MTU of all destinations for this... 

> 
> > This may be a perfectly reasonable behavior but the wording above
> > dis-allows that... I know the intent was to prevent folks from
> > just setting the DF bit and NOT doing P-MTU discovery... but what
> > we probably need here is a little clearification for the implementors
> > guide... something like.
> >
> > "
> >  Note: If an implementation that supports fragmentation makes
> >  available to its upper layer a mechanism to turn off fragmentation
> >  it may do so. However in so doing, it MUST react just like an
> >  implementation that does NOT support fragmentation i.e. it MUST
> >  reject sends that exceed the current P-MTU.
> > "
> >
> > I think this note (+wordsmitthing :>) would allow the behavior
> > I mention above without harming the intent of the P-MTU/Fragmentation.
> >
> > (Word smithing suggestions would be welcome here :>)
> >
> > What do you'all think?
> 
>         Well, I think the Note doesn't damage to anybody. I'm just
> curious why somebody would like to behave like that... (uhm, I think I
> have made to myself this same question in some other situations... :->)


I believe also Doug Otis posted something along this line
as well for RDMA... not sure of the details.. I will have
to go back and look in the email archives...

R

-- 
Randall R. Stewart
randall@stewart.chicago.il.us 815-342-5222 (cell phone)

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Tue Jan 15 09:58:34 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id JAA25835
	for <tsvwg-archive@odin.ietf.org>; Tue, 15 Jan 2002 09:58:34 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id JAA09193
	for tsvwg-archive@odin.ietf.org; Tue, 15 Jan 2002 09:58:35 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id JAA08403;
	Tue, 15 Jan 2002 09:42:44 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id JAA08372
	for <tsvwg@optimus.ietf.org>; Tue, 15 Jan 2002 09:42:42 -0500 (EST)
Received: from mxic1.isus.emc.com ([168.159.129.100])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id JAA25095
	for <tsvwg@ietf.org>; Tue, 15 Jan 2002 09:42:38 -0500 (EST)
From: Black_David@emc.com
Received: by mxic1.isus.emc.com with Internet Mail Service (5.5.2653.19)
	id <ZPC7SVMH>; Tue, 15 Jan 2002 09:43:06 -0500
Message-ID: <277DD60FB639D511AC0400B0D068B71E0293736D@CORPMX14>
To: randall@stewart.chicago.il.us
Cc: tsvwg@ietf.org
Subject: RE: [Tsvwg] fragmentation and PMTU
Date: Tue, 15 Jan 2002 09:27:37 -0500
MIME-Version: 1.0
X-Mailer: Internet Mail Service (5.5.2653.19)
Content-Type: text/plain;
	charset="iso-8859-1"
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

> > > This all seems fine until one looks carefully at one little aspect:
> > >
> > > - If I an implementation supports fragmentation it MUST fragment.
> > >
> > > The issue I have with this is what if an application layer says:
> > >
> > > "hey I know you can fragment, but please don't, I want you
> > >  to reject sends that exceed the P-MTU."
> > 
> >         Excuse my ignorance, but, why would you like such kind of
> > behavior? Couldn't the upper user simply discard those messages
> > beforehand? The upper user should be aware of the variations in the
> > P-MTU but... :-/
> 
> 
> One place such a thing MIGHT be wanted is in something like
> SCSI over IP... there are other applications as well that
> may NOT want to ever send anything over a PMTU.

Not exactly.  On its own, SCSI has little interest in PMTU - SCSI
tends to transfer data in units of 2k, 4k, 16k, etc., driven by
what's convenient for the systems it connects.  IMHO, forcing a
SCSI implementation to fragment to PMTU is not a real good idea.
I believe the actual motivation here is the desire to define a
Direct Data Placement (DDP) layer between SCTP and a protocol
like SCSI over IP - one of the ways of doing this entails adding
information to every chunk and hence wants to ensure that SCTP
never fragments so that DDP can do its own fragmentation and
add its info to every fragment.

> I believe also Doug Otis posted something along this line
> as well for RDMA... not sure of the details.. I will have
> to go back and look in the email archives...

DDP is an important enabling technology for RDMA.  There's
a lot more about this on the rdma@yahoogroups.com list.

Thanks,
--David

---------------------------------------------------
David L. Black, Senior Technologist
EMC Corporation, 42 South St., Hopkinton, MA  01748
+1 (508) 249-6449 *NEW*      FAX: +1 (508) 497-8500
black_david@emc.com         Cell: +1 (978) 394-7754
---------------------------------------------------

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Tue Jan 15 10:12:47 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id KAA28396
	for <tsvwg-archive@odin.ietf.org>; Tue, 15 Jan 2002 10:12:47 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id KAA09966
	for tsvwg-archive@odin.ietf.org; Tue, 15 Jan 2002 10:12:48 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id JAA08985;
	Tue, 15 Jan 2002 09:52:42 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id JAA08909
	for <tsvwg@optimus.ietf.org>; Tue, 15 Jan 2002 09:52:38 -0500 (EST)
Received: from sj-msg-core-4.cisco.com (sj-msg-core-4.cisco.com [171.71.163.10])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id JAA25554
	for <tsvwg@ietf.org>; Tue, 15 Jan 2002 09:52:36 -0500 (EST)
Received: from mira-sjc5-2.cisco.com (mira-sjc5-2.cisco.com [171.71.163.16])
	by sj-msg-core-4.cisco.com (8.11.3/8.9.1) with ESMTP id g0FEq7J06792;
	Tue, 15 Jan 2002 06:52:07 -0800 (PST)
Received: from stewart.chicago.il.us (ssh-sj1.cisco.com [171.68.225.134])
	by mira-sjc5-2.cisco.com (Mirapoint)
	with ESMTP id AAT43227;
	Tue, 15 Jan 2002 06:52:06 -0800 (PST)
Message-ID: <3C444215.CD62B0D5@stewart.chicago.il.us>
Date: Tue, 15 Jan 2002 08:52:05 -0600
From: Randall Stewart <randall@stewart.chicago.il.us>
X-Mailer: Mozilla 4.76 [en] (X11; U; Linux 2.2.12 i386)
X-Accept-Language: en
MIME-Version: 1.0
To: Black_David@emc.com
CC: tsvwg@ietf.org
Subject: Re: [Tsvwg] fragmentation and PMTU
References: <277DD60FB639D511AC0400B0D068B71E0293736D@CORPMX14>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Transfer-Encoding: 7bit
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit

David:

Thanks for the correction.. I knew it had something
to do with IPS and RDMA... sorry about getting them
mixed up :>

R

Black_David@emc.com wrote:
> 
> > > > This all seems fine until one looks carefully at one little aspect:
> > > >
> > > > - If I an implementation supports fragmentation it MUST fragment.
> > > >
> > > > The issue I have with this is what if an application layer says:
> > > >
> > > > "hey I know you can fragment, but please don't, I want you
> > > >  to reject sends that exceed the P-MTU."
> > >
> > >         Excuse my ignorance, but, why would you like such kind of
> > > behavior? Couldn't the upper user simply discard those messages
> > > beforehand? The upper user should be aware of the variations in the
> > > P-MTU but... :-/
> >
> >
> > One place such a thing MIGHT be wanted is in something like
> > SCSI over IP... there are other applications as well that
> > may NOT want to ever send anything over a PMTU.
> 
> Not exactly.  On its own, SCSI has little interest in PMTU - SCSI
> tends to transfer data in units of 2k, 4k, 16k, etc., driven by
> what's convenient for the systems it connects.  IMHO, forcing a
> SCSI implementation to fragment to PMTU is not a real good idea.
> I believe the actual motivation here is the desire to define a
> Direct Data Placement (DDP) layer between SCTP and a protocol
> like SCSI over IP - one of the ways of doing this entails adding
> information to every chunk and hence wants to ensure that SCTP
> never fragments so that DDP can do its own fragmentation and
> add its info to every fragment.
> 
> > I believe also Doug Otis posted something along this line
> > as well for RDMA... not sure of the details.. I will have
> > to go back and look in the email archives...
> 
> DDP is an important enabling technology for RDMA.  There's
> a lot more about this on the rdma@yahoogroups.com list.
> 
> Thanks,
> --David
> 
> ---------------------------------------------------
> David L. Black, Senior Technologist
> EMC Corporation, 42 South St., Hopkinton, MA  01748
> +1 (508) 249-6449 *NEW*      FAX: +1 (508) 497-8500
> black_david@emc.com         Cell: +1 (978) 394-7754
> ---------------------------------------------------

-- 
Randall R. Stewart
randall@stewart.chicago.il.us 815-342-5222 (cell phone)

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Tue Jan 15 11:39:06 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id LAA02197
	for <tsvwg-archive@odin.ietf.org>; Tue, 15 Jan 2002 11:39:06 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id LAA12835
	for tsvwg-archive@odin.ietf.org; Tue, 15 Jan 2002 11:39:08 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id LAA12067;
	Tue, 15 Jan 2002 11:12:30 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id LAA12038
	for <tsvwg@optimus.ietf.org>; Tue, 15 Jan 2002 11:12:28 -0500 (EST)
Received: from c003.snv.cp.net (c003-h004.c003.snv.cp.net [209.228.32.218])
	by ietf.org (8.9.1a/8.9.1a) with SMTP id LAA00689
	for <tsvwg@ietf.org>; Tue, 15 Jan 2002 11:12:25 -0500 (EST)
Received: (cpmta 25456 invoked from network); 15 Jan 2002 08:11:57 -0800
Received: from 64.130.130.105 (HELO dougrmt)
  by smtp.telocity.com (209.228.32.218) with SMTP; 15 Jan 2002 08:11:57 -0800
X-Sent: 15 Jan 2002 16:11:57 GMT
From: "Douglas Otis" <dotis@sanlight.net>
To: "Tsvwg" <tsvwg@ietf.org>
Subject: RE: [Tsvwg] fragmentation and PMTU
Date: Tue, 15 Jan 2002 08:12:30 -0800
Message-ID: <NEBBJGDMMLHHCIKHGBEJKENKCOAA.dotis@sanlight.net>
MIME-Version: 1.0
Content-Type: text/plain;
	charset="iso-8859-1"
Content-Transfer-Encoding: 7bit
X-Priority: 3 (Normal)
X-MSMail-Priority: Normal
X-Mailer: Microsoft Outlook IMO, Build 9.0.2416 (9.0.2910.0)
In-Reply-To: <277DD60FB639D511AC0400B0D068B71E0293736D@CORPMX14>
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4807.1700
Importance: Normal
Content-Transfer-Encoding: 7bit
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit

David,

Allowing the Direct Data Placement shim to sit above SCTP needs two features
from the SCTP API.  One is the ability to do the chunk fragmentation.  This
allows automation of reassembly by the shim through insertion of an offset
and is independent of the message size being sent by the ULP.  This also has
advantages with respect to partial writes to a Stream, in that there is no
sequential TSN requirement across a message.  This feature has been
accommodated in the latest Socket draft -03, and should be included in the
Implementers guide.

The second requirement is equally important.  As these fragmented chunks are
sent Unordered to ensure no buffering is done by SCTP, completion signaling
then requires that the Transmission Sequence Number as well as the
Cumulative Transmission Sequence Number be visible.  As messages are
fragmented and reassembled by the shim as received, there is no need for the
Stream Sequence Number (it is invalid for Unordered delivery) and its 16 bit
size *may* represent a potential limitation at very high bandwidths in the
future.  A message is signaled complete to the ULP when the last chunk in a
message containing a marker is received and its TSN is equal to or less than
the cumulative TSN per serial math.  This marker signal is held at the shim
layer until the TSN condition is met.

The shim concept is to allow Direct Data Placement (DDP) to be added as a
layer above SCTP.  This differs substantially from framing techniques used
on TCP, such that no packet filter is inserted below the transport to
intercept packets.  SCTP remains unaltered and its path management,
bundling, retry algorithms, etc. remain intact and unaltered.

Doug

> > > > This all seems fine until one looks carefully at one little aspect:
> > > >
> > > > - If I an implementation supports fragmentation it MUST fragment.
> > > >
> > > > The issue I have with this is what if an application layer says:
> > > >
> > > > "hey I know you can fragment, but please don't, I want you
> > > >  to reject sends that exceed the P-MTU."
> > >
> > >         Excuse my ignorance, but, why would you like such kind of
> > > behavior? Couldn't the upper user simply discard those messages
> > > beforehand? The upper user should be aware of the variations in the
> > > P-MTU but... :-/
> >
> > One place such a thing MIGHT be wanted is in something like
> > SCSI over IP... there are other applications as well that
> > may NOT want to ever send anything over a PMTU.
>
> Not exactly.  On its own, SCSI has little interest in PMTU - SCSI
> tends to transfer data in units of 2k, 4k, 16k, etc., driven by
> what's convenient for the systems it connects.  IMHO, forcing a
> SCSI implementation to fragment to PMTU is not a real good idea.
> I believe the actual motivation here is the desire to define a
> Direct Data Placement (DDP) layer between SCTP and a protocol
> like SCSI over IP - one of the ways of doing this entails adding
> information to every chunk and hence wants to ensure that SCTP
> never fragments so that DDP can do its own fragmentation and
> add its info to every fragment.
>
> > I believe also Doug Otis posted something along this line
> > as well for RDMA... not sure of the details.. I will have
> > to go back and look in the email archives...
>
> DDP is an important enabling technology for RDMA.  There's
> a lot more about this on the rdma@yahoogroups.com list.
>
> Thanks,
> --David
>
> ---------------------------------------------------
> David L. Black, Senior Technologist
> EMC Corporation, 42 South St., Hopkinton, MA  01748
> +1 (508) 249-6449 *NEW*      FAX: +1 (508) 497-8500
> black_david@emc.com         Cell: +1 (978) 394-7754
> ---------------------------------------------------


_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Tue Jan 15 12:09:33 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id MAA05495
	for <tsvwg-archive@odin.ietf.org>; Tue, 15 Jan 2002 12:09:33 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id MAA14283
	for tsvwg-archive@odin.ietf.org; Tue, 15 Jan 2002 12:09:27 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id LAA12996;
	Tue, 15 Jan 2002 11:45:36 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id LAA12966
	for <tsvwg@optimus.ietf.org>; Tue, 15 Jan 2002 11:45:34 -0500 (EST)
Received: from sandmail.sandburst.com (sandmail.sandburst.com [216.57.132.42])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id LAA04170
	for <tsvwg@ietf.org>; Tue, 15 Jan 2002 11:45:31 -0500 (EST)
To: tsvwg@ietf.org
Subject: Re: [Tsvwg] fragmentation and PMTU 
In-Reply-To: Message from Randall Stewart <randall@stewart.chicago.il.us> 
   of "Mon, 14 Jan 2002 15:00:12 CST." <3C4346DC.D5F3301B@stewart.chicago.il.us> 
References: <3C4346DC.D5F3301B@stewart.chicago.il.us> 
Date: Tue, 15 Jan 2002 11:44:53 -0500
From: Stephen Bailey <steph@cs.uchicago.edu>
Message-Id: <20020115164503.2B5565113@sandmail.sandburst.com>
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

> It has come to my attention that the following
> is currently in RFC2960:
>
> [...]
> 
> What do you'all think?

No surprise, I'm for it (I've been agitating for this change for a
while).

It seemed like just an accident of wording that implementations that
CAN'T do fragmentation offer the ULP a guarantee that messages will
always be in a single chunk, where implementations that CAN do
fragmentation don't.  Personally, I'm going to run out and buy one of
those cool, unfragmenting implementations right now before they remove
them from the market :^)

As David points out, one reason for this change is to allow efficient
support of DDP in SCTP.  Another might be to permit simple processing
of unordered messages, perhaps for customized implementations of
particular ULP/SCTP pairs (maybe some telephony systems might be in
this category?).

Steph


_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Tue Jan 15 13:26:14 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id NAA08324
	for <tsvwg-archive@odin.ietf.org>; Tue, 15 Jan 2002 13:26:14 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id NAA17028
	for tsvwg-archive@odin.ietf.org; Tue, 15 Jan 2002 13:26:15 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id NAA16340;
	Tue, 15 Jan 2002 13:05:00 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id NAA16307
	for <tsvwg@optimus.ietf.org>; Tue, 15 Jan 2002 13:04:58 -0500 (EST)
Received: from localhost ([211.175.168.104])
	by ietf.org (8.9.1a/8.9.1a) with SMTP id NAA07441
	for <tsvwg@ietf.org>; Tue, 15 Jan 2002 13:04:51 -0500 (EST)
Message-Id: <200201151804.NAA07441@ietf.org>
Reply-To: office700@yahoo.co.kr
From: 창업정보<office700@yahoo.co.kr>
To: tsvwg@ietf.org
Mime-Version: 1.0
Content-Type: text/html; charset="ks_c_5601-1987"
Date: Wed, 16 Jan 2002 02:57:45 +0900
Subject: [Tsvwg] [광고] 돈되는 사업정보입니다
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

<html>
<head>
<meta http-equiv="content-type" content="text/html; charset=euc-kr">
<title> 제목 없음</title>
<meta name="GENERATOR" content="Namo WebEditor v5.0">
<meta name="description" content="스타일이 전혀 적용되지 않은 새 문서 양식을 만듭니다.">
</head>
<body>

<TABLE cellSpacing=0 cellPadding=0 width=646 bgColor=#f4f3f3 border=0>
<TBODY>
<TR>
<TD width=646 colSpan=2 height=207>
<DIV align=center>
                <img src="http://www.700office.com/images/pic4.bmp" width="600" height="201" border="0"></DIV></TD></TR>
<TR>
<TD width=645 height=9>
<P>&nbsp;</P></TD>
<TD width=1 height=9>&nbsp;</TD></TR>
<TR>
<TD width=645>
<P><img src="http://www.700office.com/images/noname6_htm_smartbutton1.gif" width="472" height="32" border="0"></P></TD>
<TD width=1>&nbsp;</TD></TR>
<TR>
<TD width=646 colSpan=2>
<TABLE cellSpacing=0 cellPadding=0 width=649 align=center border=0>
<TBODY>
<TR>
<TD height=14>
<P><IMG height=15 src="http://www.700office.com/main_01.gif" width=649></P></TD></TR>
<TR>
<TD background="http://www.700office.com/main_02.gif">
<DIV align=center>
<TABLE cellSpacing=0 cellPadding=0 width=600 border=0>
<TBODY>
<TR>
<TD vAlign=top width=13 height=22>
<P style="MARGIN-TOP: 0px; MARGIN-BOTTOM: 0px; LINE-HEIGHT: 150%"><IMG height=10 
src="http://www.700office.com/main_04.gif" width=10></P></TD>
<TD width=587 height=22>
<P style="LINE-HEIGHT: 15px"><SPAN style="FONT-SIZE: 9pt"><FONT face=돋움>가장 빠른 
광고효과입니다. 빠른 시간안에 확실한 효과를 보실수 있습니다.</FONT></SPAN> </P>
<P style="LINE-HEIGHT: 15px"><SPAN style="FONT-SIZE: 9pt"><FONT face=돋움>이보다 뛰어난 
광고효과를 기대할 수 없습니다. 저희 회원에 한하여 후불로 결재 받습니다.</FONT></SPAN></P></TD></TR>
<TR>
<TD width=13 height=22>
<P style="MARGIN-TOP: 0px; MARGIN-BOTTOM: 0px; LINE-HEIGHT: 150%">&nbsp;</P></TD>
<TD width=587 height=22>
<P style="LINE-HEIGHT: 5px"><SPAN style="FONT-SIZE: 9pt">&nbsp;</SPAN></P>
<P style="LINE-HEIGHT: 5px"><SPAN style="FONT-SIZE: 9pt">원하시는 
가격에~~~~~</SPAN></P>
<P style="LINE-HEIGHT: 5px"><SPAN style="FONT-SIZE: 9pt">원하시는 
대상에~~~~~</SPAN></P>
<P style="LINE-HEIGHT: 5px"><SPAN style="FONT-SIZE: 9pt">원하시는 
시간에~~~~~</SPAN></P>
<P style="LINE-HEIGHT: 5px"><SPAN style="FONT-SIZE: 9pt">전국의 메일광고를 대행합니다.자세한 자료는 
메일로 연락주세요...</SPAN></P>
<P style="LINE-HEIGHT: 5px"><SPAN style="FONT-SIZE: 9pt">바로 연락 
드리겠습니다.</SPAN></P>
<P style="LINE-HEIGHT: 5px">&nbsp;</P></TD></TR>
<TR>
<TD width=13 height=22>
<P style="MARGIN-TOP: 0px; MARGIN-BOTTOM: 0px; LINE-HEIGHT: 150%"><IMG height=10 
src="http://www.700office.com/main_04.gif" width=10></P></TD>
<TD width=587 height=22>
<P style="MARGIN-TOP: 0px; MARGIN-BOTTOM: 0px; LINE-HEIGHT: 150%">&nbsp;<a href="mailto:skylove0028@dreamwiz.com"><IMG height=31 alt=mail02.gif 
src="http://www.700office.com/mail02.gif" width=32 border=0></a> &nbsp; : <A title="보내주신 정성껏 답변해드리겟습니다." 
href="mailto:skylove0028@dreamwiz.com"><FONT color=#0033ff><SPAN 
style="FONT-SIZE: 9pt">궁금한 점을 마음껏 질문하세요</SPAN></FONT></A><FONT 
color=#0033ff><SPAN style="FONT-SIZE: 9pt">.</SPAN></FONT> </P></TD></TR>
<TR>
<TD width=13 height=22>
<P style="MARGIN-TOP: 0px; MARGIN-BOTTOM: 0px; LINE-HEIGHT: 150%">&nbsp;</P></TD>
<TD width=587 height=22>
<P style="MARGIN-TOP: 0px; MARGIN-BOTTOM: 0px; LINE-HEIGHT: 150%">&nbsp; </P></TD></TR>
<TR>
<TD width=13 height=22>
<P style="MARGIN-TOP: 0px; MARGIN-BOTTOM: 0px; LINE-HEIGHT: 150%"><IMG height=10 
src="http://www.700office.com/main_04.gif" width=10></P></TD>
<TD width=587 height=22>
<P style="MARGIN-TOP: 0px; MARGIN-BOTTOM: 0px; LINE-HEIGHT: 150%">&nbsp;<img src="http://www.700office.com/neww.gif" width="50" height="15" border="0"> 
                                        <font color="blue">&quot;</font><span style="font-size:9pt;"><a href="http://www.700office.com"><font face="돋움" color="blue">돈되는 
                                        창업정보&quot; &nbsp;링크해주세요 
                                        ~! </font></a></span> 
</P></TD></TR></TBODY></TABLE></DIV></TD></TR>
<TR>
<TD>
<P><IMG height=3 src="http://www.700office.com/main_03.gif" 
width=649></P></TD></TR></TBODY></TABLE></TD></TR></TBODY></TABLE>
<table border="1" cellspacing="0" width="651" bgcolor="white" bordercolordark="black" bordercolorlight="black">
    <tr>
        <td width="645"><p style="line-height:20px;" align="center"><SPAN style="FONT-SIZE: 9pt">귀하의 메일주소는 웹서핑중, <B>http://www.xxxxxxxx.com/</B> 
<BR>에서 알게 된것이며, E-Mail 주소 외에, 다른 정보는 갖고 있지 않습니다.<BR>정통부 권고사항에 의거 제목에 
</SPAN><B><SPAN style="FONT-SIZE: 9pt">[광고]</SPAN></B><SPAN style="FONT-SIZE: 9pt">라고 표기한 메일입니다. 원치 않으면 </SPAN><A 
href="mailto:joki@borahome.net?subject=수신거부" target=new><B><SPAN 
style="FONT-SIZE: 9pt">,</SPAN></B></A><A 
href="mailto:office700@yahoo.co.kr?subject=수신거부" target=new><B><SPAN 
style="FONT-SIZE: 9pt"><FONT color=red>수신거부</FONT></SPAN></B></A><SPAN style="FONT-SIZE: 9pt">를 눌러주세요</SPAN> 
</p>
        </td>
    </tr>
</table>
</body>

</html>

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Tue Jan 15 19:55:59 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id TAA17359
	for <tsvwg-archive@odin.ietf.org>; Tue, 15 Jan 2002 19:55:55 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id TAA29298
	for tsvwg-archive@odin.ietf.org; Tue, 15 Jan 2002 19:55:56 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id TAA28287;
	Tue, 15 Jan 2002 19:29:18 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id TAA28255
	for <tsvwg@optimus.ietf.org>; Tue, 15 Jan 2002 19:29:15 -0500 (EST)
Received: from mercury.Sun.COM (mercury.Sun.COM [192.9.25.1])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id TAA17031
	for <tsvwg@ietf.org>; Tue, 15 Jan 2002 19:29:13 -0500 (EST)
Received: from engmail1.Eng.Sun.COM ([129.146.1.13])
	by mercury.Sun.COM (8.9.3+Sun/8.9.3) with ESMTP id QAA29617;
	Tue, 15 Jan 2002 16:29:09 -0800 (PST)
Received: from shield (shield.EBay.Sun.COM [129.150.113.84])
	by engmail1.Eng.Sun.COM (8.9.3+Sun/8.9.3/ENSMAIL,v2.1p1) with SMTP id QAA20625;
	Tue, 15 Jan 2002 16:29:07 -0800 (PST)
Date: Tue, 15 Jan 2002 16:28:24 -0800 (PST)
From: Kacheong Poon <kcpoon@eng.sun.com>
Reply-To: Kacheong Poon <kcpoon@eng.sun.com>
Subject: RE: [Tsvwg] fragmentation and PMTU
To: Douglas Otis <dotis@sanlight.net>
Cc: Tsvwg <tsvwg@ietf.org>
In-Reply-To: "Your message with ID" <NEBBJGDMMLHHCIKHGBEJKENKCOAA.dotis@sanlight.net>
Message-ID: <Roam.SIMC.2.0.6.1011140904.31762.kcpoon@eng.sun.com>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; CHARSET=US-ASCII
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

> The second requirement is equally important.  As these fragmented chunks are
> sent Unordered to ensure no buffering is done by SCTP, completion signaling
> then requires that the Transmission Sequence Number as well as the
> Cumulative Transmission Sequence Number be visible.  As messages are
> fragmented and reassembled by the shim as received, there is no need for the
> Stream Sequence Number (it is invalid for Unordered delivery) and its 16 bit
> size *may* represent a potential limitation at very high bandwidths in the
> future.  A message is signaled complete to the ULP when the last chunk in a
> message containing a marker is received and its TSN is equal to or less than
> the cumulative TSN per serial math.  This marker signal is held at the shim
> layer until the TSN condition is met.

Can you explain why the shim cannot put in its own sequence number and
the stack has to expose the internal TSN to the ULP?

							K. Poon.
							kcpoon@eng.sun.com



_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Tue Jan 15 20:51:46 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id UAA17966
	for <tsvwg-archive@odin.ietf.org>; Tue, 15 Jan 2002 20:51:46 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id UAA01452
	for tsvwg-archive@odin.ietf.org; Tue, 15 Jan 2002 20:51:48 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id UAA00560;
	Tue, 15 Jan 2002 20:23:17 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id UAA00532
	for <tsvwg@optimus.ietf.org>; Tue, 15 Jan 2002 20:23:16 -0500 (EST)
Received: from c003.snv.cp.net (c003-h004.c003.snv.cp.net [209.228.32.218])
	by ietf.org (8.9.1a/8.9.1a) with SMTP id UAA17684
	for <tsvwg@ietf.org>; Tue, 15 Jan 2002 20:23:13 -0500 (EST)
Received: (cpmta 20218 invoked from network); 15 Jan 2002 17:22:44 -0800
Received: from 64.130.130.105 (HELO dougrmt)
  by smtp.telocity.com (209.228.32.218) with SMTP; 15 Jan 2002 17:22:44 -0800
X-Sent: 16 Jan 2002 01:22:44 GMT
From: "Douglas Otis" <dotis@sanlight.net>
Cc: "Tsvwg" <tsvwg@ietf.org>
Subject: RE: [Tsvwg] fragmentation and PMTU
Date: Tue, 15 Jan 2002 17:23:18 -0800
Message-ID: <NEBBJGDMMLHHCIKHGBEJCEOGCOAA.dotis@sanlight.net>
MIME-Version: 1.0
Content-Type: text/plain;
	charset="us-ascii"
Content-Transfer-Encoding: 7bit
X-Priority: 3 (Normal)
X-MSMail-Priority: Normal
X-Mailer: Microsoft Outlook IMO, Build 9.0.2416 (9.0.2910.0)
In-reply-to: <Roam.SIMC.2.0.6.1011140904.31762.kcpoon@eng.sun.com>
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4807.1700
Importance: Normal
Content-Transfer-Encoding: 7bit
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit

Kacheong,

I understand a desire to hide internals of SCTP from the ULP.  In the case
of TSN, the transport is doing a significant amount of work providing a
cumulative TSN as well as tracking the TSN on each chunk.  As these are
structures already available within the transport and on the wire and, at
one time were exposed, these values available to the ULP will remove
overhead both with respect to size on a per chunk basis as well as the score
boarding otherwise required by the shim.  As you know, the Stream Sequence
Number is of no use for Unordered delivery as this value is invalid.  If
there is an application running in the Unordered mode, there would be no
means to synchronize completion of a set of chunks across a Stream as SCTP
does not provide sufficient information although it could.

I see this feature assisting both a TCP model able to employ partial writes
in a software implementation as well as an SCTP off-load device that of
course would have no difficulty seeing this transport information.  I
suspect the reason for exclusion of TSN information from the ULP was that
there were few applications at the time considering Unordered delivery and
thus relied upon ordering provided by SCTP and SSN to isolate messages.

By allowing this feature, this allows the most expedient means of providing
signaling that would not get tangled in the transport's decision to hold a
stream and to transmit another out of sequence of the ULP thus increasing
possible HOL blocking at the shim.  For applications such as RDMA that make
little use of the Stream feature of SCTP, there is a desire to treat Streams
in unison and thus create more HOL blocking at the send should there be any
disparity between the shim send order and the SCTP send order.  As you may
know, there is ongoing work related to Stream flow control that could easily
exasperate this problem.

Doug

> > The second requirement is equally important.  As these
> > fragmented chunks are sent Unordered to ensure no buffering
> > is done by SCTP, completion signaling then requires that the
> > Transmission Sequence Number as well as the Cumulative
> > Transmission Sequence Number be visible.  As messages are
> > fragmented and reassembled by the shim as received, there is
> > no need for the Stream Sequence Number (it is invalid for
> > Unordered delivery) and its 16 bit size *may* represent a
> > potential limitation at very high bandwidths in the future.
> > A message is signaled complete to the ULP when the last chunk
> > in a message containing a marker is received and its TSN is
> > equal to or less than the cumulative TSN per serial math.
> > This marker signal is held at the shim layer until the TSN
> > condition is met.
>
> Can you explain why the shim cannot put in its own sequence number and
> the stack has to expose the internal TSN to the ULP?
>
> 							K. Poon.
> 							kcpoon@eng.sun.com


_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Wed Jan 16 10:00:22 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id KAA10039
	for <tsvwg-archive@odin.ietf.org>; Wed, 16 Jan 2002 10:00:17 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id KAA04041
	for tsvwg-archive@odin.ietf.org; Wed, 16 Jan 2002 10:00:20 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id JAA03035;
	Wed, 16 Jan 2002 09:36:07 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id JAA03006
	for <tsvwg@optimus.ietf.org>; Wed, 16 Jan 2002 09:36:05 -0500 (EST)
Received: from mail.tlc.polito.it (IDENT:root@prezzemolo.polito.it [130.192.9.131])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id JAA08949
	for <tsvwg@ietf.org>; Wed, 16 Jan 2002 09:36:00 -0500 (EST)
Received: from polito.it (IDENT:michela@pomodoro.polito.it [130.192.9.133])
	by mail.tlc.polito.it (8.11.6/8.11.6) with ESMTP id g0GEZZM15116;
	Wed, 16 Jan 2002 15:35:35 +0100
Message-ID: <3C458FB7.A8583AC3@polito.it>
Date: Wed, 16 Jan 2002 15:35:35 +0100
From: Michela Meo <michela@polito.it>
X-Mailer: Mozilla 4.76 [en] (X11; U; Linux 2.2.14 i686)
X-Accept-Language: en
MIME-Version: 1.0
To: tsvwg <tsvwg@ietf.org>
CC: Claudio Casetti <casetti@polito.it>,
        Marco Mellia <mellia@prezzemolo.polito.it>, michela@gramigna.polito.it
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Transfer-Encoding: 7bit
Subject: [Tsvwg] TCP Smart Framing
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit


Hi folks,
To revitalize the discussion on the Smart Framing option, let us
summarize 
the discussion so far:
We have been proposing a new segmentation algorithm that exploits
variable segment size. The goal of the scheme is to allow a TCP sender
to trigger fast recovery (FR) even when the congestion window is not
large enough (i.e., smaller than 4 segments).
The major motivations are:
1) The majority of the flow are short, which translates into few
segments
2) "few segments" means that the flow will end before exiting the
initial 
slow start, during which only the retransmission timeout (RTO) can be
used to recover from losses (since the cwnd is small)
3) Fast Recovery (FR) is a much better way to detect packet loss,
especially 
when latency is the major performance index. We feel that this is a real
problem (ever happened to click on a web link and wait 4 seconds before
anything happens? :-))
4) this is again true for every slow start phase, i.e., after every RTO.

Thus, allowing the TCP source to always have at least 4 segments in
flight 
(given that the flow is long enough...) greatly helps the sender to
receive enough acks and eventually to detect segment loss by FR.

Our scheme is engineered so that TCP-SF sender transmits exactly the
same load to the network as a classic TCP sender, so that it perfeclty
mimics its 
behavoiur in term of window growth, congestion control, 
aggressiveness/friendliness.


Some of the people there pointed out that
1) TCP-SF may have problem with the Karn's algorithm
2) By sending more segments, it is less efficient in terms of overhead 
introduced into the network
3) Other proposals (mainly TCPIW) acheive the same results

Summarizing the discussion so far, we can say that
1) as regards Karn's algorithm: TCP-SF must deal with it, and a simple 
modification is required. However, we also pointed out the Karn's
algorithm 
is effective for slow sources, rather than high speed data transfer,
while 
TCP-SF aims at enhancing TCP latency for short lived, http-like, web
traffic. 
Thus they deal with two different problems. However, we foresee no
problems in disabling the SF option when the Karn's algorithm kicks in.

2) Higher overhead introduced. Good point. As usual, you cannot have
your cake and eat it too. 
But this is ONLY true

a) when operating in the "small window" region (i.e., when the cwnd is
smaller than 4). In particular, only 5 more headers are sent considering
the variable size TCP-SF (as 4+4+4 smaller segments will be sent 
per RTT in the small window region versus 1+2+4 segments of a classic
TCP 
sender. After that TCP-SF will behave as a classic TCP sender...). 
Considering TCP-SF with fixed size option, the added overhead introduced
is 
much higher, since 4+8+12 segments are sent on the first three RTTs.

b) when the dropping probability is small. Indeed, TCP-SF is more
efficient in case of packet retransmission. To quantify this, if a
segment is dropped, classic TCP has to resend the whole segment, while
TCP-SF will resend only a small size segment. So if p is the dropping
probability (with the conservative assumption that it is the same for
both full size and small size segments), we have to transmit, due to a
packet loss: 
Classic TCP: (SMSS+Headers) + (SMSS+headers)*p
TCP-SF: (SMSS/4+Headers)*4 + (SMSS/4+Headers)*p
We can now find p=p0 so that the offered load is equal in both cases.
Clearly there is a tradeoff: for p<p0, the overhead introduced is larger
than the one of classic TCP in case of packet drop. But for p>p0, the
overhead introduced using MSS/4 is SMALLER. Considering SMSS~1460, you
can find that p0=~0.1, which amounts to saying that for high dropping
probabilities TCP-SF is more effective than classic TCP in managing
packet drops.

c) Larger number of ACKs on the reverse path. Although this is true, the
reverse path is normally less congested than the forward path.

3) It is true that TCPIW helps the sender in getting more 
packets per RTT. But, as the authors themselves recognize,
a) it is definitively more aggressive than classic TCP
b) it affects the initial slow start only
c) it does not help flows with large MSS (minor problem)

So, to keep discussing about TCP-SF, we'd like to ask you what you think 
about the key ideas and goals that it tries to exploit:
address latency of short lived TCP flows trying to minimize the RTO
penalty, 
using a novel segmentation algorithm.
Any feedback will be greatly appreciated.

Michela, Marco e Claudio

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Wed Jan 16 13:08:33 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id NAA16210
	for <tsvwg-archive@odin.ietf.org>; Wed, 16 Jan 2002 13:08:33 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id NAA12936
	for tsvwg-archive@odin.ietf.org; Wed, 16 Jan 2002 13:08:33 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id MAA11658;
	Wed, 16 Jan 2002 12:32:23 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id MAA11630
	for <tsvwg@optimus.ietf.org>; Wed, 16 Jan 2002 12:32:20 -0500 (EST)
Received: from linuxpow.com (IDENT:qmailr@linuxpow.com [12.149.2.10])
	by ietf.org (8.9.1a/8.9.1a) with SMTP id MAA15121
	for <tsvwg@ietf.org>; Wed, 16 Jan 2002 12:32:13 -0500 (EST)
Date: Wed, 16 Jan 2002 12:32:13 -0500 (EST)
Message-Id: <200201161732.MAA15121@ietf.org>
Received: (qmail 11514 invoked from network); 16 Jan 2002 16:37:05 -0000
Received: from buystainlessonline.com (HELO ) (nobody@12.149.2.55)
  by mail.buystainlessonline.com with SMTP; 16 Jan 2002 16:37:05 -0000
To: tsvwg@ietf.org
From: "BuyStainlessOnline.com" <sales@buystainlessonline.com>
Subject: [Tsvwg] The Stainless Steel Network MEGA DEAL
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

Unsubscribe by clicking the link at the bottom of this E-mail! 
It's quick and painless.

The Stainless Steel Network is brought to you by www.buystainlessonline.com

What BuyStainlessOnline users are doing...
----------------------------------------------------------------------
RECENT INQUIRIES:
904L Round Bar
168' 10mm
60'16mm
49' 25mm
NO QUOTE YET

321 Stainless Plate
1/2 x 60 x 96 (7) pcs
QUOTED

440C material, .500" dia x 144" long (840) pcs
QUOTED

310 16ga 48 x 120 20 sheets
QUOTED
------------------------------------------------------------------------
New Items for Sale

OFFER #1  (looking for Best Offer)
Commodity - Stainless Steel Wire.
Grade - AISI 304L.
Size - 0.135"
Surface finish - Matte (Medium Bright)
Condition - Soft Annealed.
U.T.S. - 85,000 to 100,000 psi max.
Quantity - aprox. 40,000 lbs.
Packing - Coils wrapped in HDPE and loaded on Mild Steel Carriers.
Remarks : All material is ex-mill in original mill's packaging.

OFFER #2
Attention Distributors!!! 3,000,000 Pounds of PRIME 304 COILS for sale
The following list of Stainless Steel Hot Rolled Coils is available FOB
Ohio for the following PRICES:
$.50 per pound (1 LOAD min)
$.47 per pound (10 LOAD min)

NOTE, these coils are MILL PRIME... Chemical Properties are listed below
------------------------------------------------------------------------
STEEL	FIN	WIDTH (mm)	THICK (mm)	Weigth (Pounds)	C	Cr	Cu	Mn	Mo	N	Ni	P	S	Si	
304	1	950.0	3.00	 37,238 	0.0450	18.070	0.330	1.470	0.230	0.055	8.090	0.027	0.001	0.350	
304	1	946.0	3.00	 31,698 	0.0470	18.070	0.280	1.320	0.250	0.042	8.040	0.030	0.001	0.360	
304	1	1256.0	3.00	 49,310 	0.0440	18.110	0.240	1.340	0.240	0.042	8.050	0.030	0.001	0.370	
304	1	1255.0	3.00	 44,451 	0.0500	18.290	0.240	1.370	0.250	0.046	8.030	0.031	0.001	0.420	
304	1	1254.0	3.00	 44,572 	0.0500	18.290	0.240	1.370	0.250	0.046	8.030	0.031	0.001	0.420	
304	1	1254.0	3.00	 46,113 	0.0500	18.060	0.320	1.350	0.350	0.042	8.070	0.030	0.001	0.420	
304	1	799.5	3.00	 31,239 	0.0432	18.110	0.250	1.350	0.270	0.042	8.100	0.029	0.001	0.340	
304	1	794.0	3.00	 30,328 	0.0410	18.220	0.280	1.420	0.300	0.047	8.120	0.030	0.001	0.350	
304	1	1256.5	3.00	 49,583 	0.0420	18.110	0.250	1.350	0.270	0.045	8.100	0.029	0.001	0.340	
304	1	1254.5	3.00	 49,453 	0.0420	18.110	0.250	1.350	0.270	0.045	8.100	0.029	0.001	0.340	
304	1	1255.0	3.00	 48,305 	0.0450	18.020	0.320	1.390	0.310	0.046	8.000	0.028	0.001	0.360	
304	1	1252.5	6.00	 49,449 	0.0420	18.030	0.330	1.360	0.240	0.050	8.070	0.025	0.001	0.390	
304	1	1251.5	6.00	 49,427 	0.0410	18.020	0.300	1.350	0.260	0.045	8.090	0.028	0.001	0.380	
304	1	1250.5	6.00	 49,414 	0.0410	18.020	0.300	1.350	0.260	0.045	8.090	0.028	0.001	0.380	
304	1	1253.5	6.00	 49,414 	0.0450	18.080	0.300	1.320	0.250	0.050	8.100	0.031	0.001	0.350	
304	1	1255.5	3.00	 41,821 	0.0430	18.070	0.340	1.390	0.270	0.051	8.100	0.026	0.001	0.480	
304	1	1255.5	2.80	 49,623 	0.0430	18.070	0.340	1.390	0.270	0.051	8.100	0.026	0.001	0.480	
304	1	1254.5	2.80	 49,438 	0.0420	18.090	0.330	1.400	0.330	0.043	8.150	0.026	0.001	0.360	
304	1	1254.0	2.80	 49,436 	0.0420	18.030	0.330	1.360	0.240	0.050	8.070	0.025	0.001	0.390	
304	1	1251.0	2.80	 48,435 	0.0410	18.060	0.270	1.330	0.250	0.051	8.150	0.026	0.001	0.340	
304	1	1256.0	2.80	 49,387 	0.0410	18.060	0.270	1.330	0.250	0.051	8.150	0.026	0.001	0.340	
304	1	1254.5	2.80	 49,493 	0.0410	18.060	0.270	1.330	0.250	0.051	8.150	0.026	0.001	0.340	
304	1	1253.5	2.80	 48,457 	0.0410	18.060	0.270	1.330	0.250	0.051	8.150	0.026	0.001	0.340	
304	1	1221.5	6.00	 49,074 	0.0430	18.060	0.410	1.350	0.240	0.049	8.070	0.027	0.001	0.470	
304	1	1254.5	6.00	 49,334 	0.0430	18.060	0.410	1.350	0.240	0.049	8.070	0.027	0.001	0.470	
304	1	1251.0	6.00	 48,155 	0.0400	18.040	0.320	1.410	0.200	0.050	8.060	0.029	0.001	0.360	
34A	1	947.0	4.70	 37,310 	0.0390	18.130	0.260	1.340	0.190	0.047	8.550	0.027	0.001	0.370	
304	1	951.5	3.00	 36,565 	0.0450	18.050	0.340	1.500	0.270	0.049	8.080	0.031	0.001	0.440	
304	1	950.0	3.00	 37,200 	0.0450	18.050	0.340	1.500	0.270	0.049	8.080	0.031	0.001	0.440	
304	1	951.0	3.00	 37,299 	0.0450	18.050	0.340	1.500	0.270	0.049	8.080	0.031	0.001	0.440	
304	1	949.5	3.00	 37,337 	0.0450	18.050	0.340	1.500	0.270	0.049	8.080	0.031	0.001	0.440	
304	1	948.5	3.00	 37,260 	0.0450	18.050	0.340	1.500	0.270	0.049	8.080	0.031	0.001	0.440	
304	1	1252.0	6.00	 43,333 	0.0450	18.080	0.300	1.320	0.250	0.050	8.100	0.031	0.001	0.350	
304	1	1251.5	6.00	 49,561 	0.0450	18.080	0.300	1.320	0.250	0.050	8.100	0.031	0.001	0.350	
304	1	946.5	3.00	 37,211 	0.0450	18.070	0.330	1.470	0.230	0.055	8.090	0.027	0.001	0.350	
304	1	947.5	3.00	 37,277 	0.0450	18.070	0.330	1.470	0.230	0.055	8.090	0.027	0.001	0.350	
304	1	951.5	3.00	 37,134 	0.0450	18.070	0.330	1.470	0.230	0.055	8.090	0.027	0.001	0.350	
304	1	949.0	3.00	 36,490 	0.0450	18.070	0.330	1.470	0.230	0.055	8.090	0.027	0.001	0.350	
304	1	1255.0	3.00	 48,450 	0.0460	18.030	0.240	1.370	0.210	0.042	8.030	0.029	0.001	0.360	
304	1	1256.0	3.00	 48,338 	0.0460	18.030	0.240	1.370	0.210	0.042	8.030	0.029	0.001	0.360	
304	1	1254.5	3.00	 49,195 	0.0460	18.030	0.240	1.370	0.210	0.042	8.030	0.029	0.001	0.360	
304	1	1255.5	3.00	 49,449 	0.0460	18.030	0.240	1.370	0.210	0.042	8.030	0.029	0.001	0.360	
304	1	1256.5	3.00	 48,638 	0.0450	18.090	0.250	1.360	0.250	0.049	8.070	0.028	0.001	0.370	
304	1	1255.5	3.00	 47,672 	0.0450	18.090	0.250	1.360	0.250	0.049	8.070	0.028	0.001	0.370	
304	1	1255.5	3.00	 49,482 	0.0450	18.090	0.250	1.360	0.250	0.049	8.070	0.028	0.001	0.370	
304	1	1254.5	3.00	 49,493 	0.0450	18.090	0.250	1.360	0.250	0.049	8.070	0.028	0.001	0.370	
304	1	1239.0	3.00	 49,151 	0.0450	18.090	0.250	1.360	0.250	0.049	8.070	0.028	0.001	0.370	
304	1	1255.0	6.00	 49,475 	0.0490	18.030	0.260	1.320	0.280	0.046	8.010	0.031	0.001	0.350	
304	1	1254.5	6.00	 49,286 	0.0490	18.030	0.260	1.320	0.280	0.046	8.010	0.031	0.001	0.350	
304	1	1251.5	6.00	 48,415 	0.0430	18.030	0.250	1.340	0.270	0.050	8.090	0.030	0.001	0.350	
304	1	1251.0	6.00	 47,824 	0.0430	18.030	0.250	1.340	0.270	0.050	8.090	0.030	0.001	0.350	
304	1	1252.5	6.00	 48,364 	0.0430	18.030	0.250	1.340	0.270	0.050	8.090	0.030	0.001	0.350	
34A	1	795.0	3.00	 28,968 	0.0400	18.010	0.270	1.370	0.200	0.048	8.550	0.028	0.001	0.300	
34A	1	798.0	3.00	 31,367 	0.0400	18.010	0.270	1.370	0.200	0.048	8.550	0.028	0.001	0.300	
34A	1	797.0	3.00	 31,389 	0.0400	18.010	0.270	1.370	0.200	0.048	8.550	0.028	0.001	0.300	
34A	1	796.5	3.00	 31,338 	0.0400	18.010	0.270	1.370	0.200	0.048	8.550	0.028	0.001	0.300	
34A	1	798.5	3.00	 31,550 	0.0400	18.010	0.270	1.370	0.200	0.048	8.550	0.028	0.001	0.300	
34A	1	797.5	3.00	 31,473 	0.0400	18.010	0.270	1.370	0.200	0.048	8.550	0.028	0.001	0.300	
34A	1	800.0	3.00	 30,613 	0.0400	18.010	0.270	1.370	0.200	0.048	8.550	0.028	0.001	0.300	
34A	1	798.0	3.00	 30,988 	0.0400	18.010	0.270	1.370	0.200	0.048	8.550	0.028	0.001	0.300	
34A	1	796.0	3.00	 31,354 	0.0400	18.010	0.270	1.370	0.200	0.048	8.550	0.028	0.001	0.300	
34A	1	798.0	3.00	 29,374 	0.0400	18.010	0.270	1.370	0.200	0.048	8.550	0.028	0.001	0.300	
304	1	1255.0	3.00	 41,960 	0.0490	18.030	0.260	1.320	0.280	0.046	8.010	0.031	0.001	0.350	
304	1	1253.5	3.00	 41,675 	0.0490	18.030	0.260	1.320	0.280	0.046	8.010	0.031	0.001	0.350	
304	1	1257.0	3.00	 43,397 	0.0420	18.060	0.300	1.380	0.340	0.048	8.040	0.028	0.001	0.350	
304	1	1258.0	3.00	 49,409 	0.0410	18.030	0.270	1.340	0.250	0.050	8.040	0.027	0.001	0.290	
304	1	1253.0	3.00	 49,433 	0.0410	18.030	0.270	1.340	0.250	0.050	8.040	0.027	0.001	0.290	
304	1	1253.0	3.00	 49,563 	0.0400	18.080	0.270	1.390	0.270	0.050	8.070	0.028	0.001	0.350	
304	1	1253.0	3.00	 49,497 	0.0400	18.080	0.270	1.390	0.270	0.050	8.070	0.028	0.001	0.350	
304	1	1256.0	3.00	 49,548 	0.0400	18.080	0.270	1.390	0.270	0.050	8.070	0.028	0.001	0.350	
304	1	1254.5	3.00	 48,616 	0.0400	18.080	0.270	1.390	0.270	0.050	8.070	0.028	0.001	0.350	
304	1	1258.0	3.00	 48,580 	0.0400	18.080	0.270	1.390	0.270	0.050	8.070	0.028	0.001	0.350	
304	1	1254.0	3.00	 49,460 	0.0420	18.060	0.300	1.380	0.340	0.048	8.040	0.028	0.001	0.350	
304	1	1254.0	3.00	 48,472 	0.0430	18.200	0.260	1.360	0.260	0.043	8.070	0.028	0.001	0.380	
34A	1	800.5	3.00	 31,534 	0.0410	18.030	0.330	1.340	0.250	0.050	8.530	0.029	0.001	0.360	
34A	1	798.0	3.00	 31,495 	0.0410	18.030	0.330	1.340	0.250	0.050	8.530	0.029	0.001	0.360	
34A	1	799.0	3.00	 31,499 	0.0410	18.030	0.330	1.340	0.250	0.050	8.530	0.029	0.001	0.360	
34A	1	797.5	3.00	 31,561 	0.0410	18.030	0.330	1.340	0.250	0.050	8.530	0.029	0.001	0.360	
				 3,343,970 											
															
OFFER #3
Stainless Steel SLIT COIL REMNANTS
FOB PHILADELPHIA, PRICES LISTED BELOW, NO MINIMUM PURCHASE.

	This information is provided by BuyStainlessOnline, Inc.				www.buystainlessonline.com					
	Please email Sales@buystainlessonline.com or call 215.604.5922 to place orders.									
	All materials are subject to prior sale.									

Tag	Description	Finish	Thickness	Width	Quantity	Weight	Location	Sale Price	Regular Price	
45146	28 GA 304	2B	0.015	18	1	5031	Phila, P.a.  USA	$1.10	$1.38	
30794	26 GA 201	2D	0.020	21.5	1	4158	Phila, P.a.  USA	$0.88	$1.09	
26202	26 GA 301	2B	0.020	1.375	1	418	Phila, P.a.  USA	$0.63	$0.78	
232512E	26 GA 304	2B	0.020	1.5	3	386	Phila, P.a.  USA	$0.64	$0.80	
45062	26 GA 304	POL/VC	0.020	11.75	1	2393	Phila, P.a.  USA	$0.63	$0.78	
229878	26 GA 430	2B	0.020	3.5	2	936	Phila, P.a.  USA	$0.19	$0.23	
22840	26 GA 430	BA	0.020	4.5	1	906	Phila, P.a.  USA	$0.25	$0.31	
29364	24 GA 301	POL	0.024	4	1	1427	Phila, P.a.  USA	$0.65	$0.81	
29363	24 GA 301	POL	0.024	4	1	1430	Phila, P.a.  USA	$0.65	$0.81	
35176	24 GA 301	POL	0.024	6	2	4727	Phila, P.a.  USA	$0.53	$0.66	
29362A	24 GA 304	POL	0.024	1.187	4	1024	Phila, P.a.  USA	$0.85	$1.06	
186896	24 GA 304	2B	0.024	2.75	9	2790	Phila, P.a.  USA	$0.75	$0.94	
19023	24 GA 304	2B	0.024	3.75	2	2762	Phila, P.a.  USA	$0.71	$0.89	
61931	24 GA 304	2B	0.024	9	1	1122	Phila, P.a.  USA	$0.59	$0.73	
29584	22 GA 301	POL	0.028	3.5	1	680	Phila, P.a.  USA	$0.66	$0.83	
32794	22 GA 301	2B	0.028	6.75	1	1578	Phila, P.a.  USA	$0.54	$0.67	
84108C	22 GA 304	POL	0.028	1.125	2	444	Phila, P.a.  USA	$0.90	$1.13	
29362B	22 GA 304	POL	0.028	1.187	6	1229	Phila, P.a.  USA	$0.85	$1.06	
32948	22 GA 304	POL	0.028	3.75	2	2858	Phila, P.a.  USA	$0.53	$0.66	
44796	22 GA 304	POL	0.028	11.25	1	2159	Phila, P.a.  USA	$0.69	$0.86	
224335	22 GA 430	POL	0.028	5.5	1	970	Phila, P.a.  USA	$0.06	$0.08	
45099	22 GA 430	BA	0.028	10	1	2175	Phila, P.a.  USA	$0.34	$0.42	
45098	22 GA 430	BA	0.028	10	1	2376	Phila, P.a.  USA	$0.34	$0.42	
243739	20 GA 304	2D	0.035	1.5	3	1194	Phila, P.a.  USA	$0.88	$1.09	
32943	20 GA 304	POL	0.035	3.125	3	2218	Phila, P.a.  USA	$0.78	$0.97	
46336C	20 GA 304	POL	0.035	3.5	1	1013	Phila, P.a.  USA	$0.99	$1.23	
48418E	20 GA 304	POL	0.035	4.125	1	877	Phila, P.a.  USA	$0.88	$1.09	
49488	20 GA 304	2B	0.036	6.1	1	948	Phila, P.a.  USA	$0.71	$0.89	
232335D	18 GA 304	POL	0.048	1.75	4	1111	Phila, P.a.  USA	$0.68	$0.84	
27329	18 GA 304	AF	0.048	1.875	1	260	Phila, P.a.  USA	$0.44	$0.55	
32953	18 GA 304	POL	0.048	2.5	1	513	Phila, P.a.  USA	$0.53	$0.66	
186895	18 GA 304	AF	0.048	3.5	1	610	Phila, P.a.  USA	$0.75	$0.94	
43255	18 GA 304	POL	0.048	5.25	4	5076	Phila, P.a.  USA	$0.44	$0.55	
212565	18 GA 409	2B	0.048	3.75	1	1143	Phila, P.a.  USA	$0.09	$0.11	
191101C	16 GA 304	POL	0.058	1	2	386	Phila, P.a.  USA	$0.98	$1.22	
191101B	16 GA 304	POL	0.058	2	3	1490	Phila, P.a.  USA	$0.98	$1.22	
44820	16 GA 304	BA	0.058	10	1	7504	Phila, P.a.  USA	$0.69	$0.86	
33017	16 GA 430	2B	0.058	4	1	1249	Phila, P.a.  USA	$0.25	$0.31	
43259	14 GA 304	POL	0.075	2.625	2	2684	Phila, P.a.  USA	$0.44	$0.55	
43258	14 GA 304	POL	0.075	4.5	2	1750	Phila, P.a.  USA	$0.44	$0.55	
43260	14 GA 304	POL	0.075	4.75	2	1977	Phila, P.a.  USA	$0.44	$0.55	
27327	14 GA 409	AF	0.075	4.562	1	550	Phila, P.a.  USA	$0.31	$0.39	
27328	14 GA 409	AF	0.075	4.687	1	531	Phila, P.a.  USA	$0.31	$0.39	
35360	MISC GA 3003	H14	0.075	1	4	319	Phila, P.a.  USA	$0.50	$0.63	
168033B	13 GA 301	2B	0.083	3.75	4	3518	Phila, P.a.  USA	$0.65	$0.81	
168033C	13 GA 301	2B	0.083	5.25	2	2442	Phila, P.a.  USA	$0.65	$0.81	
168033A	13 GA 301	2B	0.083	5.25	2	2419	Phila, P.a.  USA	$0.65	$0.81	
243738	13 GA 304	2B	0.083	2.187	3	4028	Phila, P.a.  USA	$0.53	$0.66	
215992A	12 GA 304	2B	0.105	2.75	1	1190	Phila, P.a.  USA	$0.50	$0.63	
43257	12 GA 304	POL	0.105	6.437	2	3092	Phila, P.a.  USA	$0.44	$0.55	
243494	12 GA 304	2B	0.105	20.9	1	8050	Phila, P.a.  USA	$0.48	$0.59	
41325	11 GA 304	2B	0.120	3	1	1772	Phila, P.a.  USA	$0.63	$0.78	
41326	11 GA 304	2B	0.120	3	1	1772	Phila, P.a.  USA	$0.63	$0.78	
41327	11 GA 304	2B	0.120	3	1	1752	Phila, P.a.  USA	$0.63	$0.78	
41328	11 GA 304	2B	0.120	3	1	1761	Phila, P.a.  USA	$0.63	$0.78	
215992	11 GA 304	2B	0.120	3.125	1	1150	Phila, P.a.  USA	$0.50	$0.63	
47400D	11 GA 304	POL	0.120	8.22	1	4630	Phila, P.a.  USA	$0.84	$1.05	
38870	11 GA 316	#1	0.120	3.625	2	5454	Phila, P.a.  USA	$0.60	$0.75	
49098	11 GA 430	BA	0.120	7.781	1	1828	Phila, P.a.  USA	$0.31	$0.39	
46217I	11 GA 5052	H32	0.120	1.5	2	500	Phila, P.a.  USA	$1.10	$1.38	
45031	11 GA 304	HRAP	0.123	1	2	927	Phila, P.a.  USA	$0.60	$0.75	
47400A	11 GA 304	POL	0.125	2.625	2	1272	Phila, P.a.  USA	$0.84	$1.05	
47400C	11 GA 304	POL	0.125	2.625	4	2538	Phila, P.a.  USA	$0.84	$1.05	
42144	10 GA 316L	#1	0.130	8	1	3726	Phila, P.a.  USA	$0.78	$0.97	
45027	3/16" 304L	HRAP	0.196	3	1	1745	Phila, P.a.  USA	$0.60	$0.75	
44814A	1/4" 316L	#1	0.250	1	1	286	Phila, P.a.  USA	$0.83	$1.03	
44814C	1/4" 316L	#1	0.250	1	1	344	Phila, P.a.  USA	$0.83	$1.03	
44814F	1/4" 316L	#1	0.250	1.5	2	1690	Phila, P.a.  USA	$0.83	$1.03	
71103F	1/4" 316L	#1	0.250	2	1	1620	Phila, P.a.  USA	$0.56	$0.70	
44814B	1/4" 316L	#1	0.250	2	5	6140	Phila, P.a.  USA	$0.83	$1.03	
44814D	1/4" 316L	#1	0.250	2	5	5635	Phila, P.a.  USA	$0.83	$1.03	
44814E	1/4" 316L	#1	0.250	2	4	4530	Phila, P.a.  USA	$0.83	$1.03	
44814G	1/4" 316L	#1	0.250	2	2	952	Phila, P.a.  USA	$0.83	$1.03	
44814H	1/4" 316L	#1	0.250	4	3	7385	Phila, P.a.  USA	$0.83	$1.03	


www.BuyStainlessOnline.com
Your Place for Stainless Today.

P 215.604.5922
F 240.358.8483

Click Here to REGISTER!
https://www.buystainlessonline.com/registration/registration.php

Unsubscribe By clicking below: 
 http://www.buystainlessonline.com/email/mail.php?action=delete&eval=54804&email=tsvwg@ietf.org

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Wed Jan 16 13:36:13 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id NAA17151
	for <tsvwg-archive@odin.ietf.org>; Wed, 16 Jan 2002 13:36:07 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id NAA14408
	for tsvwg-archive@odin.ietf.org; Wed, 16 Jan 2002 13:36:09 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id NAA12709;
	Wed, 16 Jan 2002 13:02:13 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id NAA12680
	for <tsvwg@optimus.ietf.org>; Wed, 16 Jan 2002 13:02:11 -0500 (EST)
Received: from motgate.mot.com (motgate.mot.com [129.188.136.100])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id NAA16030
	for <tsvwg@ietf.org>; Wed, 16 Jan 2002 13:02:08 -0500 (EST)
Received: [from mothost.mot.com (mothost.mot.com [129.188.137.101]) by motgate.mot.com (motgate 2.1) with ESMTP id LAA24683 for <tsvwg@ietf.org>; Wed, 16 Jan 2002 11:02:07 -0700 (MST)]
Received: [from relay2.cig.mot.com (relay2.cig.mot.com [136.182.15.24]) by mothost.mot.com (MOT-mothost 2.0) with ESMTP id LAA11246 for <tsvwg@ietf.org>; Wed, 16 Jan 2002 11:02:06 -0700 (MST)]
Received: from quantix20.cig.mot.com (quantix20 [160.47.58.31])
	by relay2.cig.mot.com (8.11.4/8.11.4) with ESMTP id g0GI23718296;
	Wed, 16 Jan 2002 12:02:03 -0600 (CST)
Received: from cig.mot.com (d50-38e1.cig.mot.com [160.47.56.225])
	by quantix20.cig.mot.com (8.8.8+Sun/8.8.8) with ESMTP id MAA27535;
	Wed, 16 Jan 2002 12:06:48 -0600 (CST)
Message-ID: <3C45C04B.DD04F504@cig.mot.com>
Date: Wed, 16 Jan 2002 12:02:51 -0600
From: Qiaobing Xie <xieqb@cig.mot.com>
X-Mailer: Mozilla 4.77 [en] (Win98; U)
X-Accept-Language: en
MIME-Version: 1.0
To: Randall Stewart <randall@stewart.chicago.il.us>
CC: TSV WG list <tsvwg@ietf.org>
Subject: Re: [Tsvwg] fragmentation and PMTU
References: <3C4346DC.D5F3301B@stewart.chicago.il.us>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Transfer-Encoding: 7bit
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit

Randy,

I fully understand what you want to do with the implementation and I
think it can be a very useful feature for certain applications. However,
even with the current wording in RFC2960, I think it is already allowed
and if you do so nothing is going to be violated. In particular, there
is nothing in RFC2960 to prevent one from building a "dual-mode" stack
that can operate in either "fragmentation-capable" or
"fragmentation-incapable" state via an application controlled switch :-)

-Qiaobing

Randall Stewart wrote:
> 
> Dear All:
> 
> It has come to my attention that the following
> is currently in RFC2960:
> 
> "
> 6.9 Fragmentation and Reassembly
> 
>    An endpoint MAY support fragmentation when sending DATA chunks, but
>    MUST support reassembly when receiving DATA chunks.  If an endpoint
>    supports fragmentation, it MUST fragment a user message if the size
>    of the user message to be sent causes the outbound SCTP packet size
>    to exceed the current MTU.  If an implementation does not support
>    fragmentation of outbound user messages, the endpoint must return an
>    error to its upper layer and not attempt to send the user message.
> "
> 
> This all seems fine until one looks carefully at one little aspect:
> 
> - If I an implementation supports fragmentation it MUST fragment.
> 
> The issue I have with this is what if an application layer says:
> 
> "hey I know you can fragment, but please don't, I want you
>  to reject sends that exceed the P-MTU."
> 
> This may be a perfectly reasonable behavior but the wording above
> dis-allows that... I know the intent was to prevent folks from
> just setting the DF bit and NOT doing P-MTU discovery... but what
> we probably need here is a little clearification for the implementors
> guide... something like.
> 
> "
>  Note: If an implementation that supports fragmentation makes
>  available to its upper layer a mechanism to turn off fragmentation
>  it may do so. However in so doing, it MUST react just like an
>  implementation that does NOT support fragmentation i.e. it MUST
>  reject sends that exceed the current P-MTU.
> "
> 
> I think this note (+wordsmitthing :>) would allow the behavior
> I mention above without harming the intent of the P-MTU/Fragmentation.
> 
> (Word smithing suggestions would be welcome here :>)
> 
> What do you'all think?
> 
> Thanks
> 
> R
> 
> --
> Randall R. Stewart
> randall@stewart.chicago.il.us 815-342-5222 (cell phone)
> 
> _______________________________________________
> tsvwg mailing list
> tsvwg@ietf.org
> http://www1.ietf.org/mailman/listinfo/tsvwg


_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Wed Jan 16 14:00:19 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id OAA17684
	for <tsvwg-archive@odin.ietf.org>; Wed, 16 Jan 2002 14:00:19 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id OAA15220
	for tsvwg-archive@odin.ietf.org; Wed, 16 Jan 2002 14:00:22 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id NAA14371;
	Wed, 16 Jan 2002 13:34:42 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id NAA14343
	for <tsvwg@optimus.ietf.org>; Wed, 16 Jan 2002 13:34:41 -0500 (EST)
Received: from sj-msg-core-1.cisco.com (sj-msg-core-1.cisco.com [171.71.163.11])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id NAA17111
	for <tsvwg@ietf.org>; Wed, 16 Jan 2002 13:34:37 -0500 (EST)
Received: from mira-sjc5-2.cisco.com (mira-sjc5-2.cisco.com [171.71.163.16])
	by sj-msg-core-1.cisco.com (8.11.3/8.9.1) with ESMTP id g0GIY8F10385;
	Wed, 16 Jan 2002 10:34:08 -0800 (PST)
Received: from stewart.chicago.il.us (ssh-sj1.cisco.com [171.68.225.134])
	by mira-sjc5-2.cisco.com (Mirapoint)
	with ESMTP id AAT79232;
	Wed, 16 Jan 2002 10:34:07 -0800 (PST)
Message-ID: <3C45C79E.77D23774@stewart.chicago.il.us>
Date: Wed, 16 Jan 2002 12:34:06 -0600
From: Randall Stewart <randall@stewart.chicago.il.us>
X-Mailer: Mozilla 4.76 [en] (X11; U; Linux 2.2.12 i386)
X-Accept-Language: en
MIME-Version: 1.0
To: Qiaobing Xie <xieqb@cig.mot.com>
CC: TSV WG list <tsvwg@ietf.org>
Subject: Re: [Tsvwg] fragmentation and PMTU
References: <3C4346DC.D5F3301B@stewart.chicago.il.us> <3C45C04B.DD04F504@cig.mot.com>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Transfer-Encoding: 7bit
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit

Qiaobing:

I understand what you are after here but I think I dis-agree :0

I think we intended that if you can't fragment you get an error.
This is clearly stated in RFC2960. But if you read the words
carefully the only way you COULD disable fragmentation on
an implementation that is fragmentation capable is to "pretend"
that your flag made your whole stack incapable of fragmentation.

This is only because of our poor wording to try to prevent
folks from incorrectly implementing SCTP without PMTU discovery.

Now I think you are really streching when the same stack
may have one endpoint that has disabled fragmentation and another
that has NOT disabled fragmentation. Thus the stack is really
"streching" the truth if it trys to obey RFC2960 ... in order
to reject the sends...

I think the wording below is not harmful and prevents a stack
from violating the RFC by allowing the lee-way we should have
allowed in the first place :>

R


Qiaobing Xie wrote:
> 
> Randy,
> 
> I fully understand what you want to do with the implementation and I
> think it can be a very useful feature for certain applications. However,
> even with the current wording in RFC2960, I think it is already allowed
> and if you do so nothing is going to be violated. In particular, there
> is nothing in RFC2960 to prevent one from building a "dual-mode" stack
> that can operate in either "fragmentation-capable" or
> "fragmentation-incapable" state via an application controlled switch :-)
> 
> -Qiaobing
> 
> Randall Stewart wrote:
> >
> > Dear All:
> >
> > It has come to my attention that the following
> > is currently in RFC2960:
> >
> > "
> > 6.9 Fragmentation and Reassembly
> >
> >    An endpoint MAY support fragmentation when sending DATA chunks, but
> >    MUST support reassembly when receiving DATA chunks.  If an endpoint
> >    supports fragmentation, it MUST fragment a user message if the size
> >    of the user message to be sent causes the outbound SCTP packet size
> >    to exceed the current MTU.  If an implementation does not support
> >    fragmentation of outbound user messages, the endpoint must return an
> >    error to its upper layer and not attempt to send the user message.
> > "
> >
> > This all seems fine until one looks carefully at one little aspect:
> >
> > - If I an implementation supports fragmentation it MUST fragment.
> >
> > The issue I have with this is what if an application layer says:
> >
> > "hey I know you can fragment, but please don't, I want you
> >  to reject sends that exceed the P-MTU."
> >
> > This may be a perfectly reasonable behavior but the wording above
> > dis-allows that... I know the intent was to prevent folks from
> > just setting the DF bit and NOT doing P-MTU discovery... but what
> > we probably need here is a little clearification for the implementors
> > guide... something like.
> >
> > "
> >  Note: If an implementation that supports fragmentation makes
> >  available to its upper layer a mechanism to turn off fragmentation
> >  it may do so. However in so doing, it MUST react just like an
> >  implementation that does NOT support fragmentation i.e. it MUST
> >  reject sends that exceed the current P-MTU.
> > "
> >
> > I think this note (+wordsmitthing :>) would allow the behavior
> > I mention above without harming the intent of the P-MTU/Fragmentation.
> >
> > (Word smithing suggestions would be welcome here :>)
> >
> > What do you'all think?
> >
> > Thanks
> >
> > R
> >
> > --
> > Randall R. Stewart
> > randall@stewart.chicago.il.us 815-342-5222 (cell phone)
> >
> > _______________________________________________
> > tsvwg mailing list
> > tsvwg@ietf.org
> > http://www1.ietf.org/mailman/listinfo/tsvwg
> 
> _______________________________________________
> tsvwg mailing list
> tsvwg@ietf.org
> http://www1.ietf.org/mailman/listinfo/tsvwg

-- 
Randall R. Stewart
randall@stewart.chicago.il.us 815-342-5222 (cell phone)

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Wed Jan 16 23:56:52 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id XAA29971
	for <tsvwg-archive@odin.ietf.org>; Wed, 16 Jan 2002 23:56:47 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id XAA04413
	for tsvwg-archive@odin.ietf.org; Wed, 16 Jan 2002 23:56:52 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id XAA04112;
	Wed, 16 Jan 2002 23:39:08 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id XAA04084
	for <tsvwg@optimus.ietf.org>; Wed, 16 Jan 2002 23:39:06 -0500 (EST)
Received: from elk.icir.org (elk.icir.org [192.150.187.21])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id XAA29686
	for <tsvwg@ietf.org>; Wed, 16 Jan 2002 23:38:59 -0500 (EST)
Received: from elk.icir.org (localhost [127.0.0.1])
	by elk.icir.org (8.11.3/8.11.3) with ESMTP id g0H4cub87630;
	Wed, 16 Jan 2002 20:38:56 -0800 (PST)
	(envelope-from floyd@elk.icir.org)
Message-Id: <200201170438.g0H4cub87630@elk.icir.org>
To: Michela Meo <michela@polito.it>
cc: Claudio Casetti <casetti@polito.it>,
        Marco Mellia <mellia@prezzemolo.polito.it>, tsvwg@ietf.org
From: Sally Floyd <floyd@aciri.org>
Subject: Re: [Tsvwg] TCP Smart Framing 
Date: Wed, 16 Jan 2002 20:38:56 -0800
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

Michela -

>To revitalize the discussion on the Smart Framing option, let us
>summarize 
>the discussion so far:

I have not followed this discussion in complete detail, but I would
suggest that in summarizing the discussion so far, it would also
be useful to summarize the relationship of SF with Limited Transmit
(LT), RFC 3042, which has been Proposed Standard for a year now.

If the goal is "to allow a TCP sender to trigger fast recovery (FR)
even when the congestion window is not large enough", Limited
Transmit lets you do this in many but not all cases.  Limited
Transmit sends a new packet (if there is a new packet to send) when
the first and second dup acks are received, so in some scenarios
it lets you keep sending data, and keep the dup-ack clock going,
until Fast Retransmit is invoked.

But there are cases where Limited Transmit wouldn't prevent a
timeout, and SF would.  An example would be when the sender had no
new data to send.  But instead of using smaller-than-normal packets
to induce a Fast Retransmit, it might be better to ask the question:
if you get a single duplicate acknowledgement, and there is no new
data to send, should you be able to retransmit old data that might
or might not be lost?  A "yes" answer to this question would take
care of many of the additional cases where Limited Transmit in its
current form is not sufficient to prevent a timeout, without
unnecessarily reducing the packet size, and without increasing the
number of packets used by all connections when they have small
windows.

And if the last packet of a transmission is lost, nothing is going
to prevent the need for a timeout, I am afraid.  

...
>Our scheme is engineered so that TCP-SF sender transmits exactly the
>same load to the network as a classic TCP sender,...

I don't buy it that four small packets presents the same load to
the network as one larger packet.  Aside from the issue of header
size, there can be router limitations in terms of packets per second
processing capacity in addition to the link bandwidth limitation
in terms of bytes per second.  It is still not clear (to me) which
limitation, packets per second or bytes per second, is likely to
dominate some years in the future.

- Sally
http://www.icir.org/floyd/

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Thu Jan 17 06:05:05 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id GAA11624
	for <tsvwg-archive@odin.ietf.org>; Thu, 17 Jan 2002 06:05:05 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id GAA26077
	for tsvwg-archive@odin.ietf.org; Thu, 17 Jan 2002 06:05:08 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id FAA24588;
	Thu, 17 Jan 2002 05:19:16 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id FAA24521
	for <tsvwg@optimus.ietf.org>; Thu, 17 Jan 2002 05:19:12 -0500 (EST)
Date: Thu, 17 Jan 2002 05:19:12 -0500 (EST)
Received: from excel-careers.be (mail.excel-careers.be [195.74.208.117])
	by ietf.org (8.9.1a/8.9.1a) with SMTP id FAA11136
	for <tsvwg@ietf.org>; Thu, 17 Jan 2002 05:19:02 -0500 (EST)
Received: from dbjj.com [64.86.192.77] by excel-careers.be [195.74.208.117] with SMTP (MDaemon.v2.7.SP5.R) for <tsvwg@ietf.org>; Thu, 17 Jan 2002 07:45:18 +0100
Message-ID: <52ec2197$2a8b$5933>
Mime-Version: 1.0
Content-Type: multipart/mixed;
	boundary="----=_NextPart_30570_30389_448C.346F7801F40B"
To: "tsvwg@ietf.org" <tsvwg@ietf.org>
From: "" <email@ewasher.org>
Reply-to: email@ewasher.org
X-MDaemon-Deliver-To: tsvwg@ietf.org
X-Return-Path: email@ewasher.org
Subject: [Tsvwg] Effective Online Marketing
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org


------=_NextPart_30570_30389_448C.346F7801F40B
Content-Type: text/html; 
	charset="us-ascii"
Content-Transfer-Encoding: 8bit

<html xmlns:o="urn:schemas-microsoft-com:office:office"
xmlns:w="urn:schemas-microsoft-com:office:word"
xmlns="http://www.w3.org/TR/REC-html40">

<head>
<meta http-equiv=Content-Type content="text/html; charset=windows-1252">
<meta name=ProgId content=Word.Document>
<meta name=Generator content="Microsoft Word 9">
<meta name=Originator content="Microsoft Word 9">
<link rel=File-List href="./bulkispmsg_files/filelist.xml">
<title>        </title>
<!--[if gte mso 9]><xml>
 <o:DocumentProperties>
  <o:Author>admin</o:Author>
  <o:LastAuthor>gfghfhg</o:LastAuthor>
  <o:Revision>2</o:Revision>
  <o:TotalTime>49</o:TotalTime>
  <o:LastPrinted>2002-01-13T16:15:00Z</o:LastPrinted>
  <o:Created>2002-01-13T16:15:00Z</o:Created>
  <o:LastSaved>2002-01-13T16:15:00Z</o:LastSaved>
  <o:Pages>2</o:Pages>
  <o:Words>269</o:Words>
  <o:Characters>1534</o:Characters>
  <o:Lines>12</o:Lines>
  <o:Paragraphs>3</o:Paragraphs>
  <o:CharactersWithSpaces>1883</o:CharactersWithSpaces>
  <o:Version>9.2720</o:Version>
 </o:DocumentProperties>
</xml><![endif]-->
<style>
<!--
 /* Font Definitions */
@font-face
	{font-family:Wingdings;
	panose-1:5 0 0 0 0 0 0 0 0 0;
	mso-font-charset:2;
	mso-generic-font-family:auto;
	mso-font-pitch:variable;
	mso-font-signature:0 268435456 0 0 -2147483648 0;}
 /* Style Definitions */
p.MsoNormal, li.MsoNormal, div.MsoNormal
	{mso-style-parent:"";
	margin:0in;
	margin-bottom:.0001pt;
	mso-pagination:widow-orphan;
	font-size:12.0pt;
	font-family:"Times New Roman";
	mso-fareast-font-family:"Times New Roman";}
p.MsoBodyText, li.MsoBodyText, div.MsoBodyText
	{margin:0in;
	margin-bottom:.0001pt;
	mso-pagination:widow-orphan;
	font-size:12.0pt;
	font-family:"Times New Roman";
	mso-fareast-font-family:"Times New Roman";
	color:navy;
	font-weight:bold;}
p.MsoBodyText2, li.MsoBodyText2, div.MsoBodyText2
	{margin:0in;
	margin-bottom:.0001pt;
	mso-pagination:widow-orphan;
	font-size:12.0pt;
	font-family:"Times New Roman";
	mso-fareast-font-family:"Times New Roman";
	color:blue;
	font-weight:bold;}
p.MsoBodyText3, li.MsoBodyText3, div.MsoBodyText3
	{margin:0in;
	margin-bottom:.0001pt;
	mso-pagination:widow-orphan;
	font-size:12.0pt;
	font-family:"Times New Roman";
	mso-fareast-font-family:"Times New Roman";
	font-weight:bold;}
a:link, span.MsoHyperlink
	{color:blue;
	text-decoration:underline;
	text-underline:single;}
a:visited, span.MsoHyperlinkFollowed
	{color:purple;
	text-decoration:underline;
	text-underline:single;}
@page Section1
	{size:8.5in 11.0in;
	margin:1.0in 1.25in 1.0in 1.25in;
	mso-header-margin:.5in;
	mso-footer-margin:.5in;
	mso-paper-source:0;}
div.Section1
	{page:Section1;}
 /* List Definitions */
@list l0
	{mso-list-id:392043152;
	mso-list-type:hybrid;
	mso-list-template-ids:-2098311420 67698689 67698691 67698693 67698689 67698691 67698693 67698689 67698691 67698693;}
@list l0:level1
	{mso-level-number-format:bullet;
	mso-level-text:\F0B7;
	mso-level-tab-stop:.75in;
	mso-level-number-position:left;
	margin-left:.75in;
	text-indent:-.25in;
	font-family:Symbol;}
@list l1
	{mso-list-id:828518779;
	mso-list-type:hybrid;
	mso-list-template-ids:-881447762 67698689 67698691 67698693 67698689 67698691 67698693 67698689 67698691 67698693;}
@list l1:level1
	{mso-level-number-format:bullet;
	mso-level-text:\F0B7;
	mso-level-tab-stop:111.0pt;
	mso-level-number-position:left;
	margin-left:111.0pt;
	text-indent:-.25in;
	font-family:Symbol;}
@list l2
	{mso-list-id:882787630;
	mso-list-type:hybrid;
	mso-list-template-ids:1278536142 67698689 67698691 67698693 67698689 67698691 67698693 67698689 67698691 67698693;}
@list l2:level1
	{mso-level-number-format:bullet;
	mso-level-text:\F0B7;
	mso-level-tab-stop:.5in;
	mso-level-number-position:left;
	text-indent:-.25in;
	font-family:Symbol;}
@list l3
	{mso-list-id:1026370737;
	mso-list-type:hybrid;
	mso-list-template-ids:-2116024552 67698689 67698691 67698693 67698689 67698691 67698693 67698689 67698691 67698693;}
@list l3:level1
	{mso-level-number-format:bullet;
	mso-level-text:\F0B7;
	mso-level-tab-stop:.5in;
	mso-level-number-position:left;
	text-indent:-.25in;
	font-family:Symbol;}
@list l3:level3
	{mso-level-number-format:bullet;
	mso-level-text:\F0A7;
	mso-level-tab-stop:1.5in;
	mso-level-number-position:left;
	text-indent:-.25in;
	font-family:Wingdings;}
ol
	{margin-bottom:0in;}
ul
	{margin-bottom:0in;}
-->
</style>
</head>

<body lang=EN-US link=blue vlink=purple style='tab-interval:.5in'>

<div class=Section1>

<p class=MsoNormal><span style="mso-spacerun: yes">젨젨젨� </span></p>

<p class=MsoNormal><b><span style='font-size:14.0pt;mso-bidi-font-size:12.0pt;
color:#0033CC'>We will put your product or service instantly into the hands of
millions!<o:p></o:p></span></b></p>

  <p class=MsoNormal><b><span style='font-size:14.0pt;mso-bidi-font-size:12.0pt'> 
    &nbsp;<o:p></o:p></span></b></p>

<p class=MsoNormal><b><span style='color:navy'>Since 1996, Bulk Email Network
has provided bulk email marketing to thousands of <o:p></o:p></span></b></p>

<p class=MsoNormal><b><span style='color:navy'>well-satisfied customers. We
offer the most competitive prices in the industry, made <o:p></o:p></span></b></p>

<p class=MsoBodyText>possible by our high percentage of repeat business. We
have the most advanced direct email technology employed by only a knowledgeable
few in the world. </p>

  <p class=MsoNormal><b> &nbsp;<o:p></o:p></b></p>

<p class=MsoNormal><b><span style='color:navy'>We have over 160 million active
email addresses All sorted by country, state, city and target.<o:p></o:p></span></b></p>

  <p class=MsoNormal><b><span style='color:navy'> &nbsp;<o:p></o:p></span></b></p>

<p class=MsoBodyText>Call us for a free consultation at 323 876 6148<span
style="mso-spacerun: yes">� </span>[U.S.A.]. </p>

  <p class=MsoNormal><b><span style='color:navy'> &nbsp;<o:p></o:p></span></b></p>

<p class=MsoBodyText>We guarantee the <span style='color:red'>lowest prices or
your service is free!<o:p></o:p></span></p>

  <p class=MsoNormal><b><span style='color:navy'> &nbsp;<o:p></o:p></span></b></p>

<p class=MsoNormal><b><span style='color:navy'>You will get at least </span><span
style='color:red'>1% response rate</span><span style='color:navy'> or we will
repeat the mailing to a new list for no extra charge.</span></b><b><span
style='font-size:8.0pt;mso-bidi-font-size:12.0pt;color:navy'><o:p></o:p></span></b></p>

<p class=MsoNormal><b><span style='color:navy'><span style="mso-spacerun:
yes">젨젨젨젨젨젨젨젨젨젨젨젨젨젨젨젨젨젨젨젨젨 </span><span
style="mso-spacerun: yes">젨젨젨젨젨젨젨젨젨젨젨젨젨젨젨젨젨젨</span></span></b><span
style='font-size:8.0pt;mso-bidi-font-size:12.0pt'><o:p></o:p></span></p>

<p class=MsoNormal><span style='color:black'><span style="mso-spacerun:
yes">�</span><b>1) Let's say you... Sell a $24.95 PRODUCT or SERVICE.<o:p></o:p></b></span></p>

<p class=MsoNormal><b><span style='color:black'><span style="mso-spacerun:
yes">�</span>2) Let's say you... Mass Email to 1,000,000 PEOPLE DAILY.<o:p></o:p></span></b></p>

<p class=MsoNormal><b><span style='color:black'><span style="mso-spacerun:
yes">�</span>3) Let's say you... Receive JUST 1 ORDER for EVERY 2,500 EMAILS.<o:p></o:p></span></b></p>

  <p class=MsoNormal><b> &nbsp;<o:p></o:p></b></p>

<p class=MsoNormal><b><span style="mso-spacerun: yes">�</span>CALCULATION OF
YOUR EARNINGS BASED ON THE ABOVE STATISTICS:<o:p></o:p></b></p>

<p class=MsoNormal><b><span style="mso-spacerun: yes">�</span>[Day 1]: <span
style='color:#0033CC'>$9,980</span><span style="mso-spacerun: yes">�
</span>[Week <span style='color:black'>1]:</span><span style='color:red'> </span><span
style='color:#0033CC'>$69,860</span><span style="mso-spacerun: yes">�
</span>[Month 1]: <span style='color:#0033CC'>$279,440<o:p></o:p></span></b></p>

  <p class=MsoNormal><b><span style='color:#0033CC'> &nbsp;<o:p></o:p></span></b></p>

<p class=MsoNormal><b>Now you know why you receive so many email
advertisements...<o:p></o:p></b></p>

  <p class=MsoNormal><b><span style='color:#0033CC'> &nbsp;<o:p></o:p></span></b></p>

<p class=MsoBodyText2><span style='color:#0033CC'>Best of ALL, Bulk Email
Network can be used as a 100% TAX WRITE OFF for your Business!<o:p></o:p></span></p>

  <p class=MsoNormal><span style='color:#0033CC'> &nbsp;<o:p></o:p></span></p>

<p class=MsoNormal>===============================================================</p>

<p class=MsoNormal><b>&quot;Many business people are finding out that they can
now advertise in ways that they <o:p></o:p></b></p>

<p class=MsoBodyText3>never could have afforded in the past.<span
style="mso-spacerun: yes">� </span>The cost of sending mass e-mail is extremely
low, and the response rate is high and quick.&quot; - <span style='color:red'>USA
TODAY</span></p>

<p class=MsoNormal>===============================================================</p>

  <p class=MsoNormal>&nbsp; <o:p></o:p></p>

  <p class=MsoNormal><b> &nbsp;<o:p></o:p></b></p>

<p class=MsoNormal><b>Best Regards,<o:p></o:p></b></p>

  <p class=MsoNormal><b> &nbsp;<o:p></o:p></b></p>

<p class=MsoNormal><b>Jennifer O묬onner,<o:p></o:p></b></p>

<p class=MsoNormal><b>Bulk Email Network<o:p></o:p></b></p>

<p class=MsoNormal><b>Marketing Dept.<o:p></o:p></b></p>

  <p class=MsoNormal><b>&nbsp; <o:p></o:p></b></p>

<p class=MsoNormal><b>Phone:1(323) 876 6148 [U.S.A.]<o:p></o:p></b></p>

  <p class=MsoNormal><b> &nbsp;<o:p></o:p></b></p>

  <p class=MsoNormal><b> &nbsp;<o:p></o:p></b></p>

<p class=MsoNormal><b>To be Opt-out Please email: <a
href="mailto:optout@ewasher.org">optout@ewasher.org</a> with the email address
that you would like removed.<o:p></o:p></b></p>

  <p class=MsoNormal><b> &nbsp;<o:p></o:p></b></p>

  <p class=MsoNormal><b> &nbsp;<o:p></o:p></b></p>

</div>

</body>

</html><!--5b2a5ea66e1d519028d558233e9455bba165acf-->

------=_NextPart_30570_30389_448C.346F7801F40B--


_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Thu Jan 17 09:24:16 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id JAA16336
	for <tsvwg-archive@odin.ietf.org>; Thu, 17 Jan 2002 09:24:15 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id JAA03560
	for tsvwg-archive@odin.ietf.org; Thu, 17 Jan 2002 09:24:18 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id JAA03000;
	Thu, 17 Jan 2002 09:05:09 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id JAB02970
	for <tsvwg@optimus.ietf.org>; Thu, 17 Jan 2002 09:05:06 -0500 (EST)
Received: from mercury.Sun.COM (mercury.Sun.COM [192.9.25.1])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id JAA15688
	for <tsvwg@ietf.org>; Thu, 17 Jan 2002 09:05:02 -0500 (EST)
Received: from sunuk.UK.Sun.COM ([129.156.85.58])
	by mercury.Sun.COM (8.9.3+Sun/8.9.3) with ESMTP id GAA18685;
	Thu, 17 Jan 2002 06:05:02 -0800 (PST)
Received: from watford-109 (watford-109.UK.Sun.COM [129.156.199.109])
	by sunuk.UK.Sun.COM (8.9.3+Sun/8.9.3/ENSMAIL,v2.1p1) with SMTP id OAA16214;
	Thu, 17 Jan 2002 14:05:00 GMT
Message-Id: <200201171405.OAA16214@sunuk.UK.Sun.COM>
Date: Thu, 17 Jan 2002 14:05:46 +0000 (GMT)
From: "Jeremy Harris [RU-UK]" <jgh@uk.sun.com>
Reply-To: "Jeremy Harris [RU-UK]" <jgh@uk.sun.com>
To: mallman@grc.nasa.gov
Cc: tsvwg@ietf.org
MIME-Version: 1.0
Content-Type: TEXT/plain; charset=us-ascii
Content-MD5: ygvtA8SUfhHAV6wDtkJowA==
X-Mailer: dtmail 1.3.0 @(#)CDE Version 1.4.2 SunOS 5.9 sun4u sparc 
Subject: [Tsvwg] TCP's performance in the face of packet reordering
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

>Ethan Blanton, Mark Allman. On Making TCP More Robust to Packet 
>Reordering. ACM Computer Communication Review, 32(1), January
>2002.
>http://roland.grc.nasa.gov/~mallman/papers/tcp-reorder-ccr.ps

Wouldn't the use of SACK make most of this moot?

Jeremy				jeremy.harris@sun.com


_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Thu Jan 17 09:25:35 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id JAA16366
	for <tsvwg-archive@odin.ietf.org>; Thu, 17 Jan 2002 09:25:34 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id JAA03639
	for tsvwg-archive@odin.ietf.org; Thu, 17 Jan 2002 09:25:38 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id JAA02957;
	Thu, 17 Jan 2002 09:05:00 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id JAA02925
	for <tsvwg@optimus.ietf.org>; Thu, 17 Jan 2002 09:04:58 -0500 (EST)
Received: from seraph3.grc.nasa.gov (firewall-user@seraph3.lerc.nasa.gov [128.156.10.12])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id JAA15678
	for <tsvwg@ietf.org>; Thu, 17 Jan 2002 09:04:53 -0500 (EST)
Received: by seraph3.grc.nasa.gov (Postfix, from userid 5)
	id 43DFC640EA; Thu, 17 Jan 2002 09:04:43 -0500 (EST)
Received: from lombok-fi.lerc.nasa.gov(139.88.112.33) by seraph3.grc.nasa.gov via csmap (V6.0)
	id srcAAArZaiUs; Thu, 17 Jan 02 09:04:43 -0500
Received: from guns.lerc.nasa.gov (guns.lerc.nasa.gov [139.88.87.35])
	by lombok-fi.lerc.nasa.gov (NASA LeRC 8.9.1.1/8.9.1) with ESMTP id JAA01729;
	Thu, 17 Jan 2002 09:04:42 -0500 (EST)
Received: from guns.lerc.nasa.gov (localhost.lerc.nasa.gov [127.0.0.1]) by guns.lerc.nasa.gov with ESMTP (NASA LeRC 8.7.4.1/2.01-local)
        id JAA23404; Thu, 17 Jan 2002 09:04:41 -0500 (EST)
Message-Id: <200201171404.JAA23404@guns.lerc.nasa.gov>
To: Sally Floyd <floyd@aciri.org>
From: Mark Allman <mallman@grc.nasa.gov>
Reply-To: mallman@grc.nasa.gov
Cc: Michela Meo <michela@polito.it>, Claudio Casetti <casetti@polito.it>,
        Marco Mellia <mellia@prezzemolo.polito.it>, tsvwg@ietf.org
Subject: Re: [Tsvwg] TCP Smart Framing 
Organization: BBN Technologies/NASA GRC
Song-of-the-Day: Blue on Black
Date: Thu, 17 Jan 2002 09:04:41 -0500
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org


> But there are cases where Limited Transmit wouldn't prevent a
> timeout, and SF would.  An example would be when the sender had no
> new data to send.  But instead of using smaller-than-normal
> packets to induce a Fast Retransmit, it might be better to ask the
> question: if you get a single duplicate acknowledgement, and there
> is no new data to send, should you be able to retransmit old data
> that might or might not be lost?  A "yes" answer to this question
> would take care of many of the additional cases where Limited
> Transmit in its current form is not sufficient to prevent a
> timeout, without unnecessarily reducing the packet size, and
> without increasing the number of packets used by all connections
> when they have small windows.

As a note, in the draft version of the LT spec we did include
something we called "early retransmit" (if I remember correctly).
The algorithm was basically if (a) your cwnd is <= 3 packets *and*
(b) you have no new data to send then you are allowed to trigger
fast retransmit on cwnd-1 dup ACKs (where cwnd is measured in
segments).  The idea being that if you get a loss with a 2 or 3
packet cwnd and have nothing new to send you will never get 3 dup
ACKs and so you can trigger early.  (OK, maybe not never -- the
network could duplicate an ACK, but that is hopefully rare and
doesn't seem like something to count on! ;)

Mysteriously, consensus for this idea was not reached.  There are,
in fact, re-ordering pathologies that can cause this scheme to
inject a lot of useless traffic into the network.  I would hope that
maybe you could mitigate those.  I still think the idea has some
merit (especially if used with a couple of conditions -- i.e., you
can do it only once per connection or you have to be able to detect
if you're persistently dumping spurious retransmits into the network
and take action to avoid that or ...).

allman


---
Mark Allman -- BBN/NASA GRC -- http://roland.grc.nasa.gov/~mallman/

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Thu Jan 17 09:33:26 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id JAA16610
	for <tsvwg-archive@odin.ietf.org>; Thu, 17 Jan 2002 09:33:26 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id JAA04262
	for tsvwg-archive@odin.ietf.org; Thu, 17 Jan 2002 09:33:29 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id JAA03248;
	Thu, 17 Jan 2002 09:12:13 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id JAA03222
	for <tsvwg@optimus.ietf.org>; Thu, 17 Jan 2002 09:12:11 -0500 (EST)
Received: from seraph3.grc.nasa.gov (firewall-user@seraph3.lerc.nasa.gov [128.156.10.12])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id JAA15924
	for <tsvwg@ietf.org>; Thu, 17 Jan 2002 09:12:08 -0500 (EST)
Received: by seraph3.grc.nasa.gov (Postfix, from userid 5)
	id 4FC5B640F4; Thu, 17 Jan 2002 09:09:50 -0500 (EST)
Received: from lombok-fi.lerc.nasa.gov(139.88.112.33) by seraph3.grc.nasa.gov via csmap (V6.0)
	id srcAAAyWaqVt; Thu, 17 Jan 02 09:09:50 -0500
Received: from guns.lerc.nasa.gov (guns.lerc.nasa.gov [139.88.87.35])
	by lombok-fi.lerc.nasa.gov (NASA LeRC 8.9.1.1/8.9.1) with ESMTP id JAA03072;
	Thu, 17 Jan 2002 09:09:49 -0500 (EST)
Received: from guns.lerc.nasa.gov (localhost.lerc.nasa.gov [127.0.0.1]) by guns.lerc.nasa.gov with ESMTP (NASA LeRC 8.7.4.1/2.01-local)
        id JAA23461; Thu, 17 Jan 2002 09:09:48 -0500 (EST)
Message-Id: <200201171409.JAA23461@guns.lerc.nasa.gov>
To: "Jeremy Harris [RU-UK]" <jgh@uk.sun.com>
Cc: "Ethan Blanton" <eblanton@prime.cs.ohiou.edu>
From: Mark Allman <mallman@grc.nasa.gov>
Reply-To: mallman@grc.nasa.gov
Cc: tsvwg@ietf.org
Organization: BBN Technologies/NASA GRC
Song-of-the-Day: Blue on Black
Date: Thu, 17 Jan 2002 09:09:48 -0500
Subject: [Tsvwg] Re: TCP's performance in the face of packet reordering
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org


> >Ethan Blanton, Mark Allman. On Making TCP More Robust to Packet 
> >Reordering. ACM Computer Communication Review, 32(1), January
> >2002.
> >http://roland.grc.nasa.gov/~mallman/papers/tcp-reorder-ccr.ps
> 
> Wouldn't the use of SACK make most of this moot?

I do not follow.  But, my gut reaction is the answer is "no".  I
can't see how SACK solves the problems we were looking at.  And, we
used SACK TCP as the baseline in the paper (in fact, every variant
of TCP we used in the paper included SACK) and it clearly has
problems with reordering (as expected).

allman


---
Mark Allman -- BBN/NASA GRC -- http://roland.grc.nasa.gov/~mallman/

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Thu Jan 17 09:35:08 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id JAA16659
	for <tsvwg-archive@odin.ietf.org>; Thu, 17 Jan 2002 09:35:07 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id JAA04348
	for tsvwg-archive@odin.ietf.org; Thu, 17 Jan 2002 09:35:11 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id JAA03354;
	Thu, 17 Jan 2002 09:17:16 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id JAA03327
	for <tsvwg@optimus.ietf.org>; Thu, 17 Jan 2002 09:17:09 -0500 (EST)
Received: from patan.sun.com (patan.Sun.COM [192.18.98.43])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id JAA16103
	for <tsvwg@ietf.org>; Thu, 17 Jan 2002 09:17:05 -0500 (EST)
Received: from sunuk.UK.Sun.COM ([129.156.85.58])
	by patan.sun.com (8.9.3+Sun/8.9.3) with ESMTP id HAA14412;
	Thu, 17 Jan 2002 07:17:06 -0700 (MST)
Received: from watford-109 (watford-109.UK.Sun.COM [129.156.199.109])
	by sunuk.UK.Sun.COM (8.9.3+Sun/8.9.3/ENSMAIL,v2.1p1) with SMTP id OAA23080;
	Thu, 17 Jan 2002 14:17:04 GMT
Message-Id: <200201171417.OAA23080@sunuk.UK.Sun.COM>
Date: Thu, 17 Jan 2002 14:17:51 +0000 (GMT)
From: "Jeremy Harris [RU-UK]" <jgh@uk.sun.com>
Reply-To: "Jeremy Harris [RU-UK]" <jgh@uk.sun.com>
To: Jeremy.Harris@Sun.COM, mallman@grc.nasa.gov
Cc: eblanton@prime.cs.ohiou.edu, tsvwg@ietf.org
MIME-Version: 1.0
Content-Type: TEXT/plain; charset=us-ascii
Content-MD5: ZIXuXkoWNEc2nlwUWoZ64Q==
X-Mailer: dtmail 1.3.0 @(#)CDE Version 1.4.2 SunOS 5.9 sun4u sparc 
Subject: [Tsvwg] Re: TCP's performance in the face of packet reordering
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org


> > >Ethan Blanton, Mark Allman. On Making TCP More Robust to Packet 
> > >Reordering. ACM Computer Communication Review, 32(1), January
> > >2002.
> > >http://roland.grc.nasa.gov/~mallman/papers/tcp-reorder-ccr.ps
> > 
> > Wouldn't the use of SACK make most of this moot?
> 
> I do not follow.  But, my gut reaction is the answer is "no".  I
> can't see how SACK solves the problems we were looking at.  And, we
> used SACK TCP as the baseline in the paper (in fact, every variant
> of TCP we used in the paper included SACK) and it clearly has
> problems with reordering (as expected).

I think you were assuming that enough of any ACK rxd, including SACK,
would trigger a "bad decision" - which you then reasonably went on to
consider how to undo.  But having a SACK shouldn't, I think, count
as such.  Am I wrong?  Why were the "bad decisions" being made?

Cheers,
    Jeremy
    


_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@ns.ietf.org  Thu Jan 17 10:33:37 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id KAA18961
	for <tsvwg-archive@odin.ietf.org>; Thu, 17 Jan 2002 10:33:36 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id KAA06814
	for tsvwg-archive@odin.ietf.org; Thu, 17 Jan 2002 10:33:38 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id KAA05615;
	Thu, 17 Jan 2002 10:09:04 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id KAA05585
	for <tsvwg@ns.ietf.org>; Thu, 17 Jan 2002 10:09:01 -0500 (EST)
Received: from mclean5-mail.usae.bah.com (mclean5-mail.usae.bah.com [156.80.5.101])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id KAA18149
	for <tsvwg@ietf.org>; Thu, 17 Jan 2002 10:08:59 -0500 (EST)
Received: from BAH507013 ([156.80.71.108]) by
          mclean5-mail.usae.bah.com (Netscape Messaging Server 4.15) with
          SMTP id GQ38R100.ELQ for <tsvwg@ietf.org>; Thu, 17 Jan 2002
          10:09:01 -0500 
From: "Davenport Michael" <davenport_michael@bah.com>
To: <tsvwg@ietf.org>
Date: Thu, 17 Jan 2002 10:09:01 -0500
Message-ID: <NFBBKDFGMKODILIOLBDHKEOMCJAA.davenport_michael@bah.com>
MIME-Version: 1.0
Content-Type: text/plain;
	charset="iso-8859-1"
Content-Transfer-Encoding: 7bit
X-Priority: 3 (Normal)
X-MSMail-Priority: Normal
X-Mailer: Microsoft Outlook IMO, Build 9.0.2416 (9.0.2911.0)
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4133.2400
Importance: Normal
Content-Transfer-Encoding: 7bit
Subject: [Tsvwg] Question regarding the SCPS protocol
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit

All,

I hope this is the right forum for this question and if it isn't, please let
me know.  I am doing some testing of the SCPS protocol (along w/ other
proprietary devices) for transport over satellite links w/ chara. of high
BER and long delays; a comparison/contrast/performance study.  I basically
know what this protocol can do, but I wanted to poll the community who may
have extensive use with this protocol on how they feel about it and if it is
worth the investment/time to investigate further.  Any comments would be
welcome (dislikes, likes, issues, etc...).  If anyone is interested I could
supply the results when we are finished.  Thanks...


v/r,

Michael Davenport




_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Thu Jan 17 15:11:01 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id PAA07047
	for <tsvwg-archive@odin.ietf.org>; Thu, 17 Jan 2002 15:11:00 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id PAA21865
	for tsvwg-archive@odin.ietf.org; Thu, 17 Jan 2002 15:11:03 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id OAA21141;
	Thu, 17 Jan 2002 14:55:22 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id OAA21112
	for <tsvwg@optimus.ietf.org>; Thu, 17 Jan 2002 14:55:20 -0500 (EST)
Received: from boreas.isi.edu (boreas.isi.edu [128.9.160.161])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id OAA05174
	for <tsvwg@ietf.org>; Thu, 17 Jan 2002 14:55:16 -0500 (EST)
Received: from nit (nit.isi.edu [128.9.160.116])
	by boreas.isi.edu (8.11.6/8.11.2) with ESMTP id g0HJt9N22865;
	Thu, 17 Jan 2002 11:55:09 -0800 (PST)
Date: Thu, 17 Jan 2002 11:55:09 -0800
From: Aaron Falk <falk@ISI.EDU>
To: Davenport Michael <davenport_michael@bah.com>
cc: tsvwg@ietf.org
Subject: Re: [Tsvwg] Question regarding the SCPS protocol
Message-ID: <18430000.1011297309@nit>
In-Reply-To: <NFBBKDFGMKODILIOLBDHKEOMCJAA.davenport_michael@bah.com>
References:  <NFBBKDFGMKODILIOLBDHKEOMCJAA.davenport_michael@bah.com>
X-Mailer: Mulberry/2.1.2 (Linux/x86)
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Disposition: inline
Content-Transfer-Encoding: 7bit
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit

Michael-

There is a community of people who have played with SCPS and you'll be
better able to reach them via the TCP over Satellite mailing list
(tcpsat@grc.nasa.gov).  You also might consider reading the following:

RFC2488: Enhancing TCP Over Satellite Channels using Standard Mechanisms
RFC2760: Ongoing TCP Research Related to Satellites

Since you are doing testing, you might also consider reading

http://roland.grc.nasa.gov/~mallman/papers/tcp-evaluation.ps

Hope this helps,

--aaron


--On Thursday, January 17, 2002 10:09:01 AM -0500 Davenport Michael
<davenport_michael@bah.com> wrote:

> All,
> 
> I hope this is the right forum for this question and if it isn't, please
> let me know.  I am doing some testing of the SCPS protocol (along w/ other
> proprietary devices) for transport over satellite links w/ chara. of high
> BER and long delays; a comparison/contrast/performance study.  I basically
> know what this protocol can do, but I wanted to poll the community who may
> have extensive use with this protocol on how they feel about it and if it
> is worth the investment/time to investigate further.  Any comments would
> be welcome (dislikes, likes, issues, etc...).  If anyone is interested I
> could supply the results when we are finished.  Thanks...
> 
> 
> v/r,
> 
> Michael Davenport
> 
> 
> 
> 
> _______________________________________________
> tsvwg mailing list
> tsvwg@ietf.org
> http://www1.ietf.org/mailman/listinfo/tsvwg



_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
http://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Fri Jan 18 06:26:56 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id GAA06546
	for <tsvwg-archive@odin.ietf.org>; Fri, 18 Jan 2002 06:26:56 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id GAA05866
	for tsvwg-archive@odin.ietf.org; Fri, 18 Jan 2002 06:26:58 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id GAA05363;
	Fri, 18 Jan 2002 06:01:49 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id GAA05335
	for <tsvwg@optimus.ietf.org>; Fri, 18 Jan 2002 06:01:47 -0500 (EST)
Received: from beamer.mchh.siemens.de (beamer.mchh.siemens.de [194.138.158.163])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id GAA06184
	for <tsvwg@ietf.org>; Fri, 18 Jan 2002 06:01:42 -0500 (EST)
Received: from moody.mchh.siemens.de (mail2.mchh.siemens.de [194.138.158.226])
	by beamer.mchh.siemens.de (8.9.3/8.9.3) with ESMTP id MAA10433
	for <tsvwg@ietf.org>; Fri, 18 Jan 2002 12:01:44 +0100 (MET)
Received: from mchh247e.demchh201e.icn.siemens.de ([139.21.200.57])
	by moody.mchh.siemens.de (8.9.1/8.9.1) with ESMTP id MAA23455
	for <tsvwg@ietf.org>; Fri, 18 Jan 2002 12:01:44 +0100 (MET)
Received: by MCHH247E with Internet Mail Service (5.5.2653.19)
	id <XD3ZYPXZ>; Fri, 18 Jan 2002 12:01:43 +0100
Message-ID: <F92978223B01D311ACFF0008C71EE19D015DF6FA@MCHH225E>
From: Tuexen Michael <Michael.Tuexen@icn.siemens.de>
To: "'tsvwg@ietf.org'" <tsvwg@ietf.org>
Date: Fri, 18 Jan 2002 12:01:17 +0100
MIME-Version: 1.0
X-Mailer: Internet Mail Service (5.5.2653.19)
Content-Type: text/plain
Subject: [Tsvwg] FW: SCTP is an official Connectathon Technology!
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org



Michael Tuexen    Tel.:   +49 89 722 47210
Siemens AG        Fax:    +49 89 722 48212
ICN WN CS SE 5    E-mail: Michael.Tuexen@icn.siemens.de

> -----Original Message-----
> From:	Steven M. Glass [SMTP:Steven.Glass@sun.com]
> Sent:	Thursday, January 17, 2002 10:29 PM
> To:	sctp@sun.com
> Subject:	SCTP is an official Connectathon Technology!
> 
> 
>     This is to announce that we've got our 3 enrolled participants (you know
> who you are), so SCTP is officially on the list of Connectathon technologies! 
> It will be added to the web page shortly.
> 
>     To clarify the latest on the controversial USCTP draft - as with *all*
> extensions, testing is optional, that is nobody is expected to test USCTP. 
> For those who do test any extension, it MUST NOT interfere with the testing of
> the core protocol.
> 
>     That said, I'd like to know if anyone has any other comments, complaints,
> or additions they'd like to voice for our test plan.  While participants
> obviously have more control over what will be tested, even those who may not
> (yet) be planning on participating have opinions, and can contribute to the
> big picture!
> 
>     As I announced Tuesday, the early deadline date for SCTP participants has
> been extended to tomorrow, Friday, January 18th, 2002.  The overall
> registration deadline remains February 7th for all participants.  I'm
> expecting at least 2 or 3 others for SCTP, so this should be a great
> opportunity for those participating!
> 
>                               Cheers,
>                                   Steve

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Fri Jan 18 08:41:29 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id IAA10308
	for <tsvwg-archive@odin.ietf.org>; Fri, 18 Jan 2002 08:41:28 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id IAA12397
	for tsvwg-archive@odin.ietf.org; Fri, 18 Jan 2002 08:41:26 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id IAA11562;
	Fri, 18 Jan 2002 08:28:05 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id IAA11534
	for <tsvwg@optimus.ietf.org>; Fri, 18 Jan 2002 08:28:04 -0500 (EST)
Received: from sj-msg-core-2.cisco.com (sj-msg-core-2.cisco.com [171.69.24.11])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id IAA09856
	for <tsvwg@ietf.org>; Fri, 18 Jan 2002 08:28:01 -0500 (EST)
Received: from mira-sjc5-2.cisco.com (mira-sjc5-2.cisco.com [171.71.163.16])
	by sj-msg-core-2.cisco.com (8.11.3/8.9.1) with ESMTP id g0IDRXj19449
	for <tsvwg@ietf.org>; Fri, 18 Jan 2002 05:27:33 -0800 (PST)
Received: from stewart.chicago.il.us (ssh-sj1.cisco.com [171.68.225.134])
	by mira-sjc5-2.cisco.com (Mirapoint)
	with ESMTP id AAU32634;
	Fri, 18 Jan 2002 05:27:32 -0800 (PST)
Message-ID: <3C4822C3.9FA639E0@stewart.chicago.il.us>
Date: Fri, 18 Jan 2002 07:27:31 -0600
From: Randall Stewart <randall@stewart.chicago.il.us>
X-Mailer: Mozilla 4.76 [en] (X11; U; Linux 2.2.12 i386)
X-Accept-Language: en
MIME-Version: 1.0
To: TSV WG list <tsvwg@ietf.org>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Transfer-Encoding: 7bit
Subject: [Tsvwg] An interesting student project with SCTP
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit

Dear All:

These two links provide some interesting information on
SCTP and u-SCTP. Interesting reading...

http://www.cs.ucla.edu/classes/fall01/cs218/l1/project/student/12-06.8.doc

http://www.cs.ucla.edu/classes/fall01/cs218/l1/project/student/12-06.8.ppt

Enjoy

R
-- 
Randall R. Stewart
randall@stewart.chicago.il.us 815-342-5222 (cell phone)

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@ns.ietf.org  Fri Jan 18 19:04:09 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id TAA01284
	for <tsvwg-archive@odin.ietf.org>; Fri, 18 Jan 2002 19:04:09 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id TAA10610
	for tsvwg-archive@odin.ietf.org; Fri, 18 Jan 2002 19:04:11 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id SAA10036;
	Fri, 18 Jan 2002 18:43:05 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id SAA10008
	for <tsvwg@ns.ietf.org>; Fri, 18 Jan 2002 18:43:01 -0500 (EST)
Received: from localhost ([211.216.8.17])
	by ietf.org (8.9.1a/8.9.1a) with SMTP id SAA00989
	for <tsvwg@ietf.org>; Fri, 18 Jan 2002 18:42:55 -0500 (EST)
Message-Id: <200201182342.SAA00989@ietf.org>
Reply-To: acaba2@dreamwiz.com
From: 최석훈<acaba2@hitel.net>
To: tsvwg@ietf.org
Mime-Version: 1.0
Content-Type: text/html; charset="ks_c_5601-1987"
Date: Sat, 19 Jan 2002 08:45:39 +0900
Subject: [Tsvwg] 신계념 쇼핑몰 하우리더 오픈! 사업기회를 잡자! [광고]
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
<!-- saved from url=(0167)http://himail.hitel.net/cgi-bin/readcontent.cgi?id=acaba2&uid=121&Status=0&TotalCount=87&CurrentNum=3&version=0&TempLocation=1009983657&hanusername=최석훈&folder=inbox -->
<HTML><HEAD>
<META content="Microsoft FrontPage 5.0" name=GENERATOR>
<STYLE>A {
TEXT-DECORATION: none
}
</STYLE>
<LINK href="a" rel=stylesheet>
<TITLE>편지읽기</TITLE>
<META http-equiv=Content-Type content="text/html; charset=ks_c_5601-1987">
<META content="Microsoft FrontPage 5.0" name=GENERATOR>
<META http-equiv=Description name=Description contents="하이텔 홈"><LINK
href="하우리더.files/hitelcss.css" rel=stylesheet><LINK title=WebMailer
href="하우리더.files/WebMailer.css" type=text/css rel=StyleSheet>
<SCRIPT src="하우리더.files/WebMailer.js"
type=text/javascript>
<!--
-->
</SCRIPT>
<SCRIPT language=JavaScript>
<!--
function allcheck()
{
for(var j=0; j < document.folderlist.elements.length; j++)
{
var checke = document.folderlist.elements[j];
if(checke.name != "allcheckbox")
checke.checked = document.folderlist.allcheckbox.checked;
}
}
function SubmitFormMove()
{
var UidList;
UidList = "|";
alert(document.folderlist.elements.length);
for(var j=0; j < document.folderlist.elements.length; j++)
{
var checke = document.folderlist.elements[j];
if(checke.name != "allcheckbox" && checke.checked)
{
UidList = UidList + document.folderlist.elements[j].name + "|";
}
}
alert(UidList);
document.movefolder.uid.value = UidList;
var temp;
index = document.upcommand.selectfolder.options.selectedIndex;
temp = document.upcommand.selectfolder.options[index].value;
document.movefolder.targetfolder.value = document.upcommand.selectfolder.options[index].value;
alert(temp);
document.movefolder.submit();
}
function openwindowdetail(url,folder,name)
{
if(navigator.appName.indexOf("Netscape") > -1){
urls = url +  "&foldername=" + folder;
}
else {
urls = url +  "&foldername=" + escape(folder);
}
openbufwin = window.open(urls, name, "toolbar=yes,location=no,status=no,resizable=yes,scrollbars=yes,width=640,height=480");
openbufwin.focus();
return;
}
// -->
</SCRIPT>
</HEAD>
<BODY bgColor=#ffffff leftMargin=0 topMargin=0 marginwidth="0" marginheight="0">
<TABLE height=400 cellSpacing=0 cellPadding=0 width=569 border=0>
<TR>
<TD vAlign=top align=left width=15 height=376>　</TD>
<TD vAlign=top align=left width=539 height=376>
<META content="MSHTML 6.00.2600.0" name=GENERATOR>
<STYLE>A {
TEXT-DECORATION: none
}
</STYLE>
<LINK href="a" rel=stylesheet>
<DIV style="width: 562; height: 1593">
<TABLE cellSpacing=0 cellPadding=0 width=562 align=center
border=0><TBODY>
<TR vAlign=top>
<TD class=main align=middle>
<CENTER>
<TABLE cellSpacing=0 cellPadding=0 width=560 align=center>
<TBODY>
<TR>
<TD colSpan=3><!--<a href="">--><SPAN
style="FONT-SIZE: 9pt">
<IMG height=110 alt="" hspace=0
src="http://acaba3.hihome.com/mailtop.gif" width=561
border=0></SPAN></TD></TR>
<TR>
<TD width=5 bgColor=#92d24d><SPAN
style="FONT-SIZE: 9pt">&nbsp;</SPAN></TD>
<TD align=middle width=551>
<TABLE cellSpacing=0 cellPadding=5 width=520>
<TBODY>
<TR borderColor=#ff9933>
<TD borderColor=#3399ff width=496 bgColor=#f2f7f7>
<P align=center><SPAN
style="FONT-SIZE: 10pt"><B><A
href="http://www.etimesi.com/news/detail.html?id=200112210096"
target=_blank><FONT color=fuchsia>[전자신문 보도
기사]</FONT> </A>☜ 클릭</B></SPAN><SPAN
style="FONT-SIZE: 9pt">&nbsp;(홍보)<BR></SPAN>
<P><SPAN style="FONT-SIZE: 9pt">허락없이 메일을 드려 죄송합니다.
<BR>　</SPAN></P>
<P><SPAN style="FONT-SIZE: 9pt">호소하오니 이 글을 꼭 읽어 보시고 당신도 저와 같은
행복감을 느껴 보실 수 있기를  </SPAN></P>
<P><SPAN style="FONT-SIZE: 9pt">바랍니다. 새해에는 님의 소망하시는 바가 꼭
성취되시기를 간절히 기원합니다.</SPAN></P>
<HR>
<P><SPAN style="FONT-SIZE: 9pt">국내 굴지의 인터넷 쇼핑몰들이
적자경영에서 헤어나지 못하고 있는 가운데 소비자에게 홍보비를 지급하는 이른바
<a target="_blank" href="http://www.howleader.com/">마핑몰(www.HowLeader.com)</a>이 오픈했습니다.
마핑몰이란 마케팅과 쇼핑몰을 합성한 용어로 고객은 단순히 쇼핑하는 수준에서 그치지 않고
자신의 광고 노력에 따른 보상을 지급 받게 됩니다.</SPAN></P>
<P><SPAN style="FONT-SIZE: 9pt">즉 회사는 마케팅 영업을 하지
않는 대신 이러한 노력을 소비자들에게 위임함으로써 기존에 마케팅에 소요되던 비용을
소비자들의 매출기여도에 맞게 보상으로 제공하는 것입니다.</SPAN></P>
<P><SPAN style="FONT-SIZE: 9pt">일반적으로 소비자들은 매체 광고에
의존하기 보다 주위의 친구나 직장 동료들에게 정보를 얻어 제품을 구입하는 경향이 있습니다.
그런데도 직접적인 정보 전달자에게는 아무런 혜택이 주어지지 않는 게 현실이지요. 이점에
착안하여 직접적으로 유용한 정보를 전달해준 당사자에게 광고비를 지급하는 마핑몰이 탄생하게
된 것입니다.</SPAN></P>
<P><SPAN style="FONT-SIZE: 9pt">또한 인터넷 쇼핑몰의 편리성과
네트워크 마케팅의 장점인 독특한 보상시스템을 도입하여 자신이 홍보하여 가입한 회원들 뿐만
아니라 그 회원들이 홍보하여 가입한 회원들이 제품을 구입할 때도 광고비를 지급합니다.
유통의 변화는 소비자에게 유리한 방향으로 진행되어 왔습니다. 이제 인터넷 쇼핑몰의 궁극적인
지향점은 소비자의 권리가 존중받는 마핑몰인 것입니다.</SPAN></P>
<P><SPAN style="FONT-SIZE: 9pt">저희는 현재 국내 유명 브랜드의
화장품을 총 망라하여 최저가로 공급하고 있으며, 농특산물, 생활자기, 전통공예품, 꽃배달
서비스, 보험, 생활용품 및 고품격 가구 등 2800여종의 다양한 상품을 구비하고
있습니다. 또한 내년 상반기내로 5000여 품목으로 확대할 계획을 가지고
있습니다.</SPAN></P>
<P><SPAN style="FONT-SIZE: 9pt">기존의 인터넷 쇼핑몰들이 충성도가
높은 고객의 유지에 어려움을 겪고 있는데다가 새로운 고객 유치에도 많은 마케팅 비용을
할애하고 있는 가운데 이 비용을 고객의 광고 노력에 대한 보상으로 환원시키는 하우리더
쇼핑몰의 탄생은 향후 인터넷 쇼핑몰의 새로운 비즈니스 모델로 제시될
것입니다.</SPAN></P>
<P><SPAN style="FONT-SIZE: 9pt">하우리더 쇼핑몰을 통해서 고객이
받게되는 혜택은 항구적인 인세수입입니다.</SPAN></P>
<P><SPAN style="FONT-SIZE: 9pt">인세수입이란 책이나 음반처럼 그
제작과정은 어렵고 힘이 들지만 일단 출시되고 나면 더 이상 일을 하지 않아도 물건이 팔림과
동시에 지속적인 수입이 발생하는 것을 의미합니다. 즉 자신의 노력의 결과가 지속적으로
이어져 꼬박꼬박 일정 수입을 제공받는 것을 말합니다. 인세수입 소득자는 자신이 아파서 더
이상 일을 할 수 없을 때에도 이미 쌓아놓은 노력으로 인해 안정적인 생활을 보장받게
됩니다.</SPAN></P>
<P><SPAN style="FONT-SIZE: 9pt">하우리더는 최고의 제품만을
엄선하였으며 마케팅 자체를 없애고 가급적 산지 직거래를 추진하여 유통 거품을 제거함으로써
제품 구입가의 10-30%를 고객에게 광고비 명목으로 환원시키고 있습니다. 또한 중국의
WTO 가입으로 많은 어려움을 겪고 있는 신토불이 농산물 제품의 활로를 제공하여, 하우리더
쇼핑몰은 우리 농촌 살리기의 견인차 역할을 담당하고 있습니다. </SPAN></P>
<P><SPAN style="FONT-SIZE: 9pt">고객이 하는 일은 단지 저처럼
주위 사람들에게 쇼핑몰에 대한 정보를 알리는 일 뿐입니다. 기존의 다단계 판매처럼 물건을
들고 팔러 다닐 필요도 없고, 일일이 사람을 만나러 다닐 필요도 없습니다. 또한 판매원이
아니기 때문에 복잡한 서류를 제출하여 개인 사업자 등록증을 낼 필요도 없습니다. 그럼에도
불구하고 홍보를 열심히 실시한 고객이면 누구나 1년 안에 월 평균 300만원 이상의 수입을
올릴 수 있습니다.</SPAN></P>
<P><SPAN style="FONT-SIZE: 9pt">그런데 세상은 아직도 고정관념이
강해서 이 사실을 믿으려 하지 않습니다. 하지만 모든사람이 고정관념에 사로잡혀 있을 때 그
때가 바로 우리에게는 커다란 기회인 것입니다.</SPAN></P>
<P><SPAN style="FONT-SIZE: 9pt">저희는 결코 피라미드도 다단계
판매도 아닙니다. 인터파크나 삼성몰과 똑같은 인터넷 쇼핑몰입니다. 이들은 단지 자신이
구입한 물품에 대해서만 마일리지를 제공하지만, 저희는 한 걸음 더 나아가서 본인이 홍보하여
가입한 회원이 제품을 구입할 때도 마일리지를 제공한다는 점이 다를
뿐입니다.</SPAN></P>
<P><SPAN style="FONT-SIZE: 9pt">그것은 유명연예인과 광고
관련사에게 쏟아 붓고있는 막대한 광고비를 직접적인 광고 행위자인 고객에게 지급하는,
어디까지나 고객의 홍보 노력에 대한 정당한 보상인 것입니다. 인터파크나 삼성몰이 다단계
판매가 아니듯이 저희도 아닌 것입니다.</SPAN></P>
<P><SPAN style="FONT-SIZE: 9pt">누구나 열심히 홍보노력을 기울이면
하우리더 쇼핑몰을 통해서 행복한 미래를 보장받을 수 있습니다. 평생 인세수입을 향유하며
자녀의 미래까지도 보장이 됩니다. 최소한 시간이 없어서 돈이 없어서 화가나 음악가나
문학가가 되려는 자녀의 꿈을 가로막는 능력없는 부모상은 면할 수 있을
것입니다.</SPAN></P>
<P><SPAN style="FONT-SIZE: 9pt">시간도, 자본도, 재능도 필요없이
누구나 할 수 있는 하우리더 비즈니스에 당신을 초대합니다. 지금 하우리더 쇼핑몰에
가입하시고 이 정보를 주위분들에게 알려주십시오.<BR>회원가입은 무료입니다. 귀하의 소중한
권리를 당신의 제품구입과 전혀 관계없는 유명 연예인들에게 헌납하지 마십시오. 부디 귀하의
현명한 판단을 기원합니다.<BR></SPAN></P>
<P align=center><SPAN style="FONT-SIZE: 9pt"><A
href="http://www.howleader.com/HowHelp/System.html"
target=_blank><FONT color=#3399ff><B>[고객 보상 규정
보기]</B></FONT></A><FONT color=#3399ff><B>
&nbsp;</B></FONT><A
href="http://www.howleader.com/BBS/BBS_Read.html?BBS_Kind=알림사항&amp;SerialNo=53"
target=_blank><B><FONT color=#3399ff>[다단계판매와 다른
점]</FONT></B></A><FONT
color=#3399ff><B><BR><BR></B></FONT><FONT
color=#cc6633><B>지금 가입하시면 5000원을 적립해 드리고 1만원 적립시
계좌로 입금시켜 드립니다.</B></FONT><FONT
color=#3399ff><B><BR></B></FONT>&nbsp;&nbsp;</SPAN></P></TD></TR></TBODY></TABLE>
<p><SPAN
style="FONT-SIZE: 9pt"><BR></SPAN>
</p>
<TABLE cellSpacing=0 cellPadding=0 width=520>
<TBODY>
<TR>
<TD vAlign=center bgColor=#e2ddf3 height=61>
<P align=center><B><SPAN style="FONT-SIZE: 9pt">지금
바로<FONT color=#3399ff> </FONT><A
href="http://www.howleader.com/"
target=_blank><FONT
color=#3399ff>하우리더</FONT></A></SPAN></B><SPAN
style="FONT-SIZE: 9pt"><B><FONT color=#3399ff>
</FONT>쇼핑몰로 가셔서 무료 회원 가입을 하시고 <BR>자신의 신상내용을 변경하여
주위 분들에게 이 정보를 전달하십시오.</B></SPAN><B><FONT
color=black><SPAN style="FONT-SIZE: 9pt"><BR>이것이 이
사업 전개 방식의 전부입니다.</SPAN></FONT><FONT
color=#cc6633><SPAN
style="FONT-SIZE: 9pt"><BR></SPAN></FONT></B></P></TD></TR>
<TBODY>
<TR>
<TD bgColor=#f2f7f7>
<P><SPAN
style="FONT-SIZE: 9pt"><BR>&nbsp;&nbsp;<B>&nbsp; [하우리더
비즈니스]<BR></B><BR>&nbsp;&nbsp;&nbsp;&nbsp;정보 전달자 :
최석훈<BR>&nbsp;&nbsp;&nbsp;&nbsp;이 &nbsp; 메 &nbsp; 일
: acaba2@hitel.net&nbsp;&nbsp; 전&nbsp; 달 자&nbsp;
ID :</SPAN><b><font size="2" color="#FF0000">
acaba2</font></b></P>
<p align="center"><font size="2"><br>귀하의 메일주소는 웹 서핑중에 알게 된 것이며<br>정보통신부 권고 사항에 의거 제목에 <b>[광고]</b>라고 표기한 광고 메일입니다.<br>
이런 메일을 받으셔서 불편하셨다면 다시한번 정중히 사과드립니다.수신을 원치 않으시면<a href="mailto:acaba2@dreamwiz.com?subject=[수신거부]">
[수신거부]</a> 의사를 밝힌 응답메일을 발송해 주십시요.</font></p>
<P><SPAN
style="FONT-SIZE: 9pt">&nbsp;<BR>　</SPAN></P></TD></TR></TBODY></TABLE></TD>
<TD width=5 bgColor=#92d24d><SPAN
style="FONT-SIZE: 9pt">&nbsp;</SPAN></TD></TR>
<TR>
<TD colSpan=3><SPAN style="FONT-SIZE: 9pt">
<IMG
src="http://acaba3.hihome.com/girl05.gif"
border=0 width="561" height="6"></SPAN></TD></TR></TBODY></TABLE></CENTER></TD></TR></TBODY></TABLE></DIV>
</TR>
</TABLE> </BODY></HTML>

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Sat Jan 19 14:27:01 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id OAA21775
	for <tsvwg-archive@odin.ietf.org>; Sat, 19 Jan 2002 14:27:01 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id OAA15318
	for tsvwg-archive@odin.ietf.org; Sat, 19 Jan 2002 14:27:04 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id OAA15086;
	Sat, 19 Jan 2002 14:03:46 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id OAA15054
	for <tsvwg@optimus.ietf.org>; Sat, 19 Jan 2002 14:03:43 -0500 (EST)
Received: from nazare.cin.ufpe.br ([200.249.235.15])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id OAA21581
	for <tsvwg@ietf.org>; Sat, 19 Jan 2002 14:03:40 -0500 (EST)
Received: from jurema.cin.ufpe.br (jurema [172.17.33.106])
	by nazare.cin.ufpe.br (8.11.6/8.11.6) with ESMTP id g0JJ3BH95930
	for <tsvwg@ietf.org>; Sat, 19 Jan 2002 17:03:11 -0200 (EDT)
	(envelope-from kkco@nazare.cin.ufpe.br)
Received: from kkco (helo=localhost)
	by jurema.cin.ufpe.br with local-esmtp (Exim 3.12 #1 (Debian))
	id 16S0li-0000Pi-00
	for <tsvwg@ietf.org>; Sat, 19 Jan 2002 17:03:10 -0200
Date: Sat, 19 Jan 2002 17:03:10 -0200 (EDT)
From: "Karina Karla C. de Oliveira" <kkco@cin.ufpe.br>
To: tsvwg@ietf.org
Message-ID: <Pine.LNX.4.21.0201191658510.901-100000@jurema.cin.ufpe.br>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=ISO-8859-1
Content-Transfer-Encoding: 8bit
X-MIME-Autoconverted: from QUOTED-PRINTABLE to 8bit by optimus.ietf.org id OAA15055
Subject: [Tsvwg] Checksum Simulation.
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 8bit


Hi, People.

I'm doing my master's thesis and I need to simulate the UDP checksum. I
would like to know how is made the checksum's simulation to SCTP.

Some idea is welcome!

Thanks in advance,
Karina Oliveira.

*********************************************************
        Karina Karla Cavalcante de Oliveira
	Computer Science 
        Universidade Federal de Pernambuco
        Centro de Inform�tica - CIn
        Caixa Postal 7851
        CEP: 50732-970
        Recife-PE - Brasil
        Phone: +55 (081) 3271-8430

        Email: kkco@cin.ufpe.br
	WWW: www.cin.ufpe.br/~kkco
	Phones: (HOME) +55 (081) 3338-2054
*********************************************************


_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Mon Jan 21 05:26:06 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id FAA04272
	for <tsvwg-archive@odin.ietf.org>; Mon, 21 Jan 2002 05:26:06 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id FAA29774
	for tsvwg-archive@odin.ietf.org; Mon, 21 Jan 2002 05:26:05 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id FAA29466;
	Mon, 21 Jan 2002 05:11:03 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id FAA29436
	for <tsvwg@optimus.ietf.org>; Mon, 21 Jan 2002 05:11:01 -0500 (EST)
Received: from CNRI.Reston.VA.US (localhost [127.0.0.1])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id FAA03863;
	Mon, 21 Jan 2002 05:10:57 -0500 (EST)
Message-Id: <200201211010.FAA03863@ietf.org>
Mime-Version: 1.0
Content-Type: Multipart/Mixed; Boundary="NextPart"
To: IETF-Announce: ;
Cc: tsvwg@ietf.org
From: Internet-Drafts@ietf.org
Reply-to: Internet-Drafts@ietf.org
Date: Mon, 21 Jan 2002 05:10:56 -0500
Subject: [Tsvwg] I-D ACTION:draft-ietf-tsvwg-sctpcsum-02.txt
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

--NextPart

A New Internet-Draft is available from the on-line Internet-Drafts directories.
This draft is a work item of the Transport Area Working Group Working Group of the IETF.

	Title		: SCTP Checksum Change
	Author(s)	: R. Stewart, J. Stone, D. Otis
	Filename	: draft-ietf-tsvwg-sctpcsum-02.txt
	Pages		: 10
	Date		: 18-Jan-02
	
SCTP [RFC2960] currently uses an Adler-32 checksum.  For small packets 
Adler-32 provides weak detection of errors.  This document changes that 
checksum and updates SCTP to use a 32 bit CRC checksum.

A URL for this Internet-Draft is:
http://www.ietf.org/internet-drafts/draft-ietf-tsvwg-sctpcsum-02.txt

To remove yourself from the IETF Announcement list, send a message to 
ietf-announce-request with the word unsubscribe in the body of the message.

Internet-Drafts are also available by anonymous FTP. Login with the username
"anonymous" and a password of your e-mail address. After logging in,
type "cd internet-drafts" and then
	"get draft-ietf-tsvwg-sctpcsum-02.txt".

A list of Internet-Drafts directories can be found in
http://www.ietf.org/shadow.html 
or ftp://ftp.ietf.org/ietf/1shadow-sites.txt


Internet-Drafts can also be obtained by e-mail.

Send a message to:
	mailserv@ietf.org.
In the body type:
	"FILE /internet-drafts/draft-ietf-tsvwg-sctpcsum-02.txt".
	
NOTE:	The mail server at ietf.org can return the document in
	MIME-encoded form by using the "mpack" utility.  To use this
	feature, insert the command "ENCODING mime" before the "FILE"
	command.  To decode the response(s), you will need "munpack" or
	a MIME-compliant mail reader.  Different MIME-compliant mail readers
	exhibit different behavior, especially when dealing with
	"multipart" MIME messages (i.e. documents which have been split
	up into multiple messages), so check your local documentation on
	how to manipulate these messages.
		
		
Below is the data which will enable a MIME compliant mail reader
implementation to automatically retrieve the ASCII version of the
Internet-Draft.

--NextPart
Content-Type: Multipart/Alternative; Boundary="OtherAccess"

--OtherAccess
Content-Type: Message/External-body;
	access-type="mail-server";
	server="mailserv@ietf.org"

Content-Type: text/plain
Content-ID:	<20020118134526.I-D@ietf.org>

ENCODING mime
FILE /internet-drafts/draft-ietf-tsvwg-sctpcsum-02.txt

--OtherAccess
Content-Type: Message/External-body;
	name="draft-ietf-tsvwg-sctpcsum-02.txt";
	site="ftp.ietf.org";
	access-type="anon-ftp";
	directory="internet-drafts"

Content-Type: text/plain
Content-ID:	<20020118134526.I-D@ietf.org>

--OtherAccess--

--NextPart--



_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Mon Jan 21 06:55:37 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id GAA06302
	for <tsvwg-archive@odin.ietf.org>; Mon, 21 Jan 2002 06:55:32 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id GAA03187
	for tsvwg-archive@odin.ietf.org; Mon, 21 Jan 2002 06:55:35 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id GAA02744;
	Mon, 21 Jan 2002 06:36:12 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id GAA02713
	for <tsvwg@optimus.ietf.org>; Mon, 21 Jan 2002 06:36:10 -0500 (EST)
Received: from mgw-x2.nokia.com (mgw-x2.nokia.com [131.228.20.22])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id GAA05826
	for <tsvwg@ietf.org>; Mon, 21 Jan 2002 06:36:06 -0500 (EST)
From: john.loughney@nokia.com
Received: from esvir02nok.ntc.nokia.com (esvir02nokt.ntc.nokia.com [172.21.143.34])
	by mgw-x2.nokia.com (Switch-2.1.0/Switch-2.1.0) with ESMTP id g0LBaB911050
	for <tsvwg@ietf.org>; Mon, 21 Jan 2002 13:36:11 +0200 (EET)
Received: from esebh002.NOE.Nokia.com (unverified) by esvir02nok.ntc.nokia.com
 (Content Technologies SMTPRS 4.2.5) with ESMTP id <T5895ab83d1ac158f22078@esvir02nok.ntc.nokia.com> for <tsvwg@ietf.org>;
 Mon, 21 Jan 2002 13:36:07 +0200
Received: from esebe004.NOE.Nokia.com ([172.21.138.44]) by esebh002.NOE.Nokia.com with Microsoft SMTPSVC(5.0.2195.3779);
	 Mon, 21 Jan 2002 13:36:07 +0200
content-class: urn:content-classes:message
MIME-Version: 1.0
Content-Type: text/plain;
	charset="iso-8859-1"
X-MimeOLE: Produced By Microsoft Exchange V6.0.5762.3
Date: Mon, 21 Jan 2002 13:36:07 +0200
Message-ID: <0C1353ABB1DEB74DB067ADFF749C4EEF09BC5E@esebe004.NOE.Nokia.com>
Thread-Topic: [Tsvwg] I-D ACTION:draft-ietf-tsvwg-sctpcsum-02.txt
Thread-Index: AcGiZAtJHsuUL9oGTVeAI9nQBUfhHQACvuzA
To: <tsvwg@ietf.org>
X-OriginalArrivalTime: 21 Jan 2002 11:36:07.0681 (UTC) FILETIME=[D121D710:01C1A26F]
Content-Transfer-Encoding: 8bit
X-MIME-Autoconverted: from quoted-printable to 8bit by optimus.ietf.org id GAA02714
Subject: [Tsvwg] Comment on:I-D ACTION:draft-ietf-tsvwg-sctpcsum-02.txt
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 8bit

Hi all,

I have a question & comment on the Security Considerations 

3 Security Considerations

    There may be a computational advantage in validating the Association
    against the Verification Tag prior to performing a checksum as
    invalid tags will result in the same action as a bad checksum in
    most cases. The exceptions for this technique would be INIT and some
    SHUTDOWN-COMPLETE exchanges as well as a stale COOKIE-ECHO.  These
    special case exchanges must represent small packets and will
    minimize the effect of the checksum calculation. ...

I think the above is important, but I do not understand how it relates 
to Security.

    ...   In general, 
    the security considerations of RFC2960 apply to the protocol 
    with the new checksum as well.

I think this is sufficient, as only the algorithm is changing, and
it can be thought of as just bits on the wire.

best regards,
John

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Mon Jan 21 08:30:30 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id IAA08962
	for <tsvwg-archive@odin.ietf.org>; Mon, 21 Jan 2002 08:30:30 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id IAA06838
	for tsvwg-archive@odin.ietf.org; Mon, 21 Jan 2002 08:30:31 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id IAA06160;
	Mon, 21 Jan 2002 08:18:38 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id IAA06132
	for <tsvwg@optimus.ietf.org>; Mon, 21 Jan 2002 08:18:37 -0500 (EST)
Received: from sj-msg-core-4.cisco.com (sj-msg-core-4.cisco.com [171.71.163.10])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id IAA08444
	for <tsvwg@ietf.org>; Mon, 21 Jan 2002 08:18:32 -0500 (EST)
Received: from mira-sjc5-2.cisco.com (mira-sjc5-2.cisco.com [171.71.163.16])
	by sj-msg-core-4.cisco.com (8.11.3/8.9.1) with ESMTP id g0LDI5d06001;
	Mon, 21 Jan 2002 05:18:05 -0800 (PST)
Received: from stewart.chicago.il.us (ssh-sj1.cisco.com [171.68.225.134])
	by mira-sjc5-2.cisco.com (Mirapoint)
	with ESMTP id AAU68042;
	Mon, 21 Jan 2002 05:18:03 -0800 (PST)
Message-ID: <3C4C150B.54AE73A4@stewart.chicago.il.us>
Date: Mon, 21 Jan 2002 07:18:03 -0600
From: Randall Stewart <randall@stewart.chicago.il.us>
X-Mailer: Mozilla 4.76 [en] (X11; U; Linux 2.2.12 i386)
X-Accept-Language: en
MIME-Version: 1.0
To: john.loughney@nokia.com
CC: tsvwg@ietf.org
Subject: Re: [Tsvwg] Comment on:I-D ACTION:draft-ietf-tsvwg-sctpcsum-02.txt
References: <0C1353ABB1DEB74DB067ADFF749C4EEF09BC5E@esebe004.NOE.Nokia.com>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Transfer-Encoding: 7bit
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit

John:

Hmm..  I too thought of this when Doug posted the original
text proposal to the WG.. 

And it is true that I don't think I would check the V-Tag
first in my implementation however there is something to the
fact that a small amount of more overhead processing as
been added to the protocol... i.e. the difference in
CPU cost of doing Adler32 and CRC32c... 

I realize this is a very small amount of difference, but
an implementor MAY want to use this fact to instead lookup
the association first and check the V-Tag before going
to the trouble of computing the CRC...

So I think it is worth mentioning...(i.e. that is why I
did not object to Doug's proposal).

R


john.loughney@nokia.com wrote:
> 
> Hi all,
> 
> I have a question & comment on the Security Considerations
> 
> 3 Security Considerations
> 
>     There may be a computational advantage in validating the Association
>     against the Verification Tag prior to performing a checksum as
>     invalid tags will result in the same action as a bad checksum in
>     most cases. The exceptions for this technique would be INIT and some
>     SHUTDOWN-COMPLETE exchanges as well as a stale COOKIE-ECHO.  These
>     special case exchanges must represent small packets and will
>     minimize the effect of the checksum calculation. ...
> 
> I think the above is important, but I do not understand how it relates
> to Security.
> 
>     ...   In general,
>     the security considerations of RFC2960 apply to the protocol
>     with the new checksum as well.
> 
> I think this is sufficient, as only the algorithm is changing, and
> it can be thought of as just bits on the wire.
> 
> best regards,
> John
> 
> _______________________________________________
> tsvwg mailing list
> tsvwg@ietf.org
> https://www1.ietf.org/mailman/listinfo/tsvwg

-- 
Randall R. Stewart
randall@stewart.chicago.il.us 815-342-5222 (cell phone)

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Mon Jan 21 08:36:33 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id IAA09349
	for <tsvwg-archive@odin.ietf.org>; Mon, 21 Jan 2002 08:36:32 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id IAA07152
	for tsvwg-archive@odin.ietf.org; Mon, 21 Jan 2002 08:36:34 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id IAA06295;
	Mon, 21 Jan 2002 08:21:37 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id IAA06264
	for <tsvwg@optimus.ietf.org>; Mon, 21 Jan 2002 08:21:35 -0500 (EST)
Received: from mgw-x3.nokia.com (mgw-x3.nokia.com [131.228.20.26])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id IAA08544
	for <tsvwg@ietf.org>; Mon, 21 Jan 2002 08:21:30 -0500 (EST)
From: john.loughney@nokia.com
Received: from esvir01nok.ntc.nokia.com (esvir01nokt.ntc.nokia.com [172.21.143.33])
	by mgw-x3.nokia.com (Switch-2.1.0/Switch-2.1.0) with ESMTP id g0LDLfk19423
	for <tsvwg@ietf.org>; Mon, 21 Jan 2002 15:21:41 +0200 (EET)
Received: from esebh001.NOE.Nokia.com (unverified) by esvir01nok.ntc.nokia.com
 (Content Technologies SMTPRS 4.2.5) with ESMTP id <T58960c0552ac158f21082@esvir01nok.ntc.nokia.com>;
 Mon, 21 Jan 2002 15:21:32 +0200
Received: from esebe004.NOE.Nokia.com ([172.21.138.44]) by esebh001.NOE.Nokia.com with Microsoft SMTPSVC(5.0.2195.3779);
	 Mon, 21 Jan 2002 15:21:19 +0200
content-class: urn:content-classes:message
MIME-Version: 1.0
Content-Type: text/plain;
	charset="iso-8859-1"
Subject: RE: [Tsvwg] Comment on:I-D ACTION:draft-ietf-tsvwg-sctpcsum-02.txt
X-MimeOLE: Produced By Microsoft Exchange V6.0.5762.3
Date: Mon, 21 Jan 2002 15:21:18 +0200
Message-ID: <0C1353ABB1DEB74DB067ADFF749C4EEF5D93CA@esebe004.NOE.Nokia.com>
Thread-Topic: [Tsvwg] Comment on:I-D ACTION:draft-ietf-tsvwg-sctpcsum-02.txt
Thread-Index: AcGifiPr0FW9wlf/SS2VfYHKaDp35gAAD17g
To: <randall@stewart.chicago.il.us>
Cc: <tsvwg@ietf.org>
X-OriginalArrivalTime: 21 Jan 2002 13:21:19.0674 (UTC) FILETIME=[835F89A0:01C1A27E]
Content-Transfer-Encoding: 8bit
X-MIME-Autoconverted: from quoted-printable to 8bit by optimus.ietf.org id IAA06265
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 8bit

Hi Randy,

> Hmm..  I too thought of this when Doug posted the original
> text proposal to the WG.. 
> 
> And it is true that I don't think I would check the V-Tag
> first in my implementation however there is something to the
> fact that a small amount of more overhead processing as
> been added to the protocol... i.e. the difference in
> CPU cost of doing Adler32 and CRC32c... 
> 
> I realize this is a very small amount of difference, but
> an implementor MAY want to use this fact to instead lookup
> the association first and check the V-Tag before going
> to the trouble of computing the CRC...
> 
> So I think it is worth mentioning...(i.e. that is why I
> did not object to Doug's proposal).

I think it is worth mentioning too, I'm just suggesting
moving it from the Security Considerations.

John

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Mon Jan 21 10:24:19 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id KAA12287
	for <tsvwg-archive@odin.ietf.org>; Mon, 21 Jan 2002 10:24:19 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id KAA12382
	for tsvwg-archive@odin.ietf.org; Mon, 21 Jan 2002 10:24:21 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id KAA11264;
	Mon, 21 Jan 2002 10:07:20 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id KAA11236
	for <tsvwg@optimus.ietf.org>; Mon, 21 Jan 2002 10:07:18 -0500 (EST)
Received: from mail.tlc.polito.it (IDENT:root@prezzemolo.polito.it [130.192.9.131])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id KAA11665
	for <tsvwg@ietf.org>; Mon, 21 Jan 2002 10:07:14 -0500 (EST)
Received: from polito.it (IDENT:michela@pomodoro.polito.it [130.192.9.133])
	by mail.tlc.polito.it (8.11.6/8.11.6) with ESMTP id g0LF7Gj04260;
	Mon, 21 Jan 2002 16:07:16 +0100
Message-ID: <3C4C2EA4.20AC0AFF@polito.it>
Date: Mon, 21 Jan 2002 16:07:16 +0100
From: Michela Meo <michela@polito.it>
X-Mailer: Mozilla 4.76 [en] (X11; U; Linux 2.2.14 i686)
X-Accept-Language: en
MIME-Version: 1.0
To: tsvwg@ietf.org
CC: Sally Floyd <floyd@aciri.org>, Michela Meo <michela@polito.it>,
        casetti@polito.it, mellia@polito.it
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Transfer-Encoding: 7bit
Subject: [Tsvwg] TCP Smart Framing
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit


Sally Floyd wrote:
>
> Michela -
>
> >To revitalize the discussion on the Smart Framing option, let us
> >summarize
> >the discussion so far:
>
> I have not followed this discussion in complete detail, but I would
> suggest that in summarizing the discussion so far, it would also
> be useful to summarize the relationship of SF with Limited Transmit
> (LT), RFC 3042, which has been Proposed Standard for a year now.

As Sally suggested in her last email, we're listing a few points
that we believe can summarize the relationship between Limited
Transmit and Smart Framing:

* Reaction time: in the small window regime (cwnd <= 4MSS), in a
single-drop scenario, it can take LT between two and three RTTs
before enough dupACKs are available to trigger FR. Conversely,
TCP-SF just needs one RTT before FR can be triggered.

* Small windows: LT does not prevent a timeout if IW=1 or when
TCP restarts with LW=1. TCP-SF always prevents timeouts in single-
loss scenarios, whatever the size of cwnd.

* RTT estimation: TCP-SF's "hidden agenda" includes the refinement
of the RTT estimate (hence possibly shorter RTO) early on in a
connection, thanks to the higher number of RTT samples. LT does not
achieve this.

* no new data: as already pointed out, not enough new data may be
available to keep the ACK clock going, i.e. because the original
file was too short; the position (inside the window)
of the dropped packet may exacerbate this problem. Such shortage
might affect TCP-SF too, though the higher number of segments makes
this
occurence less likely. Of course, if one of TCP-SF's last segments
is dropped, timeout expiration cannot be helped.

As an aside, it's probably a good idea to consider the coexistence of
LT and TCP-SF, whose joint deployment would trigger FR even in
multiple-drop scenarios.

> But instead of using smaller-than-normal packets
> to induce a Fast Retransmit, it might be better to ask the question:

> if you get a single duplicate acknowledgement, and there is no new
> data to send, should you be able to retransmit old data that might
> or might not be lost?  A "yes" answer to this question would take
> care of many of the additional cases where Limited Transmit in its
> current form is not sufficient to prevent a timeout, without
> unnecessarily reducing the packet size, and without increasing the
> number of packets used by all connections when they have small
> windows.

In our opinion, retransmitting old data is dangerously close to
getting rid of timeouts and using single dupACKs as drop
indicators. As Mark Allman himself pointed out, this should
probably be a once-per-connection solution.


> I don't buy it that four small packets presents the same load to
> the network as one larger packet.  Aside from the issue of header
> size, there can be router limitations in terms of packets per second

> processing capacity in addition to the link bandwidth limitation
> in terms of bytes per second.  It is still not clear (to me) which
> limitation, packets per second or bytes per second, is likely to
> dominate some years in the future.

Agreed, the overhead is the price you pay. However, one has to wonder
whether processing capacity is going to be the bottleneck, even in the

near future.


Michela, Claudio and Marco




_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Mon Jan 21 13:40:18 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id NAA19650
	for <tsvwg-archive@odin.ietf.org>; Mon, 21 Jan 2002 13:40:09 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id NAA22501
	for tsvwg-archive@odin.ietf.org; Mon, 21 Jan 2002 13:40:11 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id NAA21906;
	Mon, 21 Jan 2002 13:27:56 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id NAA21877
	for <tsvwg@optimus.ietf.org>; Mon, 21 Jan 2002 13:27:53 -0500 (EST)
Received: from localhost ([218.55.130.23])
	by ietf.org (8.9.1a/8.9.1a) with SMTP id NAA19475
	for <tsvwg@ietf.org>; Mon, 21 Jan 2002 13:27:46 -0500 (EST)
Message-Id: <200201211827.NAA19475@ietf.org>
Reply-To: naracom2002@naver.com
From: (주)나라콤<naracom2002@naver.com>
To: tsvwg@ietf.org
Mime-Version: 1.0
Content-Type: text/html; charset="ks_c_5601-1987"
Date: Tue, 22 Jan 2002 03:42:45 +0900
Subject: [Tsvwg] 안녕하세요[홍보]
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

<html>
<head>
<title>선불요금제 참조자료집</title>
<meta http-equiv="Content-Type" content="text/html; charset=euc-kr">
<meta name="generator" content="Namo WebEditor v4.0">
<style type="text/css">
<!--
td { font-family: "돋움", "돋움체"; font-size: 12px; color: #000000; line-height: 120%}
-->
</style>
</head>
<body text="black" link="blue" vlink="purple" alink="red">
<div>
<table width="620" border="0" cellspacing="2" cellpadding="2" align="center">
<tr bgcolor="#FFFFFF">
<td>
<p>안녕하세요 저호준이에요<br>
잘지내고 계시져?<br>
좋은정보가 이렇게 멜보내 드려요<br>
한번 꼼꼼히 따져보세요 아마 많은도움될거에요<br>
아래에 파일첨부했으니 읽어보시고요<br>
그럼 또 멜드릴께요<br>
</td>
</tr>
</table>
<table width="623" border="0" cellspacing="0" cellpadding="0" align="center">
<tr valign="top">
<td height="218" background="http://www.copyi114.co.kr/nara/img01.gif"> <br>
<br>
<br>
<br>
<br>
<br>
<br>
<br>
<table width="575" border="0" cellspacing="0" cellpadding="0" align="center">
<tr>
<td width="396" height="68"> 선불요금제서비스는 어느날 갑자기 나온 서비스가 아닌, 선진 여러나라에서는
이미 기존에 실시하고는 서비스로 회사의 입장에서는 안정적이며 미납 걱정이 없는 회원유치를 하여 회사의 안정성을 꾀할
수 있으며, 이에따른 고객은 각종 혜택 및 수당을 받는 것입니다.<br>
즉 선불제요금제란 먼저 일정부분의 통신료를 내고 캐쉬백(현금)과 기타 수당을 받는 요금체계이며, 이로인한 혜택을 고객에게
크게는 35%에 해당되는 부분을 다시 현금으로 돌려드리고 있습니다. </td>
<td width="179" height="68" valign="bottom">
<div align="center"><a href="http://www.copyi114.co.kr/nara/news.htm" target="_blank"><img src="http://www.copyi114.co.kr/nara/img16.gif" width="145" height="67" border="0"></a></div>
</td>
</tr>
</table>
</td>
</tr>
<tr>
<td background="http://www.copyi114.co.kr/nara/img02.gif">
<table width="620" border="0" cellspacing="0" cellpadding="0">
<tr>
<td width="17">&nbsp;</td>
<td width="561"> <img src="http://www.copyi114.co.kr/nara/img14.gif" width="175" height="22"><br>
<table width="493" border="0" cellspacing="0" cellpadding="0" align="center">
<tr>
<td><img src="http://www.copyi114.co.kr/nara/img15.gif" width="461" height="91"><br>
1. 악성연체 요금으로 인한 금융 <b><font color="#CC0000">신용불량자의 증가.</font></b>
<br>
2. 인적, 물적으로<b><font color="#CC0000"> 많은 관리비용 필요 </font></b>(요금고지,
채권관리 등). <br>
3. 이동통신 사용자가 사용요금의 인지없이 무분별한 사용으로 인한 <b><font color="#CC0000">과소비
초래</font></b>. <br>
<font color="#0066CC"><br>
<b>이러한, 문제로 인해 세계적으로 이동통신은 선불요금제로 전환되고 있습니다.</b></font> <br>
<font color="#990000">※ 2002년에는 IMT2000 서비스가 시작되며, 그때는 지금
사용요금의 3~4배에 이르는 서비스 요금이 사용될 것이라고 예측하고 있습니다. 통신시장이 개방되었음에도
불구하고, 외국의 거대통신기업들이 지금 잠잠한것은 바로 이 IMT2000 시대을 기다리며, 우리나라 통신시장을
넘보고 있는 것입니다. </font></td>
</tr>
</table>
<img src="http://www.copyi114.co.kr/nara/img09.gif" width="161" height="23"> <br>
<table width="493" border="0" cellspacing="0" cellpadding="0" align="center">
<tr>
<td><img src="http://www.copyi114.co.kr/nara/img10.gif" width="399" height="271"></td>
</tr>
</table>
<b><img src="http://www.copyi114.co.kr/nara/img04.gif" width="231" height="21"></b>
<table border="0" width="496" cellspacing="0" cellpadding="0" align="center">
<tr>
<td width="100%" height="20">: 개인대리점으로 구성<br>
고정된 지역 대리점을 탈피하여 지역에 상관없이 움직이는 대리점으로 구성 </td>
</tr>
<tr>
<td width="100%" height="20">
<table border="0" width="500" bordercolor="#FFFFFF" bordercolorlight="#FFFFFF" bordercolordark="#FFFFFF" bgcolor="#666666" cellspacing="1" cellpadding="6" align="center">
<tr>
<td width="20%" height="30" bordercolorlight="#008000" bordercolordark="#C9F5DD" bgcolor="#A6A6A6">
　</td>
<td width="40%" height="30" bordercolorlight="#008000" bordercolordark="#C9F5DD" bgcolor="#FFFFFF">
<p align="center"><b><font color="#003366">일반대리점 </font></b>
</td>
<td width="40%" height="30" bordercolorlight="#008000" bordercolordark="#C9F5DD" bgcolor="#FFFFFF">
<p align="center"><b><font color="#003366">KTF 선불대리점
</font></b>
</td>
</tr>
<tr>
<td width="20%" height="25" bordercolorlight="#008000" bordercolordark="#C9F5DD" bgcolor="#F6F6F6" align="center">
<p align="center"><font color="#000000"><b>범&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;
위</b></font>
</td>
<td width="40%" height="25" bordercolorlight="#008000" bordercolordark="#C9F5DD" bgcolor="#FFFFFF">지역별
한계</td>
<td width="40%" height="25" bordercolorlight="#008000" bordercolordark="#C9F5DD" bgcolor="#FFFFFF"><font color="#FF0000">지역
제한없이 전국</font></td>
</tr>
<tr>
<td width="20%" height="25" bordercolorlight="#008000" bordercolordark="#C9F5DD" bgcolor="#F6F6F6" align="center"><font color="#000000"><b>시&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;
설</b></font></td>
<td width="40%" height="25" bordercolorlight="#008000" bordercolordark="#C9F5DD" bgcolor="#FFFFFF">매장
임대 (보증금 + 시설비 등)</td>
<td width="40%" height="25" bordercolorlight="#008000" bordercolordark="#C9F5DD" bgcolor="#FFFFFF"><font color="#FF0000">매장
불필요</font></td>
</tr>
<tr>
<td width="20%" height="25" bordercolorlight="#008000" bordercolordark="#C9F5DD" bgcolor="#F6F6F6" align="center"><font color="#000000"><b>담&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;
보</b></font></td>
<td width="40%" height="25" bordercolorlight="#008000" bordercolordark="#C9F5DD" bgcolor="#FFFFFF">단말기
구매를 위해 담보물 제공</td>
<td width="40%" height="25" bordercolorlight="#008000" bordercolordark="#C9F5DD" bgcolor="#FFFFFF"><font color="#FF0000">담보
없음</font></td>
</tr>
<tr>
<td width="20%" height="25" bordercolorlight="#008000" bordercolordark="#C9F5DD" bgcolor="#F6F6F6" align="center"><font color="#000000"><b>관
리 유 지</b></font></td>
<td width="40%" height="25" bordercolorlight="#008000" bordercolordark="#C9F5DD" bgcolor="#FFFFFF">관리
인원 및 관리비 과다 지출</td>
<td width="40%" height="25" bordercolorlight="#008000" bordercolordark="#C9F5DD" bgcolor="#FFFFFF"><font color="#FF0000">없음</font></td>
</tr>
<tr>
<td width="20%" height="25" bordercolorlight="#008000" bordercolordark="#C9F5DD" bgcolor="#F6F6F6" align="center"><font color="#000000"><b>수&nbsp;&nbsp;
수&nbsp;&nbsp; 료</b></font></td>
<td width="40%" height="25" bordercolorlight="#008000" bordercolordark="#C9F5DD" bgcolor="#FFFFFF">관리수수료</td>
<td width="40%" height="25" bordercolorlight="#008000" bordercolordark="#C9F5DD" bgcolor="#FFFFFF"><font color="#FF0000">회원장려금</font></td>
</tr>
<tr>
<td width="20%" height="25" bordercolorlight="#008000" bordercolordark="#C9F5DD" bgcolor="#F6F6F6" align="center"><font color="#000000"><b>시&nbsp;&nbsp;
장&nbsp;&nbsp; 성</b></font></td>
<td width="40%" height="25" bordercolorlight="#008000" bordercolordark="#C9F5DD" bgcolor="#FFFFFF">시장성이
없다. (포화상태)</td>
<td width="40%" height="25" bordercolorlight="#008000" bordercolordark="#C9F5DD" bgcolor="#FFFFFF"><font color="#FF0000">무한한
시장성 (초기시장 선점)</font></td>
</tr>
<tr>
<td width="20%" height="25" bordercolorlight="#008000" bordercolordark="#C9F5DD" bgcolor="#F6F6F6" align="center"><font color="#000000"><b>해&nbsp;&nbsp;
지&nbsp;&nbsp; 율</b></font></td>
<td width="40%" height="25" bordercolorlight="#008000" bordercolordark="#C9F5DD" bgcolor="#FFFFFF">고객
이동이 많다.<br>
<br>
(대리점간 공용시스템)</td>
<td width="40%" height="25" bordercolorlight="#008000" bordercolordark="#C9F5DD" bgcolor="#FFFFFF"><font color="#FF0000">고객이동
없음.<br>
<br>
(전용시스템)</font></td>
</tr>
</table>
<table border="0" width="501" align="center">
<tr>
<td width="100%"><img border="0" src="http://www.copyi114.co.kr/nara/img05.gif" width="147" height="38">&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;
<img border="0" src="http://www.copyi114.co.kr/nara/img06.gif" width="95" height="18"></td>
</tr>
<tr>
<td width="100%" height="48">
<p> &nbsp;<font color="#008000">▶ </font><b>사용이 불편하다</b>
: 접속번호와&nbsp; 카드번호를 입력한 후 사용하여야 통화가 연결됩니다. <br>
&nbsp;<font color="#008000">▶ </font><b>시외전화</b> :
기업, 회사에서 많이 사용 (약 2조원)&nbsp; <font color="#0000FF">개인은
1가구당 약 4,600원 사용</font> <br>
&nbsp;<font color="#008000">▶ </font><b>국제전화</b> :
기업, 무역회사, 오파상등이 많이 사용 (약1조원) <font color="#0000FF">개인
사용은 미비하다.</font>
</td>
</tr>
<tr>
<td width="100%"></td>
</tr>
<tr>
<td width="100%"><img border="0" src="http://www.copyi114.co.kr/nara/img07.gif" width="180" height="37">&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;
<img border="0" src="http://www.copyi114.co.kr/nara/img08.gif" width="125" height="18">
</td>
</tr>
<tr valign="top">
<td width="100%" height="67">&nbsp; &nbsp;<font color="#008000">▶
</font><b><font color="#FF0000">사용이 편리하다.</font></b><font color="#0000FF">
: </font><font color="#000000">후불 전화 사용과 동일</font><font color="#0000FF">
<br>
&nbsp;</font><font color="#008000">▶ </font><b><font color="#FF0000">2,750만명이
넘는 대부분의 고객이 개인으로 구성되어 있습니다.</font></b><font color="#0000FF">
<br>
&nbsp;</font><font color="#008000">▶</font><font color="#0000FF">
</font><font color="#FF0000"><b> 선불시장은 무한합니다.</b></font><font color="#0000FF">
: </font><font color="#000000">2750만 후불요금 사용고객</font><font color="#0000FF">
<br>
&nbsp;</font><font color="#008000">▶</font><font color="#0000FF">
</font><b><font color="#FF0000"> 1가구당 11만원 이상 사용합니다.
</font><font color="#000000">(1999년 통계)</font></b></td>
</tr>
</table>
</td>
</tr>
</table>
<img src="http://www.copyi114.co.kr/nara/img18.gif" width="86" height="18">
<table width="480" border="0" cellspacing="2" cellpadding="2" align="center">
<tr>
<td height="305">
<table border="0" width="480" bgcolor="#000000" cellspacing="1" cellpadding="3">
<tr>
<td width="100" bgcolor="#CCCCCC" align="center" height="30" rowspan="2"><b><font color="#666666">요금상품</font></b></td>
<td width="100" bgcolor="#CCCCCC" align="center" height="30" rowspan="2"><b><font color="#666666">일
기본료</font></b></td>
<td width="100" bgcolor="#CCCCCC" align="center" height="30" rowspan="2"><b><font color="#666666">월
기본료</font></b></td>
<td width="20%" bgcolor="#CCCCCC" align="center" height="15" colspan="3"><b><font color="#666666">국내통화료&nbsp;<br>
(원/10초)</font></b></td>
<td width="100" bgcolor="#CCCCCC" align="center" height="30" rowspan="2"><b><font color="#666666">비
고</font></b></td>
</tr>
<tr>
<td width="50" bgcolor="#CCCCCC" align="center"><b><font color="#666666">평상</font></b></td>
<td width="50" bgcolor="#CCCCCC" align="center"><b><font color="#666666">할인</font></b></td>
<td width="50" bgcolor="#CCCCCC" align="center"><b><font color="#666666">심야</font></b></td>
</tr>
<tr>
<td width="100" bgcolor="#F7F7F7" height="30" align="center"><b><font color="#000000">선불
표준</font></b></td>
<td width="100" bgcolor="#FFFFFF" height="30" align="center"><b>533
원</b></td>
<td width="100" bgcolor="#FFFFFF" height="30" align="center"><b>16,000
원</b></td>
<td width="50" bgcolor="#FFFFFF" align="center"><b>18
원</b></td>
<td width="50" bgcolor="#FFFFFF" align="center"><b>15
원</b></td>
<td width="50" bgcolor="#FFFFFF" align="center"><b>10
원</b></td>
<td width="100" bgcolor="#FFFFFF" height="30" align="center">　</td>
</tr>
<tr>
<td width="100" bgcolor="#F7F7F7" height="30" align="center"><b><font color="#000000">선불
에이스</font></b></td>
<td width="100" bgcolor="#FFFFFF" height="30" align="center"><b>616
원</b></td>
<td width="100" bgcolor="#FFFFFF" height="30" align="center"><b>18,500
원</b></td>
<td width="50" bgcolor="#FFFFFF" align="center"><b>15
원</b></td>
<td width="50" bgcolor="#FFFFFF" align="center"><b>14
원</b></td>
<td width="50" bgcolor="#FFFFFF" align="center"><b>13
원</b></td>
<td width="100" bgcolor="#FFFFFF" height="30" align="center">　</td>
</tr>
<tr>
<td width="100" bgcolor="#F7F7F7" height="30" align="center"><b><font color="#000000">선불프리
200</font></b></td>
<td width="100" bgcolor="#FFFFFF" height="30" align="center"><b>1,183
원</b></td>
<td width="100" bgcolor="#FFFFFF" height="30" align="center"><b>35,500
원</b></td>
<td width="50" bgcolor="#FFFFFF" align="center"><b>17
원</b></td>
<td width="50" bgcolor="#FFFFFF" align="center"><b>14
원</b></td>
<td width="50" bgcolor="#FFFFFF" align="center"><b>10
원</b></td>
<td width="100" bgcolor="#FFFFFF" height="30" align="center"><font color="#FF0000"><b>200분
무료</b></font></td>
</tr>
</table>
<center>
</center>
<p align="left">&nbsp; ▶ 요금상품 변경시 적용시점 : 변경과 동시에 변경요금으로 정산되나,
월 1회에 한함. <br>
&nbsp; ▶ <font color="#008000"><b>할 인 시 간 <br>
&nbsp;&nbsp;&nbsp; </b></font><font color="#800000">1) 선불표준,
선불에이스 요금의 경우</font><font color="#000000"> <br>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; - 평일 (토요일 포함) : 0 ~
6 시 (심야) / 6 ~ 12 (할인) / 12 ~ 24 (평상) &nbsp;&nbsp;&nbsp;&nbsp;<br>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; - 공휴일 (일요일 포함) : 0
~ 6시 (심야) / 6 ~ 24 (할인) <br>
&nbsp;&nbsp;&nbsp;&nbsp; </font><font color="#800000">2)
선불프리 200 요금의 경우</font><font color="#000000"><br>
</font><font color="#008000"><b> &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;
</b></font><font color="#000000">- 평일 (토요일 포함) : 0 ~ 6 시
(심야) / 6 ~ 8, 21 ~ 24 (할인) / 8 ~ 21 (평상)<br>
&nbsp;&nbsp;&nbsp;&nbsp; &nbsp;
- 공휴일 (일요일 포함) : 0 ~ 6시 (심야) / 6 ~ 24 (할인)</font>
</td>
</tr>
</table>
<img src="http://www.copyi114.co.kr/nara/img12.gif" width="144" height="20"><br>
<table width="480" border="0" cellspacing="0" cellpadding="0" align="center">
<tr>
<td><img src="http://www.copyi114.co.kr/nara/img13.gif" width="499" height="359">
<table border="0" width="496" bgcolor="#990000" cellspacing="1" cellpadding="3" align="center">
<tr>
<td width="487" bgcolor="#FFFFFF"> &nbsp;<font color="#008080"><b>KTF</b></font><font color="#000000">(한국통신프리텔)의
탄생 당시부터 모든 시스템을 담당해 온 &nbsp;<b>대우정보시스템</b>은&nbsp; <br>
&nbsp;기존 전자상거래의 기반부재로 오는 적자요인을&nbsp;<b>주식회사 나라콤</b>의
네트워크 마케팅을 &nbsp;<br>
&nbsp;통하여 </font><font color="#FF0000"><b>최단기간 최대회원</b></font><font color="#000000">의</font><font color="#FF0000"><b>&nbsp;기반확보</b></font><font color="#000000">를</font><font color="#FF0000"><b>
목표</b></font><font color="#000000">로&nbsp; <b>KTF, 대우정보시스템,&nbsp;주식회사
<br>
&nbsp;나라콤의</b> </font><b><font color="#FF0000">전략적 제휴</font></b><font color="#000000">를
통하여&nbsp;<br>
</font> 선진국형 선불요금제도
<div align="center">- <b><font color="#008080">KTF</font><font color="#000000">
016</font></b> 선불요금제도 -</div>
<div align="right"><font color="#000000">를&nbsp; 탄생시켰습니다.</font><font color="#000080">
</font> </div>
</td>
</tr>
</table>
</td>
</tr>
</table>
<img src="http://www.copyi114.co.kr/nara/img17.gif" width="262" height="21"> <br>
<table width="499" border="0" cellspacing="0" cellpadding="0" align="center">
<tr>
E-mail : <a href="mailto:corud2@dreamwiz.com">corud2@dreamwiz.com
</tr>
</table>
</td>
<td width="42">&nbsp;</td>
</tr>
</table>
</td>
</tr>
<tr>
<td><img src="http://www.copyi114.co.kr/nara/img03.gif" width="619" height="31"></td>
</tr>
<tr>
<td>
<div align="center"><img height=10 src="http://www.enativespeaker.com/image/139-1.gif"
width=10> <font size=2>귀하의 E-MAIL은 관련 게시판을 통해 알게 되었습니다. <br>
원치 않았던 메일이라면 </font><a
href="mailto:naracom2002@hitel.net?subject=수신거부"><font size=2><b><font
color=#0000ff>여기</font></b></font></a><font size=2>를 눌러 알려 주시면 앞으로는 발송되지 않도록 하겠습니다.</font>
<font
size=2></font></div>
</td>
</tr>
</table>
<br>
</div>
</body>
</html>

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Mon Jan 21 15:42:26 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id PAA23587
	for <tsvwg-archive@odin.ietf.org>; Mon, 21 Jan 2002 15:42:21 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id PAA26636
	for tsvwg-archive@odin.ietf.org; Mon, 21 Jan 2002 15:42:21 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id PAA26077;
	Mon, 21 Jan 2002 15:29:28 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id PAA26048
	for <tsvwg@optimus.ietf.org>; Mon, 21 Jan 2002 15:29:26 -0500 (EST)
Received: from ftpbox.mot.com (ftpbox.mot.com [129.188.136.101])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id PAA23023
	for <tsvwg@ietf.org>; Mon, 21 Jan 2002 15:29:22 -0500 (EST)
Received: [from mothost.mot.com (mothost.mot.com [129.188.137.101]) by ftpbox.mot.com (ftpbox 2.1) with ESMTP id NAA08508; Mon, 21 Jan 2002 13:29:25 -0700 (MST)]
Received: [from relay2.cig.mot.com (relay2.cig.mot.com [136.182.15.24]) by mothost.mot.com (MOT-mothost 2.0) with ESMTP id NAA12571; Mon, 21 Jan 2002 13:29:24 -0700 (MST)]
Received: from quantix20.cig.mot.com (quantix20 [160.47.58.31])
	by relay2.cig.mot.com (8.11.4/8.11.4) with ESMTP id g0LKTM710112;
	Mon, 21 Jan 2002 14:29:23 -0600 (CST)
Received: from cig.mot.com (d50-384a.cig.mot.com [160.47.56.74])
	by quantix20.cig.mot.com (8.8.8+Sun/8.8.8) with ESMTP id OAA04175;
	Mon, 21 Jan 2002 14:34:09 -0600 (CST)
Message-ID: <3C4C7AF2.46C94D30@cig.mot.com>
Date: Mon, 21 Jan 2002 14:32:50 -0600
From: Qiaobing Xie <xieqb@cig.mot.com>
X-Mailer: Mozilla 4.77 [en] (Win98; U)
X-Accept-Language: en
MIME-Version: 1.0
To: john.loughney@nokia.com
CC: randall@stewart.chicago.il.us, tsvwg@ietf.org
Subject: Re: [Tsvwg] Comment on:I-D ACTION:draft-ietf-tsvwg-sctpcsum-02.txt
References: <0C1353ABB1DEB74DB067ADFF749C4EEF5D93CA@esebe004.NOE.Nokia.com>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Transfer-Encoding: 7bit
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit

In certain cases, the alternative approach proposed by Doug may enhance
the endpoint's resistance against flooding attacks - you may need less
computation to discard an attacking packet by checking the V-tag first.
I don't know whether this is the original intention of Doug when he put
this in the security section, though.

-Qiaobing

john.loughney@nokia.com wrote:
> 
> Hi Randy,
> 
> > Hmm..  I too thought of this when Doug posted the original
> > text proposal to the WG..
> >
> > And it is true that I don't think I would check the V-Tag
> > first in my implementation however there is something to the
> > fact that a small amount of more overhead processing as
> > been added to the protocol... i.e. the difference in
> > CPU cost of doing Adler32 and CRC32c...
> >
> > I realize this is a very small amount of difference, but
> > an implementor MAY want to use this fact to instead lookup
> > the association first and check the V-Tag before going
> > to the trouble of computing the CRC...
> >
> > So I think it is worth mentioning...(i.e. that is why I
> > did not object to Doug's proposal).
> 
> I think it is worth mentioning too, I'm just suggesting
> moving it from the Security Considerations.
> 
> John
> 
> _______________________________________________
> tsvwg mailing list
> tsvwg@ietf.org
> https://www1.ietf.org/mailman/listinfo/tsvwg

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@ns.ietf.org  Mon Jan 21 16:56:00 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id QAA26271
	for <tsvwg-archive@odin.ietf.org>; Mon, 21 Jan 2002 16:55:56 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id QAA28508
	for tsvwg-archive@odin.ietf.org; Mon, 21 Jan 2002 16:55:58 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id QAA28224;
	Mon, 21 Jan 2002 16:42:22 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id QAA28195
	for <tsvwg@ns.ietf.org>; Mon, 21 Jan 2002 16:42:20 -0500 (EST)
Received: from sj-msg-core-2.cisco.com (sj-msg-core-2.cisco.com [171.69.24.11])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id QAA25772
	for <tsvwg@ietf.org>; Mon, 21 Jan 2002 16:42:17 -0500 (EST)
Received: from mira-sjc5-2.cisco.com (mira-sjc5-2.cisco.com [171.71.163.16])
	by sj-msg-core-2.cisco.com (8.11.3/8.9.1) with ESMTP id g0LLfoX11641;
	Mon, 21 Jan 2002 13:41:50 -0800 (PST)
Received: from stewart.chicago.il.us (ssh-sj1.cisco.com [171.68.225.134])
	by mira-sjc5-2.cisco.com (Mirapoint)
	with ESMTP id AAU80712;
	Mon, 21 Jan 2002 13:41:49 -0800 (PST)
Message-ID: <3C4C8B1C.C214AF02@stewart.chicago.il.us>
Date: Mon, 21 Jan 2002 15:41:48 -0600
From: Randall Stewart <randall@stewart.chicago.il.us>
X-Mailer: Mozilla 4.76 [en] (X11; U; Linux 2.2.12 i386)
X-Accept-Language: en
MIME-Version: 1.0
To: Qiaobing Xie <xieqb@cig.mot.com>
CC: john.loughney@nokia.com, tsvwg@ietf.org
Subject: Re: [Tsvwg] Comment on:I-D ACTION:draft-ietf-tsvwg-sctpcsum-02.txt
References: <0C1353ABB1DEB74DB067ADFF749C4EEF5D93CA@esebe004.NOE.Nokia.com> <3C4C7AF2.46C94D30@cig.mot.com>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Transfer-Encoding: 7bit
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit

This is exactly what Doug meant (at least I think) :)

R

Qiaobing Xie wrote:
> 
> In certain cases, the alternative approach proposed by Doug may enhance
> the endpoint's resistance against flooding attacks - you may need less
> computation to discard an attacking packet by checking the V-tag first.
> I don't know whether this is the original intention of Doug when he put
> this in the security section, though.
> 
> -Qiaobing
> 
> john.loughney@nokia.com wrote:
> >
> > Hi Randy,
> >
> > > Hmm..  I too thought of this when Doug posted the original
> > > text proposal to the WG..
> > >
> > > And it is true that I don't think I would check the V-Tag
> > > first in my implementation however there is something to the
> > > fact that a small amount of more overhead processing as
> > > been added to the protocol... i.e. the difference in
> > > CPU cost of doing Adler32 and CRC32c...
> > >
> > > I realize this is a very small amount of difference, but
> > > an implementor MAY want to use this fact to instead lookup
> > > the association first and check the V-Tag before going
> > > to the trouble of computing the CRC...
> > >
> > > So I think it is worth mentioning...(i.e. that is why I
> > > did not object to Doug's proposal).
> >
> > I think it is worth mentioning too, I'm just suggesting
> > moving it from the Security Considerations.
> >
> > John
> >
> > _______________________________________________
> > tsvwg mailing list
> > tsvwg@ietf.org
> > https://www1.ietf.org/mailman/listinfo/tsvwg
> 
> _______________________________________________
> tsvwg mailing list
> tsvwg@ietf.org
> https://www1.ietf.org/mailman/listinfo/tsvwg

-- 
Randall R. Stewart
randall@stewart.chicago.il.us 815-342-5222 (cell phone)

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@ns.ietf.org  Mon Jan 21 17:57:27 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id RAA27401
	for <tsvwg-archive@odin.ietf.org>; Mon, 21 Jan 2002 17:57:22 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id RAA00408
	for tsvwg-archive@odin.ietf.org; Mon, 21 Jan 2002 17:57:25 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id RAA00292;
	Mon, 21 Jan 2002 17:46:26 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id RAA00256
	for <tsvwg@ns.ietf.org>; Mon, 21 Jan 2002 17:46:24 -0500 (EST)
Received: from penguin-ext.wise.edt.ericsson.se (penguin-ext.wise.edt.ericsson.se [194.237.142.110])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id RAA27307
	for <tsvwg@ietf.org>; Mon, 21 Jan 2002 17:46:20 -0500 (EST)
Received: from eed.ericsson.se (mailhost.eed.ericsson.se [164.48.133.33])
	by penguin.wise.edt.ericsson.se (8.11.0/8.11.0/WIREfire-1.3) with ESMTP id g0LMkNw01184;
	Mon, 21 Jan 2002 23:46:23 +0100 (MET)
Received: from res0010384da36.ericsson.com (rac184pc.edd.ericsson.se [164.48.123.184])
	by eed.ericsson.se (8.8.8+Sun/1.1.mit) with ESMTP id XAA29947;
	Mon, 21 Jan 2002 23:46:16 +0100 (MET)
Message-Id: <5.1.0.14.0.20020121230442.01b29758@mailhost.eed.ericsson.se>
X-Sender: eedrel@mailhost.eed.ericsson.se
X-Mailer: QUALCOMM Windows Eudora Version 5.1
Date: Mon, 21 Jan 2002 23:43:58 +0100
To: Kacheong Poon <kcpoon@eng.sun.com>
From: Reiner Ludwig <Reiner.Ludwig@ericsson.com>
Subject: Re: [Tsvwg] Eifel draft and usage of time-stamps. 
Cc: tsvwg@ietf.org
In-Reply-To: <Roam.SIMC.2.0.6.1011060317.29073.kcpoon@eng.sun.com>
References: <"Your message with ID" <200201122237.RAA04693@lawyers.grc.nasa.gov>
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"; format=flowed
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

sorry for joining this discussion so late ...

Some comments to this an previous mails ...

At 03:05 15.01.2002, Kacheong Poon wrote:
> > I do not understand this comment.  Eifel is pretty robust, it seems
> > to me.  But, regardless I haven't heard of any failure cases --
> > i.e., when it says there was a spurous timeout and there was not.
> > Are there some?
>
>I believe I came across a mail (reading mail after long vacation
>is a pain...) pointing out that it is not.  I guess Reiner can
>provide us with some data on his experience with Eifel.

I thought the Eifel algorithm was sufficiently robust in detecting spurious 
retransmits. However, Pasi Sarolahti found this case where the Eifel 
algorithm fails due to what appears to be a flaw in the current RFC1323. It 
also seems that there had been unfinished efforts [update to RFC1323 by 
Jacobson, Braden, Borman Feb. 97] that (among other things, I guess) fixed 
that part of RFC1323.

So, as it stands, I either require this fix in an update to the Eifel 
algorithm ID, or somebody rethinks RFC1323. I know that there are a number 
of lingering issues with it. So, maybe this is (again) the time.

Concerning a TCP sender response to a detected spurious timeout, I agree 
with Mark that reverting CC state needs to be combined with "adapting" the 
RTO. I would view it as a form of "speeding ticket" because the TCP sender 
(maybe) maintained a too aggressive RTO.

About reverting CC state after detecting a spurious timeout, don't forget 
that today's TCP (without Eifel) is probably the most aggressive choice. 
This is because today's TCP will forget about the flight of packets sitting 
somewhere in the network after a spurious timeout. Then it slow starts the 
(often unnecessary) go-back-N retransmits into the network thereby breaking 
'packet conservation'. Andrei has some very ugly ns2 TCP-Tahoe traces that 
show how a single spurious timeout causes TCP-Tahoe to do one unnecessary 
go-back-N after the other for quite a while. Maybe TCP-Tahoe doesn't even 
ensure network stability?

///Reiner



_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@ns.ietf.org  Mon Jan 21 18:25:28 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id SAA27992
	for <tsvwg-archive@odin.ietf.org>; Mon, 21 Jan 2002 18:25:27 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id SAA01399
	for tsvwg-archive@odin.ietf.org; Mon, 21 Jan 2002 18:25:31 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id SAA01208;
	Mon, 21 Jan 2002 18:14:48 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id SAA01180
	for <tsvwg@ns.ietf.org>; Mon, 21 Jan 2002 18:14:46 -0500 (EST)
Received: from CNRI.Reston.VA.US (localhost [127.0.0.1])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id SAA27749;
	Mon, 21 Jan 2002 18:14:42 -0500 (EST)
Message-Id: <200201212314.SAA27749@ietf.org>
To: IETF-Announce: ;
Cc: tsvwg@ietf.org
From: The IESG <iesg-secretary@ietf.org>
Reply-to: iesg@ietf.org
Date: Mon, 21 Jan 2002 18:14:42 -0500
Subject: [Tsvwg] Last Call: SCTP Checksum Change to Proposed Standard
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org


The IESG has received a request from the Transport Area Working Group
Working Group to consider SCTP Checksum Change
<draft-ietf-tsvwg-sctpcsum-02.txt> as a Proposed Standard.

The IESG plans to make a decision in the next few weeks, and solicits
final comments on this action.  Please send any comments to the
iesg@ietf.org or ietf@ietf.org mailing lists by February 4, 2002.

Files can be obtained via
http://www.ietf.org/internet-drafts/draft-ietf-tsvwg-sctpcsum-02.txt



_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@ns.ietf.org  Mon Jan 21 19:39:37 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id TAA28993
	for <tsvwg-archive@odin.ietf.org>; Mon, 21 Jan 2002 19:39:37 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id TAA04238
	for tsvwg-archive@odin.ietf.org; Mon, 21 Jan 2002 19:39:39 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id TAA03695;
	Mon, 21 Jan 2002 19:28:24 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id TAA03668
	for <tsvwg@ns.ietf.org>; Mon, 21 Jan 2002 19:28:22 -0500 (EST)
Received: from c003.snv.cp.net (c003-h016.c003.snv.cp.net [209.228.32.230])
	by ietf.org (8.9.1a/8.9.1a) with SMTP id TAA28890
	for <tsvwg@ietf.org>; Mon, 21 Jan 2002 19:28:19 -0500 (EST)
Received: (cpmta 3965 invoked from network); 21 Jan 2002 16:27:50 -0800
Received: from 64.130.130.105 (HELO dougrmt)
  by smtp.telocity.com (209.228.32.230) with SMTP; 21 Jan 2002 16:27:50 -0800
X-Sent: 22 Jan 2002 00:27:50 GMT
From: "Douglas Otis" <dotis@sanlight.net>
To: "Randall Stewart" <randall@stewart.chicago.il.us>,
        "Qiaobing Xie" <xieqb@cig.mot.com>
Cc: <john.loughney@nokia.com>, <tsvwg@ietf.org>
Subject: RE: [Tsvwg] Comment on:I-D ACTION:draft-ietf-tsvwg-sctpcsum-02.txt
Date: Mon, 21 Jan 2002 16:28:46 -0800
Message-ID: <NEBBJGDMMLHHCIKHGBEJOEANCPAA.dotis@sanlight.net>
MIME-Version: 1.0
Content-Type: text/plain;
	charset="us-ascii"
Content-Transfer-Encoding: 7bit
X-Priority: 3 (Normal)
X-MSMail-Priority: Normal
X-Mailer: Microsoft Outlook IMO, Build 9.0.2416 (9.0.2910.0)
Importance: Normal
In-Reply-To: <3C4C8B1C.C214AF02@stewart.chicago.il.us>
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4807.1700
Content-Transfer-Encoding: 7bit
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit

Randall,

I considered an attack designed to consume CPU cycles an aspect of security.
Perhaps this is too broad of an interpretation of what belongs within a
security section.  The method used to verify validity of a packet may
require some considerations to minimize resources (CPU cycles) consumed by
an attack.  As both the V-tag and the Checksum fields must be verified,
should the packet become rejected by the V-tag, there would be no need to
continue further validation.  It is likely an attacker would create a valid
Checksum in a large packet, but would be unable to create valid V-tags but
for exceptions listed with their incumbent overhead advantages as per the
RFC2960 design.  Checksum failures should not represent a high overhead
within valid associations.

I did not intend to suggest how implementations should validate packets or
that this approach would be elegant, only that there may be a computational
advantage in doing so.  That would depend on the relative speed of the
network and the CPU.  Those writing code will likely wish to get Checksum
operations done prior to the V-tag examination to ensure no state is created
as a result of injected errors.  That is a valid consideration and is a
general implementation design based on simplicity.  I would not wish anyone
to think of the security statement as anything but a heads-up for those
systems that may find themselves CPU starved as there are a few extra cycles
added in this checking scheme that may otherwise weaken their resistance.

Doug

> -----Original Message-----
> From: tsvwg-admin@ietf.org [mailto:tsvwg-admin@ietf.org]On Behalf Of
> Randall Stewart
> Sent: Monday, January 21, 2002 1:42 PM
> To: Qiaobing Xie
> Cc: john.loughney@nokia.com; tsvwg@ietf.org
> Subject: Re: [Tsvwg] Comment on:I-D
> ACTION:draft-ietf-tsvwg-sctpcsum-02.txt
>
>
> This is exactly what Doug meant (at least I think) :)
>
> R
>
> Qiaobing Xie wrote:
> >
> > In certain cases, the alternative approach proposed by Doug may enhance
> > the endpoint's resistance against flooding attacks - you may need less
> > computation to discard an attacking packet by checking the V-tag first.
> > I don't know whether this is the original intention of Doug when he put
> > this in the security section, though.
> >
> > -Qiaobing
> >
> > john.loughney@nokia.com wrote:
> > >
> > > Hi Randy,
> > >
> > > > Hmm..  I too thought of this when Doug posted the original
> > > > text proposal to the WG..
> > > >
> > > > And it is true that I don't think I would check the V-Tag
> > > > first in my implementation however there is something to the
> > > > fact that a small amount of more overhead processing as
> > > > been added to the protocol... i.e. the difference in
> > > > CPU cost of doing Adler32 and CRC32c...
> > > >
> > > > I realize this is a very small amount of difference, but
> > > > an implementor MAY want to use this fact to instead lookup
> > > > the association first and check the V-Tag before going
> > > > to the trouble of computing the CRC...
> > > >
> > > > So I think it is worth mentioning...(i.e. that is why I
> > > > did not object to Doug's proposal).
> > >
> > > I think it is worth mentioning too, I'm just suggesting
> > > moving it from the Security Considerations.
> > >
> > > John
> > >
> > > _______________________________________________
> > > tsvwg mailing list
> > > tsvwg@ietf.org
> > > https://www1.ietf.org/mailman/listinfo/tsvwg
> >
> > _______________________________________________
> > tsvwg mailing list
> > tsvwg@ietf.org
> > https://www1.ietf.org/mailman/listinfo/tsvwg
>
> --
> Randall R. Stewart
> randall@stewart.chicago.il.us 815-342-5222 (cell phone)
>
> _______________________________________________
> tsvwg mailing list
> tsvwg@ietf.org
> https://www1.ietf.org/mailman/listinfo/tsvwg
>


_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@ns.ietf.org  Mon Jan 21 22:40:40 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id WAA02852
	for <tsvwg-archive@odin.ietf.org>; Mon, 21 Jan 2002 22:40:40 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id WAA09810
	for tsvwg-archive@odin.ietf.org; Mon, 21 Jan 2002 22:40:44 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id WAA09561;
	Mon, 21 Jan 2002 22:30:43 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id WAA09511
	for <tsvwg@ns.ietf.org>; Mon, 21 Jan 2002 22:30:39 -0500 (EST)
Received: from postino8.int.prima.com.ar (postino8.prima.com.ar [200.42.0.179])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id WAA01989
	for <tsvwg@ietf.org>; Mon, 21 Jan 2002 22:30:34 -0500 (EST)
Received: from ciudad.com.ar (a200042120208.rev.prima.com.ar [200.42.120.208])
	by postino8.int.prima.com.ar (8.11.6/8.11.6) with SMTP id g0M3USI05918
	for <tsvwg@ietf.org>; Tue, 22 Jan 2002 00:30:29 -0300 (ART)
	(envelope-from aschreiber@ciudad.com.ar)
Message-Id: <200201220330.g0M3USI05918@postino8.int.prima.com.ar>
From: "RED DE INTERCAMBIO" <aschreiber@ciudad.com.ar>
To: <tsvwg@ietf.org>
Mime-Version: 1.0
Content-Type: text/plain; charset="ISO-8859-1"
Date: Tue, 22 Jan 2002 00:29:46 -0300
Reply-To: "RED DE INTERCAMBIO" <schreiber@hotpop.com>
Content-Transfer-Encoding: 8bit
Content-Transfer-Encoding: 8bit
Subject: [Tsvwg] (no subject)
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 8bit


_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@ns.ietf.org  Mon Jan 21 23:14:40 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id XAA03275
	for <tsvwg-archive@odin.ietf.org>; Mon, 21 Jan 2002 23:14:39 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id XAA10732
	for tsvwg-archive@odin.ietf.org; Mon, 21 Jan 2002 23:14:42 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id XAA10598;
	Mon, 21 Jan 2002 23:04:09 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id XAA10529
	for <tsvwg@ns.ietf.org>; Mon, 21 Jan 2002 23:04:05 -0500 (EST)
Received: from lawyers.grc.nasa.gov (d8.as0.clev.oh.voyager.net [209.81.165.9])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id XAA03160
	for <tsvwg@ietf.org>; Mon, 21 Jan 2002 23:03:59 -0500 (EST)
Received: from lawyers (mallman@localhost)
	by lawyers.grc.nasa.gov (8.9.3/8.9.3) with ESMTP id WAA31645;
	Mon, 21 Jan 2002 22:49:54 -0500
Message-Id: <200201220349.WAA31645@lawyers.grc.nasa.gov>
To: Michela Meo <michela@polito.it>
From: Mark Allman <mallman@grc.nasa.gov>
Reply-To: mallman@grc.nasa.gov
cc: tsvwg@ietf.org, Sally Floyd <floyd@aciri.org>, casetti@polito.it,
        mellia@polito.it
Subject: Re: [Tsvwg] TCP Smart Framing 
Organization: BBN Technologies/NASA GRC
Song-of-the-Day: Haircut
Date: Mon, 21 Jan 2002 22:49:54 -0500
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org


> * Small windows: LT does not prevent a timeout if IW=1 or when TCP
> restarts with LW=1. TCP-SF always prevents timeouts in single-
> loss scenarios, whatever the size of cwnd.

Maybe I am missing something here, but why not just use a larger
initial window per RFC 2414 (and soon an updated standards track
version, I believe)?

As for the LW.  Well, this is bogus.  I don't know how else to put
it...

My own hit is that exponential backoff of the RTO is not optional.
It is the last bit of mechanism that helps us in times of severe
congestion.  So, if I take an RTO and retransmit 1*MSS amount of
data in a few packets such that I could recover a loss in the
subsequent RTT with a fast retransmit, that is just wrong.  In my
opinion the only way to recover a loss of a lost packet
retransmitted via the RTO is via another (exponentially backed off)
RTO firing.

You are essentially trying to circumvent exponential backoff.  If we
get in that regime, we need to be there.  Period.

> In our opinion, retransmitting old data is dangerously close to
> getting rid of timeouts and using single dupACKs as drop
> indicators. As Mark Allman himself pointed out, this should
> probably be a once-per-connection solution.

You missed the point.  What I said was that reducing the duplicate
ACK threshold should be done only if we can be sure that we're not
in a situation where we are persistently sending spurious
retransmits (due to reordering).  One option that has been suggested
is to allow a FR on < 3 dup ACKs once per connection as a cheap way
to help most connections (since most connections are short).  A
better way (in my opinion) would be to use some sort of Eifel or
DSACK-like scheme to ensure that you are not sending a ton of
spurious retransmits -- in which case you could use a dup ACK
threshold of < 3 during the entire connection if your cwnd was
small (and your mechanisms indidicated no (or almost no) spurious
retransmits). 

> Agreed, the overhead is the price you pay. However, one has to
> wonder whether processing capacity is going to be the bottleneck,
> even in the near future.

I also do not buy that 4 little packets == one big one.

My understanding of routers is pretty rough.  So, someone can
correct me if I am wrong here, but here is my understanding of how
they work (very crudely)...  You allocate memory for a queue in
terms of "slots".  Each "slot" can hold one (MTU sized) packet.
Every packet that comes into the router gets a slot.  If it fills
the slot, fine.  If it doesn't, fine.  All packets get one slot.

So, consider an example where a router has a queue of 10 packets and
I send the first bit of data from a TCP connection I have setup.  In
the SF case the data arrives and takes up 4 slots.  In the standard
case the data arrives and takes up 1 slot.  The slots are used for
as long as it takes to send the 10 things in the queue.  And, for
this amount of time the router is more congested with SF than
standard TCP.  So, as more stuff comes into the queue the chances of
dropping increase more with SF than with standard TCP.

It is interesting that this enhanced loss recovery algorithm
actually *increases* the chances of loss.

At the very least I think you should be considering the downsides of
SF in terms that go beyond the additional packet header overhead.

allman


---
Mark Allman -- NASA GRC/BBN -- http://roland.grc.nasa.gov/~mallman/

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Tue Jan 22 05:52:39 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id FAA15380
	for <tsvwg-archive@odin.ietf.org>; Tue, 22 Jan 2002 05:52:39 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id FAA01421
	for tsvwg-archive@odin.ietf.org; Tue, 22 Jan 2002 05:52:43 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id FAA00713;
	Tue, 22 Jan 2002 05:29:41 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id FAA00688
	for <tsvwg@optimus.ietf.org>; Tue, 22 Jan 2002 05:29:39 -0500 (EST)
Received: from mgw-x2.nokia.com (mgw-x2.nokia.com [131.228.20.22])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id FAA15078
	for <tsvwg@ietf.org>; Tue, 22 Jan 2002 05:29:34 -0500 (EST)
From: john.loughney@nokia.com
Received: from esvir04nok.ntc.nokia.com (esvir04nokt.ntc.nokia.com [172.21.143.36])
	by mgw-x2.nokia.com (Switch-2.1.0/Switch-2.1.0) with ESMTP id g0MATd912780
	for <tsvwg@ietf.org>; Tue, 22 Jan 2002 12:29:39 +0200 (EET)
Received: from esebh001.NOE.Nokia.com (unverified) by esvir04nok.ntc.nokia.com
 (Content Technologies SMTPRS 4.2.5) with ESMTP id <T589a94ec67ac158f24077@esvir04nok.ntc.nokia.com>;
 Tue, 22 Jan 2002 12:29:33 +0200
Received: from esebe004.NOE.Nokia.com ([172.21.138.44]) by esebh001.NOE.Nokia.com with Microsoft SMTPSVC(5.0.2195.3779);
	 Tue, 22 Jan 2002 12:29:30 +0200
content-class: urn:content-classes:message
MIME-Version: 1.0
Content-Type: text/plain;
	charset="iso-8859-1"
Subject: RE: [Tsvwg] Comment on:I-D ACTION:draft-ietf-tsvwg-sctpcsum-02.txt
X-MimeOLE: Produced By Microsoft Exchange V6.0.5762.3
Date: Tue, 22 Jan 2002 12:29:30 +0200
Message-ID: <0C1353ABB1DEB74DB067ADFF749C4EEF09BC60@esebe004.NOE.Nokia.com>
Thread-Topic: [Tsvwg] Comment on:I-D ACTION:draft-ietf-tsvwg-sctpcsum-02.txt
Thread-Index: AcGiulbLaB1D/nMvQjiTjiSfOBgeygAcuYGw
To: <xieqb@cig.mot.com>
Cc: <randall@stewart.chicago.il.us>, <tsvwg@ietf.org>
X-OriginalArrivalTime: 22 Jan 2002 10:29:30.0822 (UTC) FILETIME=[AD3B4E60:01C1A32F]
Content-Transfer-Encoding: 8bit
X-MIME-Autoconverted: from quoted-printable to 8bit by optimus.ietf.org id FAA00689
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 8bit

Hi Qiaobing,

> In certain cases, the alternative approach proposed by Doug 
> may enhance the endpoint's resistance against flooding attacks - 
> you may need less computation to discard an attacking packet by 
> checking the V-tag first. I don't know whether this is the original 
> intention of Doug  when he put this in the security section, though.

Actually, I understand this - but wouldn't the attack scenario
be valid for any checksum calculation?  What I mean is that
there isn't anything special about this, other then the new
checksum calculation may be slightly more computationally
intensive.

br,
John

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Tue Jan 22 07:31:36 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id HAA16359
	for <tsvwg-archive@odin.ietf.org>; Tue, 22 Jan 2002 07:31:35 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id HAA05076
	for tsvwg-archive@odin.ietf.org; Tue, 22 Jan 2002 07:31:38 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id HAA04370;
	Tue, 22 Jan 2002 07:16:21 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id HAA04342
	for <tsvwg@optimus.ietf.org>; Tue, 22 Jan 2002 07:16:19 -0500 (EST)
Received: from prue.eim.surrey.ac.uk (IDENT:exim@prue.eim.surrey.ac.uk [131.227.76.5])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id HAA16159
	for <tsvwg@ietf.org>; Tue, 22 Jan 2002 07:16:16 -0500 (EST)
Received: from phaestos.ee.surrey.ac.uk ([131.227.88.14] ident=eep1lw)
	by prue.eim.surrey.ac.uk with esmtp (Exim 3.33 #4)
	id 16SzqO-00069s-00; Tue, 22 Jan 2002 12:16:04 +0000
Date: Tue, 22 Jan 2002 12:15:58 +0000 (GMT)
From: Lloyd Wood <l.wood@eim.surrey.ac.uk>
X-X-Sender: eep1lw@phaestos.ee.surrey.ac.uk
Reply-To: Lloyd Wood <L.Wood@eim.surrey.ac.uk>
To: Reiner Ludwig <Reiner.Ludwig@ericsson.com>
cc: Kacheong Poon <kcpoon@eng.sun.com>, <tsvwg@ietf.org>
Subject: Re: [Tsvwg] Eifel draft and usage of time-stamps.
In-Reply-To: <5.1.0.14.0.20020121230442.01b29758@mailhost.eed.ericsson.se>
Message-ID: <Pine.SOL.4.43.0201221208080.3898-100000@phaestos.ee.surrey.ac.uk>
Organization: speaking for none
X-url: http://www.ee.surrey.ac.uk/Personal/L.Wood/
X-no-archive: yes
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
X-Scanner: exiscan *16SzqO-00069s-00*lIae/4uzpSo* (SECM, UniS)
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

On Mon, 21 Jan 2002, Reiner Ludwig wrote:

> At 03:05 15.01.2002, Kacheong Poon wrote:
> > > I do not understand this comment.  Eifel is pretty robust, it seems
> > > to me.  But, regardless I haven't heard of any failure cases --
> > > i.e., when it says there was a spurous timeout and there was not.
> > > Are there some?
> >
> >I believe I came across a mail (reading mail after long vacation
> >is a pain...) pointing out that it is not.  I guess Reiner can
> >provide us with some data on his experience with Eifel.
>
> I thought the Eifel algorithm was sufficiently robust in detecting spurious
> retransmits. However, Pasi Sarolahti found this case where the Eifel
> algorithm fails due to what appears to be a flaw in the current RFC1323. It
> also seems that there had been unfinished efforts [update to RFC1323 by
> Jacobson, Braden, Borman Feb. 97] that (among other things, I guess) fixed
> that part of RFC1323.

Feb. 97? Got a full citation? There's also:

http://www.kohala.com/start/tcplw-extensions.txt
TCP Extensions for High Performance: An Update, internet draft,
Braden, 1993.

There are more unfinished attempts at updating 1323?
(as opposed to the elided sack stuff, which finally got
decided on and published.)

thanks,

L.

<L.Wood@surrey.ac.uk>PGP<http://www.ee.surrey.ac.uk/Personal/L.Wood/>



_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Tue Jan 22 08:45:46 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id IAA17951
	for <tsvwg-archive@odin.ietf.org>; Tue, 22 Jan 2002 08:45:46 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id IAA07038
	for tsvwg-archive@odin.ietf.org; Tue, 22 Jan 2002 08:45:49 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id IAA06261;
	Tue, 22 Jan 2002 08:29:40 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id IAA06234
	for <tsvwg@optimus.ietf.org>; Tue, 22 Jan 2002 08:29:38 -0500 (EST)
Received: from penguin-ext.wise.edt.ericsson.se (penguin-ext.wise.edt.ericsson.se [194.237.142.110])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id IAA17561
	for <tsvwg@ietf.org>; Tue, 22 Jan 2002 08:29:33 -0500 (EST)
Received: from eed.ericsson.se (mailhost.eed.ericsson.se [164.48.133.33])
	by penguin.wise.edt.ericsson.se (8.11.0/8.11.0/WIREfire-1.3) with ESMTP id g0MDTYw16447;
	Tue, 22 Jan 2002 14:29:34 +0100 (MET)
Received: from res0010384da36.ericsson.com (rac168pc.edd.ericsson.se [164.48.123.168])
	by eed.ericsson.se (8.8.8+Sun/1.1.mit) with ESMTP id OAA15396;
	Tue, 22 Jan 2002 14:29:31 +0100 (MET)
Message-Id: <5.1.0.14.0.20020122142839.04728288@mailhost.eed.ericsson.se>
X-Sender: eedrel@mailhost.eed.ericsson.se
X-Mailer: QUALCOMM Windows Eudora Version 5.1
Date: Tue, 22 Jan 2002 14:30:25 +0100
To: Lloyd Wood <L.Wood@eim.surrey.ac.uk>
From: Reiner Ludwig <Reiner.Ludwig@ericsson.com>
Subject: Re: [Tsvwg] Eifel draft and usage of time-stamps.
Cc: <tsvwg@ietf.org>
In-Reply-To: <Pine.SOL.4.43.0201221208080.3898-100000@phaestos.ee.surrey
 .ac.uk>
References: <5.1.0.14.0.20020121230442.01b29758@mailhost.eed.ericsson.se>
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"; format=flowed
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

At 13:15 22.01.2002, Lloyd Wood wrote:
> > I thought the Eifel algorithm was sufficiently robust in detecting spurious
> > retransmits. However, Pasi Sarolahti found this case where the Eifel
> > algorithm fails due to what appears to be a flaw in the current RFC1323. It
> > also seems that there had been unfinished efforts [update to RFC1323 by
> > Jacobson, Braden, Borman Feb. 97] that (among other things, I guess) fixed
> > that part of RFC1323.
>
>Feb. 97? Got a full citation? There's also:
>
>http://www.kohala.com/start/tcplw-extensions.txt
>TCP Extensions for High Performance: An Update, internet draft,
>Braden, 1993.

Google only finds this 
http://www.ietf.org/proceedings/97dec/97dec-final-131.htm but I have a copy 
of the mentioned Feb. 97 ID. Maybe somebody else has link of an online version.

///Reiner


_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Tue Jan 22 09:00:12 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id JAA18341
	for <tsvwg-archive@odin.ietf.org>; Tue, 22 Jan 2002 09:00:12 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id JAA07419
	for tsvwg-archive@odin.ietf.org; Tue, 22 Jan 2002 09:00:15 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id IAA06782;
	Tue, 22 Jan 2002 08:37:11 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id IAA06751
	for <tsvwg@optimus.ietf.org>; Tue, 22 Jan 2002 08:37:09 -0500 (EST)
Received: from prue.eim.surrey.ac.uk (IDENT:exim@prue.eim.surrey.ac.uk [131.227.76.5])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id IAA17782
	for <tsvwg@ietf.org>; Tue, 22 Jan 2002 08:37:04 -0500 (EST)
Received: from phaestos.ee.surrey.ac.uk ([131.227.88.14] ident=eep1lw)
	by prue.eim.surrey.ac.uk with esmtp (Exim 3.33 #4)
	id 16T16f-0007P9-00; Tue, 22 Jan 2002 13:36:57 +0000
Date: Tue, 22 Jan 2002 13:36:55 +0000 (GMT)
From: Lloyd Wood <l.wood@eim.surrey.ac.uk>
X-X-Sender: eep1lw@phaestos.ee.surrey.ac.uk
Reply-To: Lloyd Wood <L.Wood@eim.surrey.ac.uk>
To: Reiner Ludwig <Reiner.Ludwig@ericsson.com>
cc: tsvwg@ietf.org
Subject: Re: [Tsvwg] Eifel draft and usage of time-stamps.
In-Reply-To: <5.1.0.14.0.20020122142839.04728288@mailhost.eed.ericsson.se>
Message-ID: <Pine.SOL.4.43.0201221332500.3898-100000@phaestos.ee.surrey.ac.uk>
Organization: speaking for none
X-url: http://www.ee.surrey.ac.uk/Personal/L.Wood/
X-no-archive: yes
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
X-Scanner: exiscan *16T16f-0007P9-00*xMw3zPTl7Xs* (SECM, UniS)
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

On Tue, 22 Jan 2002, Reiner Ludwig wrote:

> At 13:15 22.01.2002, Lloyd Wood wrote:
> > > I thought the Eifel algorithm was sufficiently robust in detecting spurious
> > > retransmits. However, Pasi Sarolahti found this case where the Eifel
> > > algorithm fails due to what appears to be a flaw in the current RFC1323. It
> > > also seems that there had been unfinished efforts [update to RFC1323 by
> > > Jacobson, Braden, Borman Feb. 97] that (among other things, I guess) fixed
> > > that part of RFC1323.
> >
> >Feb. 97? Got a full citation? There's also:
> >
> >http://www.kohala.com/start/tcplw-extensions.txt
> >TCP Extensions for High Performance: An Update, internet draft,
> >Braden, 1993.
>
> Google only finds this
> http://www.ietf.org/proceedings/97dec/97dec-final-131.htm but I have a copy
> of the mentioned Feb. 97 ID. Maybe somebody else has link of an online version.

Jacobson, Braden, Borman, TCP Extensions for High Performance,
expired internet-draft, February 1997.

http://www.watersprings.org/pub/id/draft-ietf-tcplw-high-performance-00.txt
(watersprings RULES.)

L.

third time's a charm?

<L.Wood@surrey.ac.uk>PGP<http://www.ee.surrey.ac.uk/Personal/L.Wood/>




_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@ns.ietf.org  Tue Jan 22 15:02:24 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id PAA08412
	for <tsvwg-archive@odin.ietf.org>; Tue, 22 Jan 2002 15:02:24 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id PAA09304
	for tsvwg-archive@odin.ietf.org; Tue, 22 Jan 2002 15:02:27 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id OAA08085;
	Tue, 22 Jan 2002 14:38:09 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id OAA08057
	for <tsvwg@ns.ietf.org>; Tue, 22 Jan 2002 14:38:08 -0500 (EST)
Received: from NS1.US.PRSERV.NET ([64.180.246.83])
	by ietf.org (8.9.1a/8.9.1a) with SMTP id OAA07244
	for <tsvwg@ietf.org>; Tue, 22 Jan 2002 14:38:04 -0500 (EST)
From: sydney@vkistudios.org
Message-Id: <200201221938.OAA07244@ietf.org>
Reply-To: seo@vkistudios.org
To: tsvwg@ietf.org
Mime-Version: 1.0
Content-Type: text/html; charset="iso-8859-1"
Date: Tue, 22 Jan 2002 11:41:38 -0800
Subject: [Tsvwg] ADV: Professional Internet Marketing
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

<HTML>
<HEAD>
<TITLE>Professional Internet Marketing, VKI Studios 1-866-733-8899</TITLE>
<META name="description" content="A web services company specializing dynamic Internet marketing">
<META name="keywords" content="Dynamic Internet Marketing">
<META http-equiv="CONTENT-LANGUAGE" content="English">
<META name="RATING" content="General">
<META name="distribution" content="global">
<META name="copyright" content="copyright 2002, VKI Studios">
<META name="robots" content="index,follow">
<META http-equiv=Expires content="O">
<LINK rel="stylesheet" href="http://www.vkistudios.com/css/vki.css" type="text/css">
</HEAD>

<BODY bgcolor="#FFFFFF" text="#000000" link="#990000" vlink="663333">
<TABLE width="550" border="0" cellspacing="0" cellpadding="2" align="center">
  <TR> 
    <TD class="newsletter" bgcolor="006699" width="190" align="center"> 
      <P><B><A href="http://www.vkistudios.com/form_internet_mkt.html"><BR>
        <IMG src="http://www.vkistudios.com/images/h_internet_mkt.gif" width="183" height="37" alt="VKI Studios - Internet Marketing Is Mission Critical For Your Business" border="0"></A><BR>
        <OBJECT classid="clsid:D27CDB6E-AE6D-11cf-96B8-444553540000" codebase="http://download.macromedia.com/pub/shockwave/cabs/flash/swflash.cab#version=5,0,0,0" width="183" height="20">
          <PARAM name=movie value="http://www.vkistudios.com/swf/ani_email.swf">
          <PARAM name=quality value=high>
          <PARAM name="BGCOLOR" value="006699">
          <EMBED src="http://www.vkistudios.com/swf/ani_email.swf" quality=high pluginspage="http://www.macromedia.com/shockwave/download/index.cgi?P1_Prod_Version=ShockwaveFlash" type="application/x-shockwave-flash" width="183" height="20" bgcolor="006699">
          </EMBED> 
        </OBJECT><BR>
        &nbsp;</B></P>
    </TD>
    <TD class="promo_header" bgcolor="006699" valign="top" align="right" width="360"><FONT color="#FFFFFF"><B>1-866-733-8899</B></FONT>&nbsp;</TD>
  </TR>
  <TR> 
    <TD class="newsletter" colspan="2"> 
      <P><IMG src="http://www.vkistudios.com/images/shim_email0.gif" width="1" height="17"><B><FONT color="006699">&quot;VKI 
        increased our web traffic by 550%</FONT></B> which resulted in a 20% increase 
        in revenues in less than six months.&quot; John Keith-King, Curator, Granville 
        Island Museums.</P>
      <P>Would you like to see if you can achieve similar results? <BR>
        Call us at 1-866-733-8899 or fill out our <A href="http://www.vkistudios.com/form_internet_mkt.html">online 
        consultation</A> and receive a <B><FONT color="006699"><BR>
        FREE Search Engine Report<FONT size="-1"> (value $75.00)</FONT></FONT></B></P>
      <P><FONT color="#006699"><B><A href="http://www.vkistudios.com/form_internet_mkt.html"><IMG src="http://www.vkistudios.com/images/button_guarantee.gif" width="185" height="29" alt="Results Guaranteed" border="0"></A></B></FONT></P>
    </TD>
  </TR>
  <TR> 
    <TD class="newsletter" colspan="2">&nbsp;</TD>
  </TR>
  <TR> 
    <TD class="newsletter" valign="bottom" colspan="2"> 
      <P><FONT color="#006699"><B>THE BENEFITS</B></FONT></P>
    </TD>
  </TR>
  <TR> 
    <TD class="newsletter" colspan="2"> 
      <UL>
        <LI>Your website will become <FONT color="006699"><B>highly visible</B></FONT></LI>
        <LI>Your <FONT color="006699"><B>traffic will grow</B></FONT> exponentially</LI>
        <LI>Measurable and <FONT color="006699"><B>trackable results</B></FONT></LI>
        <LI>You will have an ongoing <FONT color="006699"><B>strategy that adapts 
          </B></FONT>as search engines change<BR>
        </LI>
      </UL>
    </TD>
  </TR>
  <TR> 
    <TD class="newsletter" colspan="2"><FONT color="#006699"><B> THE ELEVATION</B></FONT></TD>
  </TR>
  <TR> 
    <TD class="newsletter" colspan="2"> 
      <UL>
        <LI> Strategy development</LI>
        <LI>Keyword Formulation</LI>
        <LI>Creation of keyword &amp; description META tags</LI>
        <LI>Creation of keyword entry pages</LI>
        <LI>FTP file transfer to server</LI>
        <LI>Submission of web address (URL) to search engines</LI>
        <LI>Submission of web address (URL) to directory sites<BR>
          <A href="http://www.vkistudios.com/se_phase1.html">more info</A><BR>
        </LI>
      </UL>
    </TD>
  </TR>
  <TR> 
    <TD class="newsletter" colspan="2"><FONT color="#006699"><B>THE PUSH</B></FONT> 
      - ongoing maintenance includes:</TD>
  </TR>
  <TR> 
    <TD class="newsletter" colspan="2"> 
      <UL>
        <LI>Optimization - HTML optimization to be search engine friendly</LI>
        <LI>Submission - Proper submission protocol for each individual engine</LI>
        <LI>Registration - Verification of successful submissions</LI>
        <LI>Monitoring - Evaluation of server file logs for search engine activity</LI>
        <LI>Positioning - Fine-tuning of submitted pages</LI>
        <LI>Maintenance - Continuous attendance to all the above actions<BR>
          <A href="http://www.vkistudios.com/form_internet_mkt.html">Online consultation</A> 
        </LI>
      </UL>
    </TD>
  </TR>
  <TR> 
    <TD class="newsletter" colspan="2"> 
      <P>Many Search Engines are moving to paid listings so <B><FONT color="006699">ACT 
        NOW!</FONT></B><BR>
        <A href="http://www.vkistudios.com/form_internet_mkt.html">Contact us 
        today to take advantage of our limited time offer.</A></P>
      <P>Please call us: 1-866-733-8899 or (604) 733-1474 - <A href="http://www.vkistudios.com" target="_blank">VKI 
        Studios Inc.</A></P>
    </TD>
  </TR>
  <TR> 
    <TD class="newsletter" colspan="2"> 
      <P>&nbsp;</P>
    </TD>
  </TR>
  <TR>
    <TD class="newsletter" colspan="2">If you do not wish to receive further mailings, 
      please <A href="http://www.vkistudios.com/remove/unsubscribe.html">unsubscribe 
      here.</A></TD>
  </TR>
</TABLE>
<P>&nbsp;</P>
</BODY>
</HTML>

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@ns.ietf.org  Tue Jan 22 18:17:41 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id SAA14191
	for <tsvwg-archive@odin.ietf.org>; Tue, 22 Jan 2002 18:17:40 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id SAA17540
	for tsvwg-archive@odin.ietf.org; Tue, 22 Jan 2002 18:17:43 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id SAA17289;
	Tue, 22 Jan 2002 18:04:14 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id SAA17260
	for <tsvwg@ns.ietf.org>; Tue, 22 Jan 2002 18:04:12 -0500 (EST)
Received: from sandmail.sandburst.com (sandmail.sandburst.com [216.57.132.42])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id SAA14056
	for <tsvwg@ietf.org>; Tue, 22 Jan 2002 18:04:08 -0500 (EST)
To: tsvwg@ietf.org
Subject: Re: [Tsvwg] TCP Smart Framing 
In-Reply-To: Message from Mark Allman <mallman@grc.nasa.gov> 
   of "Mon, 21 Jan 2002 22:49:54 EST." <200201220349.WAA31645@lawyers.grc.nasa.gov> 
References: <200201220349.WAA31645@lawyers.grc.nasa.gov> 
Date: Tue, 22 Jan 2002 18:03:32 -0500
From: Stephen Bailey <steph@cs.uchicago.edu>
Message-Id: <20020122230341.38E9F4FC2@sandmail.sandburst.com>
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

> My understanding of routers is pretty rough.  
> [...]
> All packets get one slot.

They actually work about every different way you could imagine, so
it's probably sound not to reason about them much one way or the
other.

Most `new' routers (with hardware-based forwarding planes) don't use
the one slot per packet technique because it results in a substantial
waste of buffering space if the packet stream has many small packets.
Slot-based design usually assume they are only queuing a small
amount of data internally, as might be the case in an older Ethernet
switch. 

There are typically some slotish things in the header processing
parts, but they are provisioned to handle minimum sized packets at
line rate.  In some architectures if you start to turn on a lot of
per-packet features (policing, classification, counters, etc.)  you
may see a (mild or severe) degradation in forwarding performance as a
function of packet rate.

Steph

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@ns.ietf.org  Tue Jan 22 23:20:21 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id XAA21032
	for <tsvwg-archive@odin.ietf.org>; Tue, 22 Jan 2002 23:20:21 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id XAA27804
	for tsvwg-archive@odin.ietf.org; Tue, 22 Jan 2002 23:20:25 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id WAA26918;
	Tue, 22 Jan 2002 22:43:55 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id WAA26888
	for <tsvwg@ns.ietf.org>; Tue, 22 Jan 2002 22:43:53 -0500 (EST)
Received: from lawyers.grc.nasa.gov (d267.as1.clev.oh.voyager.net [216.214.12.78])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id WAA20632
	for <tsvwg@ietf.org>; Tue, 22 Jan 2002 22:43:48 -0500 (EST)
Received: from lawyers (mallman@localhost)
	by lawyers.grc.nasa.gov (8.9.3/8.9.3) with ESMTP id WAA02272;
	Tue, 22 Jan 2002 22:43:05 -0500
Message-Id: <200201230343.WAA02272@lawyers.grc.nasa.gov>
To: Stephen Bailey <steph@cs.uchicago.edu>
From: Mark Allman <mallman@grc.nasa.gov>
Reply-To: mallman@grc.nasa.gov
cc: tsvwg@ietf.org
Subject: Re: [Tsvwg] TCP Smart Framing 
Organization: BBN Technologies/NASA GRC
Song-of-the-Day: Boys of Summer
Date: Tue, 22 Jan 2002 22:43:05 -0500
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org


> They actually work about every different way you could imagine, so
> it's probably sound not to reason about them much one way or the
> other.

OK, I'll buy that advice.

So, to work at the real issue in question here: How do you set the
queue length?  Packets or bytes?  The last time I played with cisco
equipment it was in terms of packets (at least the defaults that
this stupid user was playing with).  So, regardless of the internals
my point regarding smart framing would still hold.  On the other
hand, the point is less valid if the router you are using ends up
queueing in terms of bytes.

allman


---
Mark Allman -- NASA GRC/BBN -- http://roland.grc.nasa.gov/~mallman/

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@ns.ietf.org  Wed Jan 23 00:02:39 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id AAA22046
	for <tsvwg-archive@odin.ietf.org>; Wed, 23 Jan 2002 00:02:39 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id AAA29177
	for tsvwg-archive@odin.ietf.org; Wed, 23 Jan 2002 00:02:40 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id XAA28327;
	Tue, 22 Jan 2002 23:35:50 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id XAA28300
	for <tsvwg@ns.ietf.org>; Tue, 22 Jan 2002 23:35:48 -0500 (EST)
Received: from c003.snv.cp.net (c003-h015.c003.snv.cp.net [209.228.32.229])
	by ietf.org (8.9.1a/8.9.1a) with SMTP id XAA21572
	for <tsvwg@ietf.org>; Tue, 22 Jan 2002 23:35:39 -0500 (EST)
Received: (cpmta 27887 invoked from network); 22 Jan 2002 20:35:11 -0800
Received: from 64.130.130.105 (HELO dougrmt)
  by smtp.telocity.com (209.228.32.229) with SMTP; 22 Jan 2002 20:35:11 -0800
X-Sent: 23 Jan 2002 04:35:11 GMT
From: "Douglas Otis" <dotis@sanlight.net>
To: <mallman@grc.nasa.gov>, "Stephen Bailey" <steph@cs.uchicago.edu>
Cc: <tsvwg@ietf.org>
Subject: RE: [Tsvwg] TCP Smart Framing 
Date: Tue, 22 Jan 2002 20:36:11 -0800
Message-ID: <NEBBJGDMMLHHCIKHGBEJCECACPAA.dotis@sanlight.net>
MIME-Version: 1.0
Content-Type: text/plain;
	charset="iso-8859-1"
Content-Transfer-Encoding: 7bit
X-Priority: 3 (Normal)
X-MSMail-Priority: Normal
X-Mailer: Microsoft Outlook IMO, Build 9.0.2416 (9.0.2910.0)
Importance: Normal
In-Reply-To: <200201230343.WAA02272@lawyers.grc.nasa.gov>
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4807.1700
Content-Transfer-Encoding: 7bit
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit

Mark,

Although memory impact would be less if allocation was done in bytes, the
overhead for such a scheme would include a more expensive allocation process
in addition to queue selections.  The fastest solution I could implement in
software was as you described for cable modem and wireless head-ends.  The
limits of these systems were packets per second and never bytes per second
as wire speed was possible if the average packet size was sufficient.
Memory conservation was not an issue, just per packet overhead against
system bus limitations.  It would be interesting to find the effect on
average packet size over a large network.

Doug

> > They actually work about every different way you could imagine, so
> > it's probably sound not to reason about them much one way or the
> > other.
>
> OK, I'll buy that advice.
>
> So, to work at the real issue in question here: How do you set the
> queue length?  Packets or bytes?  The last time I played with cisco
> equipment it was in terms of packets (at least the defaults that
> this stupid user was playing with).  So, regardless of the internals
> my point regarding smart framing would still hold.  On the other
> hand, the point is less valid if the router you are using ends up
> queueing in terms of bytes.
>
> allman
>
>
> ---
> Mark Allman -- NASA GRC/BBN -- http://roland.grc.nasa.gov/~mallman/


_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Wed Jan 23 02:36:00 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id CAA02913
	for <tsvwg-archive@odin.ietf.org>; Wed, 23 Jan 2002 02:36:00 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id CAA13170
	for tsvwg-archive@odin.ietf.org; Wed, 23 Jan 2002 02:36:01 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id CAA12332;
	Wed, 23 Jan 2002 02:23:02 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id CAA12301
	for <tsvwg@optimus.ietf.org>; Wed, 23 Jan 2002 02:23:00 -0500 (EST)
Received: from c003.snv.cp.net (c003-h015.c003.snv.cp.net [209.228.32.229])
	by ietf.org (8.9.1a/8.9.1a) with SMTP id CAA02611
	for <tsvwg@ietf.org>; Wed, 23 Jan 2002 02:22:58 -0500 (EST)
Received: (cpmta 6744 invoked from network); 22 Jan 2002 23:22:28 -0800
Received: from 64.130.130.105 (HELO dougrmt)
  by smtp.telocity.com (209.228.32.229) with SMTP; 22 Jan 2002 23:22:28 -0800
X-Sent: 23 Jan 2002 07:22:28 GMT
From: "Douglas Otis" <dotis@sanlight.net>
To: "Tsvwg" <tsvwg@ietf.org>
Date: Tue, 22 Jan 2002 23:23:28 -0800
Message-ID: <NEBBJGDMMLHHCIKHGBEJKECBCPAA.dotis@sanlight.net>
MIME-Version: 1.0
Content-Type: text/plain;
	charset="iso-8859-1"
Content-Transfer-Encoding: 7bit
X-Priority: 3 (Normal)
X-MSMail-Priority: Normal
X-Mailer: Microsoft Outlook IMO, Build 9.0.2416 (9.0.2910.0)
Importance: Normal
In-Reply-To: <200201230159.RAA04596@acropora.rose.agilent.com>
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4807.1700
Content-Transfer-Encoding: 7bit
Subject: [Tsvwg] RE: iSCSI: Markers and Framing
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit

Dave,

The iSCSI draft includes use of Fixed Interval Markers-

"Different schemes can be used to recover synchronization.
 One of these schemes is detailed in Appendix A. - Sync and
 Steering with Fixed Interval Markers -. To make these
 schemes work, iSCSI implementations have to make sure that
 the appropriate protocol layers are provided with enough
 information to implement a synchronization and/or data
 steering mechanism."

As you are suggesting there is a significant consequence in TCP behavior if
not adhering to this iSCSI recommendation that mandates a process below the
transport layer for de-encapsulation of iSCSI related structures, the impact
of this requirement should be reviewed within the appropriate workgroup.
Until then, Fixed Interval Markers should not be included within the iSCSI
documentation as this behavior is a change to TCP as you have indicated.

Doug


> -----Original Message-----
> From: owner-ips@ece.cmu.edu [mailto:owner-ips@ece.cmu.edu]On Behalf Of
> Dave Sheehy [mailto:dbs@acropora.rose.agilent.com]
> Sent: Tuesday, January 22, 2002 6:00 PM
> To: IETF IP SAN Reflector
> Subject: RE: iSCSI: Markers and Framing
>
>
> > A couple of comments on this:
> >
> > >    b. I strongly recommend SHOULD implement FIM on the send side.
> > >       Implication -> Senders that do not insert markers should be
> > >       willing to accept up to RTT*BW data drops! Headers being
> > >       "reasonably" out-of-order is OK. Of course, senders that do not
> > >       insert markers but are willing to pay big $$$ to the SSP will
> > >       get their buffer/BW allocation as usual and customary :-)
> >
> > I think the Implication is too strong, as it goes against the following
> > SHOULD from RFC 1122 (which modifies RFC 793):
> >
> >          4.2.2.20  Event Processing: RFC-793 Section 3.9
> >
> >             While it is not strictly required, a TCP SHOULD be capable
> >             of queueing out-of-order TCP segments.
> >
> > A drop causes the segments up to the retransmission of the drop to
> > be out-of-order.  This does not preclude "SHOULD implement FIM", but
> > the above Implication is overdone in my view as it appears to condone
> > drops of all out-of-order segments.
>
> It is not overdone. It is the reality of the situation for a one
> touch NIC.
> If an implementation does not want to implement some form of framing
> that is the consequence it must be willing to pay for that choice.
>
> Dave Sheehy


_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Wed Jan 23 12:19:59 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id MAA17018
	for <tsvwg-archive@odin.ietf.org>; Wed, 23 Jan 2002 12:19:59 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id MAA06289
	for tsvwg-archive@odin.ietf.org; Wed, 23 Jan 2002 12:20:02 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id LAA03896;
	Wed, 23 Jan 2002 11:41:28 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id LAA03869
	for <tsvwg@optimus.ietf.org>; Wed, 23 Jan 2002 11:41:26 -0500 (EST)
Received: from linuxpow.com (IDENT:qmailr@linuxpow.com [12.149.2.10])
	by ietf.org (8.9.1a/8.9.1a) with SMTP id LAA15679
	for <tsvwg@ietf.org>; Wed, 23 Jan 2002 11:41:23 -0500 (EST)
Date: Wed, 23 Jan 2002 11:41:23 -0500 (EST)
Message-Id: <200201231641.LAA15679@ietf.org>
Received: (qmail 1267 invoked from network); 23 Jan 2002 16:22:37 -0000
Received: from buystainlessonline.com (HELO ) (nobody@12.149.2.55)
  by mail.buystainlessonline.com with SMTP; 23 Jan 2002 16:22:37 -0000
To: tsvwg@ietf.org
From: "BuyStainlessOnline.com" <sales@buystainlessonline.com>
Subject: [Tsvwg] The Stainless Steel Network
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

The Stainless Steel Network is brought to you by www.buystainlessonline.com

MARKET PULSE
After attending an AWMI meeting last thursday evening, and gauging the current market conditions against last years poor performance, it is my professional opinion that markets are shifting. We have seen an increase of sales activity since Jan. 1st. Look for higher prices and thinning stocks for stainless. 

Mason Fine
President
BuyStainlessOnline, Inc.

-------------------------------------------------
Stainless Deals  (FOB Anywhere USA)
PRIME with Mill Test Reports 
375,000 pound Min. Order all items below

11ga 304 48 x 120 |price| $.55.lb
1/4  304 48 x 144 |price| $.58.lb
-------------------------------------------------



www.BuyStainlessOnline.com
Your Place for Stainless Today.

P 215.604.5922
F 240.358.8483

Click Here to REGISTER!
https://www.buystainlessonline.com/registration/registration.php

Unsubscribe By clicking below: 
 http://www.buystainlessonline.com/email/mail.php?action=delete&eval=54804&email=tsvwg@ietf.org

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Wed Jan 23 19:41:29 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id TAA00326
	for <tsvwg-archive@odin.ietf.org>; Wed, 23 Jan 2002 19:41:25 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id TAA25096
	for tsvwg-archive@odin.ietf.org; Wed, 23 Jan 2002 19:41:26 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id TAA24481;
	Wed, 23 Jan 2002 19:21:51 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id TAA24454
	for <tsvwg@optimus.ietf.org>; Wed, 23 Jan 2002 19:21:49 -0500 (EST)
Received: from e21.nc.us.ibm.com (e21.nc.us.ibm.com [32.97.136.227])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id TAA00135
	for <tsvwg@ietf.org>; Wed, 23 Jan 2002 19:21:48 -0500 (EST)
Received: from southrelay02.raleigh.ibm.com (southrelay02.raleigh.ibm.com [9.37.3.209])
	by e21.nc.us.ibm.com (8.9.3/8.9.3) with ESMTP id SAA45530;
	Wed, 23 Jan 2002 18:17:23 -0600
Received: from d04nm110.raleigh.ibm.com (d04nm110.raleigh.ibm.com [9.27.5.85])
	by southrelay02.raleigh.ibm.com (8.11.1m3/NCO v5.01) with ESMTP id g0O0LkU107796;
	Wed, 23 Jan 2002 19:21:46 -0500
Importance: Normal
To: randall@stewart.chicago.il.us
Cc: tsvwg@ietf.org, sctp-developers-list@cig.mot.com
X-Mailer: Lotus Notes Release 5.0.5  September 22, 2000
Message-ID: <OF43E5DA83.3F9164FF-ON85256B4A.00804CDD@raleigh.ibm.com>
From: "Daisy Chang" <daisyc@us.ibm.com>
Date: Wed, 23 Jan 2002 17:30:47 -0600
X-MIMETrack: Serialize by Router on D04NM110/04/M/IBM(Release 5.0.9 |November 16, 2001) at
 01/23/2002 07:21:46 PM
MIME-Version: 1.0
Content-type: text/plain; charset=us-ascii
Subject: [Tsvwg] question regarding the SCTP API - SCTP_INIT cmsg
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

Hi Randy,
      A question came up when I was implementing the SCTP_INIT cmsg in
sendmsg() API on Linux (the lksctp project), I hope that I can find a quick
answer here.
      The question is, what is the behavior of a  sendmsg() call with a
0-length
message, i.e., the total length of iov_len of the msg_iov in the struct
msghdr is 0, or,
both the msg_iov and msg_iovlen are 0. Now, if the msg_name is specified,
and an
existing association is identified with it, I believe that no data should
be sent out; we
will process the cmsg and msg_flags if they are specified, and we are done.
But, if
there is no existing association is found for the msg_name, should we go
ahead
establishing one?
      Considering the following cases, is there any differences among these
3 API's in terms of the effects:
1. sendmsg() with msg_name and 0-length data, and NULL cmsg?
2. sendmsg() with msg_name and 0-length data, and a SCTP_INIT cmsg?
3. sendmsg() with msg_name and 0-length data, and a SCTP_SNDRCVINFO cmsg?
      The <draft-ietf-tsvwg-sctpsocket-02.txt> doesn't seem to address this
very
specifically. Would you consider to add something to the draft to ensure
the
consistencies among implementations?


Thanks,
Daisy

Daisy Chang
IBM Linux Technology Center
daisyc@us.ibm.com
Tel: (512)-838-4194    T/L: 678-4194
FAX: (512)-838-4663


_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Wed Jan 23 21:33:40 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id VAA01606
	for <tsvwg-archive@odin.ietf.org>; Wed, 23 Jan 2002 21:33:40 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id VAA28068
	for tsvwg-archive@odin.ietf.org; Wed, 23 Jan 2002 21:33:42 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id VAA27521;
	Wed, 23 Jan 2002 21:19:56 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id VAA27489
	for <tsvwg@optimus.ietf.org>; Wed, 23 Jan 2002 21:19:54 -0500 (EST)
Received: from chmls20.mediaone.net (chmls20.mediaone.net [24.147.1.156])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id VAA01429
	for <tsvwg@ietf.org>; Wed, 23 Jan 2002 21:19:50 -0500 (EST)
Received: from cs.uchicago.edu (h005018030f43.ne.mediaone.net [66.31.16.32])
	by chmls20.mediaone.net (8.11.1/8.11.1) with ESMTP id g0O2Lbx05985;
	Wed, 23 Jan 2002 21:21:37 -0500 (EST)
Message-Id: <200201240221.g0O2Lbx05985@chmls20.mediaone.net>
To: tsvwg@ietf.org
cc: mallman@grc.nasa.gov
Subject: Re: [Tsvwg] TCP Smart Framing 
In-Reply-To: Message from Mark Allman <mallman@grc.nasa.gov> 
   of "Tue, 22 Jan 2002 22:43:05 EST." <200201230343.WAA02272@lawyers.grc.nasa.gov> 
References: <200201230343.WAA02272@lawyers.grc.nasa.gov> 
Date: Wed, 23 Jan 2002 21:19:35 -0500
From: Stephen Bailey <steph@cs.uchicago.edu>
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

> So, to work at the real issue in question here: How do you set the
> queue length?  Packets or bytes?

Either, depending on the box.  It's impossible for me to say whether
one is more common than the other.  As I suggested, I suspect bytes is
more common for newer gear, since that's the resource that is likely
to have a `physical' incarnation in the router.

RED-type algorithms (which give you probabilistic queue length limits)
operate on bytes.

With respect to routers, two small packets or one big one is probably
a toss up.  It is certainly a goal of new equipment (I think of
Juniper as being the beginning of `new' but I'm sure there's older
`news' than that), to make that the case.  Two small packets has
slightly more overhead in headers, and that's about as far as you can
go without stepping on to thin ice.

However, as far as current end stations are concerned, there's a big
difference between two small and one large.

Steph

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Thu Jan 24 05:33:50 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id FAA19082
	for <tsvwg-archive@odin.ietf.org>; Thu, 24 Jan 2002 05:33:50 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id FAA22336
	for tsvwg-archive@odin.ietf.org; Thu, 24 Jan 2002 05:33:53 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id FAA21669;
	Thu, 24 Jan 2002 05:17:10 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id FAA21638
	for <tsvwg@optimus.ietf.org>; Thu, 24 Jan 2002 05:17:08 -0500 (EST)
Received: from sj-msg-core-2.cisco.com (sj-msg-core-2.cisco.com [171.69.24.11])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id FAA18955
	for <tsvwg@ietf.org>; Thu, 24 Jan 2002 05:17:05 -0500 (EST)
Received: from mira-sjc5-2.cisco.com (mira-sjc5-2.cisco.com [171.71.163.16])
	by sj-msg-core-2.cisco.com (8.11.3/8.9.1) with ESMTP id g0OAGad14896;
	Thu, 24 Jan 2002 02:16:36 -0800 (PST)
Received: from stewart.chicago.il.us (ssh-sj1.cisco.com [171.68.225.134])
	by mira-sjc5-2.cisco.com (Mirapoint)
	with ESMTP id AAV54352;
	Thu, 24 Jan 2002 02:16:34 -0800 (PST)
Message-ID: <3C4FDF01.F70A2656@stewart.chicago.il.us>
Date: Thu, 24 Jan 2002 04:16:34 -0600
From: Randall Stewart <randall@stewart.chicago.il.us>
X-Mailer: Mozilla 4.76 [en] (X11; U; Linux 2.2.12 i386)
X-Accept-Language: en
MIME-Version: 1.0
To: Daisy Chang <daisyc@us.ibm.com>
CC: tsvwg@ietf.org, sctp-developers-list@cig.mot.com
References: <OF43E5DA83.3F9164FF-ON85256B4A.00804CDD@raleigh.ibm.com>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Transfer-Encoding: 7bit
Subject: [Tsvwg] Re: question regarding the SCTP API - SCTP_INIT cmsg
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit

Daisy:

Here is my interpretation/what we did:

1) Any 0 length send that has no msgflg's to process is an
   error and return EINVAL
2) A 0 length send that has msgflgs aka ABORT/TERM get the
   shutdown to happen.
3) To start an association with no data in the UDP model we
   use connect (much like UDP but without the binding
   effect to the peer endpoint ) :>

R

Daisy Chang wrote:
> 
> Hi Randy,
>       A question came up when I was implementing the SCTP_INIT cmsg in
> sendmsg() API on Linux (the lksctp project), I hope that I can find a quick
> answer here.
>       The question is, what is the behavior of a  sendmsg() call with a
> 0-length
> message, i.e., the total length of iov_len of the msg_iov in the struct
> msghdr is 0, or,
> both the msg_iov and msg_iovlen are 0. Now, if the msg_name is specified,
> and an
> existing association is identified with it, I believe that no data should
> be sent out; we
> will process the cmsg and msg_flags if they are specified, and we are done.
> But, if
> there is no existing association is found for the msg_name, should we go
> ahead
> establishing one?
>       Considering the following cases, is there any differences among these
> 3 API's in terms of the effects:
> 1. sendmsg() with msg_name and 0-length data, and NULL cmsg?
> 2. sendmsg() with msg_name and 0-length data, and a SCTP_INIT cmsg?
> 3. sendmsg() with msg_name and 0-length data, and a SCTP_SNDRCVINFO cmsg?
>       The <draft-ietf-tsvwg-sctpsocket-02.txt> doesn't seem to address this
> very
> specifically. Would you consider to add something to the draft to ensure
> the
> consistencies among implementations?
> 
> Thanks,
> Daisy
> 
> Daisy Chang
> IBM Linux Technology Center
> daisyc@us.ibm.com
> Tel: (512)-838-4194    T/L: 678-4194
> FAX: (512)-838-4663

-- 
Randall R. Stewart
randall@stewart.chicago.il.us 815-342-5222 (cell phone)

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Thu Jan 24 08:12:22 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id IAA20732
	for <tsvwg-archive@odin.ietf.org>; Thu, 24 Jan 2002 08:12:21 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id IAA26898
	for tsvwg-archive@odin.ietf.org; Thu, 24 Jan 2002 08:12:24 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id HAA26299;
	Thu, 24 Jan 2002 07:59:55 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id HAA26273
	for <tsvwg@optimus.ietf.org>; Thu, 24 Jan 2002 07:59:53 -0500 (EST)
Received: from mail.tlc.polito.it (IDENT:root@prezzemolo.polito.it [130.192.9.131])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id HAA20528
	for <tsvwg@ietf.org>; Thu, 24 Jan 2002 07:59:49 -0500 (EST)
Received: from polito.it (IDENT:michela@pomodoro.polito.it [130.192.9.133])
	by mail.tlc.polito.it (8.11.6/8.11.6) with ESMTP id g0OCxpj18270;
	Thu, 24 Jan 2002 13:59:51 +0100
Message-ID: <3C500547.496C6C7D@polito.it>
Date: Thu, 24 Jan 2002 13:59:51 +0100
From: Michela Meo <michela@polito.it>
X-Mailer: Mozilla 4.76 [en] (X11; U; Linux 2.2.14 i686)
X-Accept-Language: en
MIME-Version: 1.0
To: tsvwg@ietf.org
CC: Claudio Casetti <casetti@polito.it>, Michela Meo <michela@polito.it>,
        mellia@polito.it
Subject: RE:[tsvwg] TCP Smart Framing
References: <200201220349.WAA31645@lawyers.grc.nasa.gov>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Transfer-Encoding: 7bit
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit

> > * Small windows: LT does not prevent a timeout if IW=1 or when TCP
> > restarts with LW=1. TCP-SF always prevents timeouts in single-
> > loss scenarios, whatever the size of cwnd.
> Maybe I am missing something here, but why not just use a larger
> initial window per RFC 2414 (and soon an updated standards track
> version, I believe)?

True. This affects all TCP versions, and then all TCP versions will be
able 
to take advantage of "increased initial window". 
However, TCP-SF will start sending always 4 segments, whatever the
MSS, while 
"classic" TCP will start sending either 4, 3 or 2 segments depending
on the 
negotiated MSS... 
Still, if limited transmit is implemented/enabled, it will  take (at
least) 
one more RTT to trigger Fast Recovery in case of packet loss...

> As for the LW.  Well, this is bogus.  I don't know how else to put
> it... My own hit is that exponential backoff of the RTO is not optional.
> It is the last bit of mechanism that helps us in times of severe
> congestion.  So, if I take an RTO and retransmit 1*MSS amount of
> data in a few packets such that I could recover a loss in the
> subsequent RTT with a fast retransmit, that is just wrong.

We agree that the RTO mechanism is essential in times of severe
congestion.
However, if the congestion was so severe, it is highly likely that not
even 
TCP-SF will get away without dropping more than one small segments, 
and therefore RTO will fire using the backoff mechanism. 
On the other hand, if there's congestion such that the "optimal" 
value of the CWND is really small,  i.e., smaller than 4*MSS, then 
Classic TCP's only chance to detect a  segment loss is to use a RTO! 
TCP-SF (and Limited Transmit too...)  allows the sender to receive 
more feedback, _WITHOUT_  increasing the offered load to the
bottleneck link.

On top of that, TCP-SF will definitively be more efficient under 
heavy load/ high drop probability, as the sender will have to resend 
only one fourth of data... 
again (MSS/4+headers)*4 + (MSS/4+headers)*P < (MSS+headers) +
(MSS+headers)*P 
if P is large...
This will reduce the (bytewise) offered load to the congested link,
and 
thus will alleviate the congestion ...

This is if routers are bandwidth-limited. If limitations are due to
packet-
processing time, we agree with you that TCP-SF may pose some problems. 
In the latter case, you should blame to your network solution vendor!
:)
As several people on this mailing list pointed out, nowadays routers
are
engineered so that they can easily switch batches of short packets at
linespeed...

If the sender enters that regime it is because CWND=1MSS is too large,
so 
one had better send smaller-size segments, to ensure that at least a 
portion of data arrives to the receiver. Also the sender has the
chance 
to keep sending smaller-size segments, instead of retrying sending 
1 full-size segments.

In other words, TCP-SF is giving the sender the chance to operate with
a 
granularity equal to  MSS/4.

> > In our opinion, retransmitting old data is dangerously close to
> > getting rid of timeouts and using single dupACKs as drop
> > indicators. As Mark Allman himself pointed out, this should
> > probably be a once-per-connection solution.
>
> You missed the point.  What I said was that reducing the duplicate
> ACK threshold should be done only if we can be sure that we're not
> in a situation where we are persistently sending spurious
> retransmits (due to reordering).  One option that has been suggested
> is to allow a FR on < 3 dup ACKs once per connection as a cheap way
> to help most connections (since most connections are short).  A
> better way (in my opinion) would be to use some sort of Eifel or
> DSACK-like scheme to ensure that you are not sending a ton of
> spurious retransmits -- in which case you could use a dup ACK
> threshold of < 3 during the entire connection if your cwnd was
> small (and your mechanisms indidicated no (or almost no) spurious
> retransmits).

Is this any simpler than keeping the dup-ack threshold at 3 and using 
Smart-Framing? 

> > Agreed, the overhead is the price you pay. However, one has to
> > wonder whether processing capacity is going to be the bottleneck,
> > even in the near future.
>
> I also do not buy that 4 little packets == one big one.

Indeed, 4 little packets is better than one big packet :-)

> My understanding of routers is pretty rough.  So, someone can
> correct me if I am wrong here, but here is my understanding of how
> they work (very crudely)...  You allocate memory for a queue in
> terms of "slots".  Each "slot" can hold one (MTU sized) packet.
> Every packet that comes into the router gets a slot.  If it fills
> the slot, fine.  If it doesn't, fine.  All packets get one slot.

This is not true, as already pointed out by other people on the list. 
There are architectures out there that are working like you 
are saying. But there are also architectures where buffers are
partitioned in 
variable-size chunks.
In this case, small-sized packets will suffer a lower dropping
probability. 
And given that a flow will use small-sized packets only when it is in
the 
small window regime (and thus sending at low rate, and sensitive to
packet 
drop), it is much better for the network to force large-window
(and thus higher-rate, more bandwidth-consuming, more reactive) 
connections to slow down.

> It is interesting that this enhanced loss recovery algorithm
> actually *increases* the chances of loss.

One cannot fail to notice that "Increasing TCP initial window" does 
exacly the same...

> At the very least I think you should be considering the downsides of
> SF in terms that go beyond the additional packet header overhead.

We took care of these issues in the first version of the I-D, as 
well as in the Globecom paper.
(http://www1.tlc.polito.it/mellia/papers/globecom01_TCP-FS.pdf)

Michela, Marco and Claudio

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Thu Jan 24 09:45:09 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id JAA22535
	for <tsvwg-archive@odin.ietf.org>; Thu, 24 Jan 2002 09:45:04 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id JAA01269
	for tsvwg-archive@odin.ietf.org; Thu, 24 Jan 2002 09:45:08 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id JAA00748;
	Thu, 24 Jan 2002 09:31:01 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id JAA00710
	for <tsvwg@optimus.ietf.org>; Thu, 24 Jan 2002 09:30:58 -0500 (EST)
Received: from seraph2.grc.nasa.gov (firewall-user@seraph2.lerc.nasa.gov [128.156.10.11])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id JAA22243
	for <tsvwg@ietf.org>; Thu, 24 Jan 2002 09:30:53 -0500 (EST)
Received: from lombok-fi.lerc.nasa.gov (lombok-fi.lerc.nasa.gov [139.88.112.33])
	by seraph2.grc.nasa.gov (Postfix) with ESMTP id 85FA6C6952
	for <tsvwg@ietf.org>; Thu, 24 Jan 2002 09:30:55 -0500 (EST)
Received: from guns.lerc.nasa.gov (guns.lerc.nasa.gov [139.88.87.35])
	by lombok-fi.lerc.nasa.gov (NASA LeRC 8.9.1.1/8.9.1) with ESMTP id JAA23836;
	Thu, 24 Jan 2002 09:30:46 -0500 (EST)
Received: from guns.lerc.nasa.gov (localhost.lerc.nasa.gov [127.0.0.1]) by guns.lerc.nasa.gov with ESMTP (NASA LeRC 8.7.4.1/2.01-local)
        id JAA51186; Thu, 24 Jan 2002 09:30:41 -0500 (EST)
Message-Id: <200201241430.JAA51186@guns.lerc.nasa.gov>
To: Michela Meo <michela@polito.it>
From: Mark Allman <mallman@grc.nasa.gov>
Reply-To: mallman@grc.nasa.gov
Cc: tsvwg@ietf.org, Claudio Casetti <casetti@polito.it>, mellia@polito.it
Subject: Re: [tsvwg] TCP Smart Framing 
Organization: BBN Technologies/NASA GRC
Song-of-the-Day: Big Bang Baby
Date: Thu, 24 Jan 2002 09:30:41 -0500
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org


> Still, if limited transmit is implemented/enabled, it will take
> (at least) one more RTT to trigger Fast Recovery in case of packet
> loss...

Why is this a huge problem?  It's not like TCP is sitting around
doing nothing.  It is doing something useful while trying to elicit
these additional ACKs.  Sure, there is a modest delivery penalty to
the application, but is this really that big of an issue?

> We agree that the RTO mechanism is essential in times of severe
> congestion.  However, if the congestion was so severe, it is
> highly likely that not even TCP-SF will get away without dropping
> more than one small segments, and therefore RTO will fire using
> the backoff mechanism.

Two points...

  * First, I'd rather not bet on "highly likely" after an RTO.

  * Second, even if you drop more than one segment after an RTO you
    can still fast retransmit.  Consider sending your four small
    segments, with the second two being lost.  The first two are
    ACKed, which sends out new data that can then trigger duplicate
    ACKs.  Especially if you consider limited transmit in
    conjunction with this.

And, even if we discount the second item as an edge case, I still
think that a loss of an segment resent via RTO must be recovered via
another (exponentially backed off) RTO.  This is not the place to
add aggressiveness.

Think about it another way...  If we're in this small window case
with SF and we take an RTO that means the congestion has already
proved too much for SF to successfully handle.  So, why would it be
appropriate to be anything but very conservative?

> On top of that, TCP-SF will definitively be more efficient under 
> heavy load/ high drop probability, as the sender will have to resend 
> only one fourth of data... 

Which matters on bandwidth dominated paths, but not on delay
dominated paths.

> In other words, TCP-SF is giving the sender the chance to operate
> with a granularity equal to MSS/4.

Yep.  So, the real question is whether this is useful.  On the one
end we can operate at the path MTU.  On the other end we can send 1
byte per packet.  Those are the bounds.  From the packet processing
and overhead standpoints we want to be at the MTU-sized packets end
of things.  From the standpoint of wanting to retransmit the least
number of bits and generate the most duplicate ACKs we want to send
1 byte per packet.  Packet sizes in the middle are a mix of
tradeoffs.  

So, my question on this remains: Is there a compelling reason to
move from the MTU sized packets?  I, personally, have not yet been
convinced.

> > You missed the point.  What I said was that reducing the duplicate
> > ACK threshold should be done only if we can be sure that we're not
> > in a situation where we are persistently sending spurious
> > retransmits (due to reordering).  
> 
> Is this any simpler than keeping the dup-ack threshold at 3 and
> using Smart-Framing?

???

I don't know.  I don't think of either of them as terribly hard.
And, some of us want to be able to track spurious retransmissions
and hence would have most of the necessary machinary for doing what
I suggested in place already.

> This is not true, as already pointed out by other people on the
> list.  There are architectures out there that are working like you
> are saying. But there are also architectures where buffers are
> partitioned in variable-size chunks.  In this case, small-sized
> packets will suffer a lower dropping probability.  And given that
> a flow will use small-sized packets only when it is in the small
> window regime (and thus sending at low rate, and sensitive to
> packet drop), it is much better for the network to force
> large-window (and thus higher-rate, more bandwidth-consuming, more
> reactive) connections to slow down.

What I took away from the comments was that there are bunches of
different schemes in use and making any one assumption is dicey.
That seems reasonable to me.  So, I think that means that you need
to think about the case when the router is setup as I described and
we need to think of the impact on queueing and congestion in that
case.

> > It is interesting that this enhanced loss recovery algorithm
> > actually *increases* the chances of loss.
> 
> One cannot fail to notice that "Increasing TCP initial window"
> does exacly the same...

(True -- all the measuerments I have seen show this.  At the same
time, we never claimed it to be a better loss recovery mechanism.)

allman


---
Mark Allman -- BBN/NASA GRC -- http://roland.grc.nasa.gov/~mallman/

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@ns.ietf.org  Thu Jan 24 11:53:44 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id LAA26989
	for <tsvwg-archive@odin.ietf.org>; Thu, 24 Jan 2002 11:53:44 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id LAA07851
	for tsvwg-archive@odin.ietf.org; Thu, 24 Jan 2002 11:53:47 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id LAA06574;
	Thu, 24 Jan 2002 11:35:58 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id LAA06546
	for <tsvwg@ns.ietf.org>; Thu, 24 Jan 2002 11:35:56 -0500 (EST)
Received: from sj-msg-core-1.cisco.com (sj-msg-core-1.cisco.com [171.71.163.11])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id LAA26265
	for <tsvwg@ietf.org>; Thu, 24 Jan 2002 11:35:52 -0500 (EST)
Received: from mira-sjc5-2.cisco.com (mira-sjc5-2.cisco.com [171.71.163.16])
	by sj-msg-core-1.cisco.com (8.11.3/8.9.1) with ESMTP id g0OGZNL13525;
	Thu, 24 Jan 2002 08:35:23 -0800 (PST)
Received: from stewart.chicago.il.us (ssh-sj1.cisco.com [171.68.225.134])
	by mira-sjc5-2.cisco.com (Mirapoint)
	with ESMTP id AAV57912;
	Thu, 24 Jan 2002 08:35:22 -0800 (PST)
Message-ID: <3C5037C9.99DBCF19@stewart.chicago.il.us>
Date: Thu, 24 Jan 2002 10:35:21 -0600
From: Randall Stewart <randall@stewart.chicago.il.us>
X-Mailer: Mozilla 4.76 [en] (X11; U; Linux 2.2.12 i386)
X-Accept-Language: en
MIME-Version: 1.0
To: Daisy Chang <daisyc@us.ibm.com>
CC: tsvwg@ietf.org, sctp-developers-list@cig.mot.com
References: <OF824418F4.275C7F17-ON85256B4B.0055F883@raleigh.ibm.com>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Transfer-Encoding: 7bit
Subject: [Tsvwg] Re: question regarding the SCTP API - SCTP_INIT cmsg
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit

Daisy Chang wrote:
> 
> Randy,
>       Ah, connect()! I've got the impression a while back that a definition
> for
> "connect() - UDP style" is contemplated. Now that you mentioned it, I
> somehow
> couldn't find any reference to it in the API draft. Is this something that
> we should
> consider to add to our UDP-style support?

Yes, we overlooked this one.. it is already in the -03 (pending serious
objection from my co-authors :>).

>       I tend to agree that a 0-length send should not do much. I'm not sure
> if a EINVAL is necessary though, given that TCP and UDP both seem to take
> it without complaining - as far as I can tell on Linux. I could be wrong
> though
> because I haven't checked the library routines to see if there is any
> specific
> length verification there.

I believe that TCP and UDP do allow a 0 length send.. not sure what
happens when it is done :>... In SCTP it is forbidden explicitly by
the RFC.. this is why I think a EINVAL should be returned since it
is NOT allowed and I want to make sure the user understands it is an
error :>


>       Another thing is that, to me, SCTP_INIT seems to have slightly
> different
> functionality than a straight connect(). For a UDP-style socket, SCTP_INIT
> could provide user with finer granularity of control of those init
> parameters per
> connection. While setsockopt(SCTP_INITMSG) and connect() can provide
> the similar results, I'm wondering, rather than requiring SCTP_INIT with
> data,
> if we should allow my case #1 to work as "a connect() with init parms"?
> What
> is your insight?
> 

I think this gets into a grey area of implementation decision. I can
see that a send() with no length COULD be interepreted as a connect.
And yes you no longer need to do a setsockopt()/connect()... 

It would be more efficient.. but I am hesitant to give the user
the impression that sending() 0 bytes is ok :>

I don't know... I can see both sides.. I will be keeping connect()
in our BSD implementation and leave it at that ... i.e. no send() 0
bytes ..

But as I said.. I think this is more an implementation choice.

R
--
Randall R. Stewart
randall@stewart.chicago.il.us 815-342-5222 (cell phone)

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@ns.ietf.org  Thu Jan 24 12:12:09 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id MAA27963
	for <tsvwg-archive@odin.ietf.org>; Thu, 24 Jan 2002 12:12:08 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id MAA09954
	for tsvwg-archive@odin.ietf.org; Thu, 24 Jan 2002 12:12:11 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id LAA05931;
	Thu, 24 Jan 2002 11:27:18 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id LAA05901
	for <tsvwg@ns.ietf.org>; Thu, 24 Jan 2002 11:27:16 -0500 (EST)
Received: from e21.nc.us.ibm.com (e21.nc.us.ibm.com [32.97.136.227])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id LAA25972
	for <tsvwg@ietf.org>; Thu, 24 Jan 2002 11:27:13 -0500 (EST)
Received: from southrelay02.raleigh.ibm.com (southrelay02.raleigh.ibm.com [9.37.3.209])
	by e21.nc.us.ibm.com (8.9.3/8.9.3) with ESMTP id KAA111474;
	Thu, 24 Jan 2002 10:22:38 -0600
Received: from d04nm110.raleigh.ibm.com (d04nm110.raleigh.ibm.com [9.27.5.85])
	by southrelay02.raleigh.ibm.com (8.11.1m3/NCO v5.01) with ESMTP id g0OGQ9t216524;
	Thu, 24 Jan 2002 11:26:09 -0500
Importance: Normal
To: Randall Stewart <randall@stewart.chicago.il.us>
Cc: <tsvwg@ietf.org>, <sctp-developers-list@cig.mot.com>
X-Mailer: Lotus Notes Release 5.0.5  September 22, 2000
Message-ID: <OF824418F4.275C7F17-ON85256B4B.0055F883@raleigh.ibm.com>
From: "Daisy Chang" <daisyc@us.ibm.com>
Date: Thu, 24 Jan 2002 09:57:58 -0600
X-MIMETrack: Serialize by Router on D04NM110/04/M/IBM(Release 5.0.9 |November 16, 2001) at
 01/24/2002 11:26:09 AM
MIME-Version: 1.0
Content-type: text/plain; charset=us-ascii
Subject: [Tsvwg] Re: question regarding the SCTP API - SCTP_INIT cmsg
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

Randy,
      Ah, connect()! I've got the impression a while back that a definition
for
"connect() - UDP style" is contemplated. Now that you mentioned it, I
somehow
couldn't find any reference to it in the API draft. Is this something that
we should
consider to add to our UDP-style support?
      I tend to agree that a 0-length send should not do much. I'm not sure
if a EINVAL is necessary though, given that TCP and UDP both seem to take
it without complaining - as far as I can tell on Linux. I could be wrong
though
because I haven't checked the library routines to see if there is any
specific
length verification there.
      Another thing is that, to me, SCTP_INIT seems to have slightly
different
functionality than a straight connect(). For a UDP-style socket, SCTP_INIT
could provide user with finer granularity of control of those init
parameters per
connection. While setsockopt(SCTP_INITMSG) and connect() can provide
the similar results, I'm wondering, rather than requiring SCTP_INIT with
data,
if we should allow my case #1 to work as "a connect() with init parms"?
What
is your insight?


Thanks,
Daisy

Daisy Chang
IBM Linux Technology Center
daisyc@us.ibm.com
Tel: (512)-838-4194    T/L: 678-4194
FAX: (512)-838-4663


Randall Stewart<randall@stewart.chicago.il.us>@cig.mot.com on 01/24/2002
04:16:34 AM

Please respond to Randall Stewart<randall@stewart.chicago.il.us>

Sent by:    owner-sctp-developers-list@cig.mot.com


To:    Daisy Chang/Austin/IBM@IBMUS
cc:    <tsvwg@ietf.org>, <sctp-developers-list@cig.mot.com>
Subject:    Re: question regarding the SCTP API - SCTP_INIT cmsg


Daisy:

Here is my interpretation/what we did:

1) Any 0 length send that has no msgflg's to process is an
   error and return EINVAL
2) A 0 length send that has msgflgs aka ABORT/TERM get the
   shutdown to happen.
3) To start an association with no data in the UDP model we
   use connect (much like UDP but without the binding
   effect to the peer endpoint ) :>

R

Daisy Chang wrote:
>
> Hi Randy,
>       A question came up when I was implementing the SCTP_INIT cmsg in
> sendmsg() API on Linux (the lksctp project), I hope that I can find a
quick
> answer here.
>       The question is, what is the behavior of a  sendmsg() call with a
> 0-length
> message, i.e., the total length of iov_len of the msg_iov in the struct
> msghdr is 0, or,
> both the msg_iov and msg_iovlen are 0. Now, if the msg_name is specified,
> and an
> existing association is identified with it, I believe that no data should
> be sent out; we
> will process the cmsg and msg_flags if they are specified, and we are
done.
> But, if
> there is no existing association is found for the msg_name, should we go
> ahead
> establishing one?
>       Considering the following cases, is there any differences among
these
> 3 API's in terms of the effects:
> 1. sendmsg() with msg_name and 0-length data, and NULL cmsg?
> 2. sendmsg() with msg_name and 0-length data, and a SCTP_INIT cmsg?
> 3. sendmsg() with msg_name and 0-length data, and a SCTP_SNDRCVINFO cmsg?
>       The <draft-ietf-tsvwg-sctpsocket-02.txt> doesn't seem to address
this
> very
> specifically. Would you consider to add something to the draft to ensure
> the
> consistencies among implementations?
>
> Thanks,
> Daisy
>
> Daisy Chang
> IBM Linux Technology Center
> daisyc@us.ibm.com
> Tel: (512)-838-4194    T/L: 678-4194
> FAX: (512)-838-4663

--
Randall R. Stewart
randall@stewart.chicago.il.us 815-342-5222 (cell phone)
|
                                     |
|
                                     |
---------------------- Mailing List commands
--------------------------------
Send the following in the body of a message to
Majordomo@majordomo.cig.mot.com
SUBSCRIBE TO LIST    : subscribe list-name   | LIST OF LISTS   : lists
UNSUBSCRIBE TO LIST  : unsubscribe list-name | RETRIEVE HELP   : help
WHO ON ON THE LIST   : who list-name         | GET DESCRIPTION : info
list-name
|
|
Or you can use http://majordomo.cig.mot.com
------------------------------------------------------------------------------





_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@ns.ietf.org  Thu Jan 24 12:20:44 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id MAA28230
	for <tsvwg-archive@odin.ietf.org>; Thu, 24 Jan 2002 12:20:39 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id MAA10377
	for tsvwg-archive@odin.ietf.org; Thu, 24 Jan 2002 12:20:42 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id LAA07568;
	Thu, 24 Jan 2002 11:49:19 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id LAA07536
	for <tsvwg@ns.ietf.org>; Thu, 24 Jan 2002 11:49:14 -0500 (EST)
Received: from e21.nc.us.ibm.com (e21.nc.us.ibm.com [32.97.136.227])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id LAA26761
	for <tsvwg@ietf.org>; Thu, 24 Jan 2002 11:49:10 -0500 (EST)
Received: from southrelay02.raleigh.ibm.com (southrelay02.raleigh.ibm.com [9.37.3.209])
	by e21.nc.us.ibm.com (8.9.3/8.9.3) with ESMTP id KAA82606;
	Thu, 24 Jan 2002 10:44:49 -0600
Received: from d04nm110.raleigh.ibm.com (d04nm110.raleigh.ibm.com [9.27.5.85])
	by southrelay02.raleigh.ibm.com (8.11.1m3/NCO v5.01) with ESMTP id g0OGnBt77592;
	Thu, 24 Jan 2002 11:49:12 -0500
Importance: Normal
To: randall@stewart.chicago.il.us
Cc: tsvwg@ietf.org, sctp-developers-list@cig.mot.com
X-Mailer: Lotus Notes Release 5.0.5  September 22, 2000
Message-ID: <OFFEACB48A.2F178E48-ON85256B4B.005C0B97@raleigh.ibm.com>
From: "Daisy Chang" <daisyc@us.ibm.com>
Date: Thu, 24 Jan 2002 10:20:39 -0600
X-MIMETrack: Serialize by Router on D04NM110/04/M/IBM(Release 5.0.9 |November 16, 2001) at
 01/24/2002 11:49:11 AM
MIME-Version: 1.0
Content-type: text/plain; charset=us-ascii
Subject: [Tsvwg] Re: question regarding the SCTP API - SCTP_INIT cmsg
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

Oops, I mean "my case #2", not case #1. Sorry.

Regards,
Daisy

Daisy Chang
IBM Linux Technology Center
daisyc@us.ibm.com
Tel: (512)-838-4194    T/L: 678-4194
FAX: (512)-838-4663


---------------------- Forwarded by Daisy Chang/Austin/IBM on 01/24/2002
10:17 AM ---------------------------

Daisy Chang
01/24/2002 09:57 AM

To:    Randall Stewart <randall@stewart.chicago.il.us>
cc:    <tsvwg@ietf.org>, <sctp-developers-list@cig.mot.com>
From:  Daisy Chang/Austin/IBM@IBMUS
Subject:    Re: question regarding the SCTP API - SCTP_INIT cmsg  (Document
       link: Daisy Chang)

Randy,
      Ah, connect()! I've got the impression a while back that a definition
for
"connect() - UDP style" is contemplated. Now that you mentioned it, I
somehow
couldn't find any reference to it in the API draft. Is this something that
we should
consider to add to our UDP-style support?
      I tend to agree that a 0-length send should not do much. I'm not sure
if a EINVAL is necessary though, given that TCP and UDP both seem to take
it without complaining - as far as I can tell on Linux. I could be wrong
though
because I haven't checked the library routines to see if there is any
specific
length verification there.
      Another thing is that, to me, SCTP_INIT seems to have slightly
different
functionality than a straight connect(). For a UDP-style socket, SCTP_INIT
could provide user with finer granularity of control of those init
parameters per
connection. While setsockopt(SCTP_INITMSG) and connect() can provide
the similar results, I'm wondering, rather than requiring SCTP_INIT with
data,
if we should allow my case #1 to work as "a connect() with init parms"?
What
is your insight?


Thanks,
Daisy

Daisy Chang
IBM Linux Technology Center
daisyc@us.ibm.com
Tel: (512)-838-4194    T/L: 678-4194
FAX: (512)-838-4663


Randall Stewart<randall@stewart.chicago.il.us>@cig.mot.com on 01/24/2002
04:16:34 AM

Please respond to Randall Stewart<randall@stewart.chicago.il.us>

Sent by:    owner-sctp-developers-list@cig.mot.com


To:    Daisy Chang/Austin/IBM@IBMUS
cc:    <tsvwg@ietf.org>, <sctp-developers-list@cig.mot.com>
Subject:    Re: question regarding the SCTP API - SCTP_INIT cmsg


Daisy:

Here is my interpretation/what we did:

1) Any 0 length send that has no msgflg's to process is an
   error and return EINVAL
2) A 0 length send that has msgflgs aka ABORT/TERM get the
   shutdown to happen.
3) To start an association with no data in the UDP model we
   use connect (much like UDP but without the binding
   effect to the peer endpoint ) :>

R

Daisy Chang wrote:
>
> Hi Randy,
>       A question came up when I was implementing the SCTP_INIT cmsg in
> sendmsg() API on Linux (the lksctp project), I hope that I can find a
quick
> answer here.
>       The question is, what is the behavior of a  sendmsg() call with a
> 0-length
> message, i.e., the total length of iov_len of the msg_iov in the struct
> msghdr is 0, or,
> both the msg_iov and msg_iovlen are 0. Now, if the msg_name is specified,
> and an
> existing association is identified with it, I believe that no data should
> be sent out; we
> will process the cmsg and msg_flags if they are specified, and we are
done.
> But, if
> there is no existing association is found for the msg_name, should we go
> ahead
> establishing one?
>       Considering the following cases, is there any differences among
these
> 3 API's in terms of the effects:
> 1. sendmsg() with msg_name and 0-length data, and NULL cmsg?
> 2. sendmsg() with msg_name and 0-length data, and a SCTP_INIT cmsg?
> 3. sendmsg() with msg_name and 0-length data, and a SCTP_SNDRCVINFO cmsg?
>       The <draft-ietf-tsvwg-sctpsocket-02.txt> doesn't seem to address
this
> very
> specifically. Would you consider to add something to the draft to ensure
> the
> consistencies among implementations?
>
> Thanks,
> Daisy
>
> Daisy Chang
> IBM Linux Technology Center
> daisyc@us.ibm.com
> Tel: (512)-838-4194    T/L: 678-4194
> FAX: (512)-838-4663

--
Randall R. Stewart
randall@stewart.chicago.il.us 815-342-5222 (cell phone)
|
                                     |
|
                                     |
---------------------- Mailing List commands
--------------------------------
Send the following in the body of a message to
Majordomo@majordomo.cig.mot.com
SUBSCRIBE TO LIST    : subscribe list-name   | LIST OF LISTS   : lists
UNSUBSCRIBE TO LIST  : unsubscribe list-name | RETRIEVE HELP   : help
WHO ON ON THE LIST   : who list-name         | GET DESCRIPTION : info
list-name
|
|
Or you can use http://majordomo.cig.mot.com
------------------------------------------------------------------------------







_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@ns.ietf.org  Thu Jan 24 13:12:12 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id NAA00120
	for <tsvwg-archive@odin.ietf.org>; Thu, 24 Jan 2002 13:12:06 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id NAA13237
	for tsvwg-archive@odin.ietf.org; Thu, 24 Jan 2002 13:12:07 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id MAA10337;
	Thu, 24 Jan 2002 12:20:06 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id MAA10308
	for <tsvwg@ns.ietf.org>; Thu, 24 Jan 2002 12:20:04 -0500 (EST)
Received: from sj-msg-core-3.cisco.com (sj-msg-core-3.cisco.com [171.70.157.152])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id MAA28196
	for <tsvwg@ietf.org>; Thu, 24 Jan 2002 12:20:00 -0500 (EST)
Received: from mira-sjc5-2.cisco.com (mira-sjc5-2.cisco.com [171.71.163.16])
	by sj-msg-core-3.cisco.com (8.11.3/8.9.1) with ESMTP id g0OHJRn11651;
	Thu, 24 Jan 2002 09:19:27 -0800 (PST)
Received: from stewart.chicago.il.us (ssh-sj1.cisco.com [171.68.225.134])
	by mira-sjc5-2.cisco.com (Mirapoint)
	with ESMTP id AAV59079;
	Thu, 24 Jan 2002 09:19:17 -0800 (PST)
Message-ID: <3C504214.DDF0EDC2@stewart.chicago.il.us>
Date: Thu, 24 Jan 2002 11:19:16 -0600
From: Randall Stewart <randall@stewart.chicago.il.us>
X-Mailer: Mozilla 4.76 [en] (X11; U; Linux 2.2.12 i386)
X-Accept-Language: en
MIME-Version: 1.0
To: Daisy Chang <daisyc@us.ibm.com>
CC: tsvwg@ietf.org, sctp-developers-list@cig.mot.com
References: <OFFEACB48A.2F178E48-ON85256B4B.005C0B97@raleigh.ibm.com>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Transfer-Encoding: 7bit
Subject: [Tsvwg] Re: question regarding the SCTP API - SCTP_INIT cmsg
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit


Daisy:

Ok looking a bit further it appears that the BSD
stack under UDP will go ahead and send out
a packet with no user data... 

On the TCP side it does append the mbuf to
the TCB for output.. not sure how it
handles this down in the stack (I did not want
to dig down that far).

So the bottom line is when you do send a message with
0 length.. in UDP it may go out :>

So one may not want to accept this since the app may
expect a packet to be generated...

R


Daisy Chang wrote:
> 
> Oops, I mean "my case #2", not case #1. Sorry.
> 
> Regards,
> Daisy
> 
> Daisy Chang
> IBM Linux Technology Center
> daisyc@us.ibm.com
> Tel: (512)-838-4194    T/L: 678-4194
> FAX: (512)-838-4663
> 
> ---------------------- Forwarded by Daisy Chang/Austin/IBM on 01/24/2002
> 10:17 AM ---------------------------
> 
> Daisy Chang
> 01/24/2002 09:57 AM
> 
> To:    Randall Stewart <randall@stewart.chicago.il.us>
> cc:    <tsvwg@ietf.org>, <sctp-developers-list@cig.mot.com>
> From:  Daisy Chang/Austin/IBM@IBMUS
> Subject:    Re: question regarding the SCTP API - SCTP_INIT cmsg  (Document
>        link: Daisy Chang)
> 
> Randy,
>       Ah, connect()! I've got the impression a while back that a definition
> for
> "connect() - UDP style" is contemplated. Now that you mentioned it, I
> somehow
> couldn't find any reference to it in the API draft. Is this something that
> we should
> consider to add to our UDP-style support?
>       I tend to agree that a 0-length send should not do much. I'm not sure
> if a EINVAL is necessary though, given that TCP and UDP both seem to take
> it without complaining - as far as I can tell on Linux. I could be wrong
> though
> because I haven't checked the library routines to see if there is any
> specific
> length verification there.
>       Another thing is that, to me, SCTP_INIT seems to have slightly
> different
> functionality than a straight connect(). For a UDP-style socket, SCTP_INIT
> could provide user with finer granularity of control of those init
> parameters per
> connection. While setsockopt(SCTP_INITMSG) and connect() can provide
> the similar results, I'm wondering, rather than requiring SCTP_INIT with
> data,
> if we should allow my case #1 to work as "a connect() with init parms"?
> What
> is your insight?
> 
> Thanks,
> Daisy
> 
> Daisy Chang
> IBM Linux Technology Center
> daisyc@us.ibm.com
> Tel: (512)-838-4194    T/L: 678-4194
> FAX: (512)-838-4663
> 
> Randall Stewart<randall@stewart.chicago.il.us>@cig.mot.com on 01/24/2002
> 04:16:34 AM
> 
> Please respond to Randall Stewart<randall@stewart.chicago.il.us>
> 
> Sent by:    owner-sctp-developers-list@cig.mot.com
> 
> To:    Daisy Chang/Austin/IBM@IBMUS
> cc:    <tsvwg@ietf.org>, <sctp-developers-list@cig.mot.com>
> Subject:    Re: question regarding the SCTP API - SCTP_INIT cmsg
> 
> Daisy:
> 
> Here is my interpretation/what we did:
> 
> 1) Any 0 length send that has no msgflg's to process is an
>    error and return EINVAL
> 2) A 0 length send that has msgflgs aka ABORT/TERM get the
>    shutdown to happen.
> 3) To start an association with no data in the UDP model we
>    use connect (much like UDP but without the binding
>    effect to the peer endpoint ) :>
> 
> R
> 
> Daisy Chang wrote:
> >
> > Hi Randy,
> >       A question came up when I was implementing the SCTP_INIT cmsg in
> > sendmsg() API on Linux (the lksctp project), I hope that I can find a
> quick
> > answer here.
> >       The question is, what is the behavior of a  sendmsg() call with a
> > 0-length
> > message, i.e., the total length of iov_len of the msg_iov in the struct
> > msghdr is 0, or,
> > both the msg_iov and msg_iovlen are 0. Now, if the msg_name is specified,
> > and an
> > existing association is identified with it, I believe that no data should
> > be sent out; we
> > will process the cmsg and msg_flags if they are specified, and we are
> done.
> > But, if
> > there is no existing association is found for the msg_name, should we go
> > ahead
> > establishing one?
> >       Considering the following cases, is there any differences among
> these
> > 3 API's in terms of the effects:
> > 1. sendmsg() with msg_name and 0-length data, and NULL cmsg?
> > 2. sendmsg() with msg_name and 0-length data, and a SCTP_INIT cmsg?
> > 3. sendmsg() with msg_name and 0-length data, and a SCTP_SNDRCVINFO cmsg?
> >       The <draft-ietf-tsvwg-sctpsocket-02.txt> doesn't seem to address
> this
> > very
> > specifically. Would you consider to add something to the draft to ensure
> > the
> > consistencies among implementations?
> >
> > Thanks,
> > Daisy
> >
> > Daisy Chang
> > IBM Linux Technology Center
> > daisyc@us.ibm.com
> > Tel: (512)-838-4194    T/L: 678-4194
> > FAX: (512)-838-4663
> 
> --
> Randall R. Stewart
> randall@stewart.chicago.il.us 815-342-5222 (cell phone)
> |
>                                      |
> |
>                                      |
> ---------------------- Mailing List commands
> --------------------------------
> Send the following in the body of a message to
> Majordomo@majordomo.cig.mot.com
> SUBSCRIBE TO LIST    : subscribe list-name   | LIST OF LISTS   : lists
> UNSUBSCRIBE TO LIST  : unsubscribe list-name | RETRIEVE HELP   : help
> WHO ON ON THE LIST   : who list-name         | GET DESCRIPTION : info
> list-name
> |
> |
> Or you can use http://majordomo.cig.mot.com
> ------------------------------------------------------------------------------
> 
> |                                                                             |
> |                                                                             |
> ---------------------- Mailing List commands --------------------------------
> Send the following in the body of a message to Majordomo@majordomo.cig.mot.com
> SUBSCRIBE TO LIST    : subscribe list-name   | LIST OF LISTS   : lists
> UNSUBSCRIBE TO LIST  : unsubscribe list-name | RETRIEVE HELP   : help
> WHO ON ON THE LIST   : who list-name         | GET DESCRIPTION : info list-name
> |                                                                             |
> Or you can use http://majordomo.cig.mot.com
> ------------------------------------------------------------------------------

-- 
Randall R. Stewart
randall@stewart.chicago.il.us 815-342-5222 (cell phone)

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@ns.ietf.org  Thu Jan 24 13:30:14 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id NAA01141
	for <tsvwg-archive@odin.ietf.org>; Thu, 24 Jan 2002 13:30:14 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id NAA13833
	for tsvwg-archive@odin.ietf.org; Thu, 24 Jan 2002 13:30:16 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id NAA12980;
	Thu, 24 Jan 2002 13:05:42 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id NAA12931
	for <tsvwg@ns.ietf.org>; Thu, 24 Jan 2002 13:05:39 -0500 (EST)
Received: from sj-msg-core-2.cisco.com (sj-msg-core-2.cisco.com [171.69.24.11])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id NAA29834
	for <tsvwg@ietf.org>; Thu, 24 Jan 2002 13:05:35 -0500 (EST)
Received: from mira-sjc5-2.cisco.com (mira-sjc5-2.cisco.com [171.71.163.16])
	by sj-msg-core-2.cisco.com (8.11.3/8.9.1) with ESMTP id g0OI51d21126;
	Thu, 24 Jan 2002 10:05:02 -0800 (PST)
Received: from stewart.chicago.il.us (ssh-sj1.cisco.com [171.68.225.134])
	by mira-sjc5-2.cisco.com (Mirapoint)
	with ESMTP id AAV60857;
	Thu, 24 Jan 2002 10:04:59 -0800 (PST)
Message-ID: <3C504CCA.30C200FB@stewart.chicago.il.us>
Date: Thu, 24 Jan 2002 12:04:58 -0600
From: Randall Stewart <randall@stewart.chicago.il.us>
X-Mailer: Mozilla 4.76 [en] (X11; U; Linux 2.2.12 i386)
X-Accept-Language: en
MIME-Version: 1.0
To: Daisy Chang <daisyc@us.ibm.com>
CC: tsvwg@ietf.org, sctp-developers-list@cig.mot.com
References: <OF90F7AADB.698AC3CA-ON85256B4B.00607EC5@raleigh.ibm.com>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Transfer-Encoding: 7bit
Subject: [Tsvwg] Re: question regarding the SCTP API - SCTP_INIT cmsg
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit

Daisy:

Yes.. somewhere in the RFC it requires that if you
do recieve a Data chunk with only a header (i.e. no user
data) you MUST abort the association.

R

Daisy Chang wrote:
> 
> Yes, Randy, I think that the same behavior exists in Linux for TCP and UDP.
> Well, for TCP, appending 0 bytes to the output queue really is harmless,
> nothing is affected, so it is perfectly OK. For UDP, it seems a little
> wierd if it
> does result into a NULL message on the wire. But for a datagram protocol,
> users are taking more control than they want to in their own hands, so I
> guess
> the protocol has no reason to reject it.
> 
> As for SCTP, I think that it makes a lot of sense to disallow NULL user
> messages. I was a bit confused by the wording in 3.3.1 Payload Data
> (RFC2960) where it says "A DATA chunk with no user data field will have
> Length set to 16." But Jon Grimm graciously cleared it up by pointing me to
> the similar place in your book where it says "It should have a value equal
> to or greater than 17, because a DATA chunk is required to have at least
> one byte of user data." I gathered that this is what you refered to when
> you
> said that the RFC disallows 0-length send, right?
> 
> Thanks for your quick responses. I would suggest that we shall follow
> your interpretation for the sendmsg and connect implementation.
> 
> Thanks,
> Daisy
> 
> Daisy Chang
> IBM Linux Technology Center
> daisyc@us.ibm.com
> Tel: (512)-838-4194    T/L: 678-4194
> FAX: (512)-838-4663
> 
> Randall Stewart <randall@stewart.chicago.il.us>@cisco.com on 01/24/2002
> 11:19:16 AM
> 
> Sent by:    randall@cisco.com
> 
> To:    Daisy Chang/Austin/IBM@IBMUS
> cc:    tsvwg@ietf.org, sctp-developers-list@cig.mot.com
> Subject:    Re: question regarding the SCTP API - SCTP_INIT cmsg
> 
> Daisy:
> 
> Ok looking a bit further it appears that the BSD
> stack under UDP will go ahead and send out
> a packet with no user data...
> 
> On the TCP side it does append the mbuf to
> the TCB for output.. not sure how it
> handles this down in the stack (I did not want
> to dig down that far).
> 
> So the bottom line is when you do send a message with
> 0 length.. in UDP it may go out :>
> 
> So one may not want to accept this since the app may
> expect a packet to be generated...
> 
> R
> 
> Daisy Chang wrote:
> >
> > Oops, I mean "my case #2", not case #1. Sorry.
> >
> > Regards,
> > Daisy
> >
> > Daisy Chang
> > IBM Linux Technology Center
> > daisyc@us.ibm.com
> > Tel: (512)-838-4194    T/L: 678-4194
> > FAX: (512)-838-4663
> >
> > ---------------------- Forwarded by Daisy Chang/Austin/IBM on 01/24/2002
> > 10:17 AM ---------------------------
> >
> > Daisy Chang
> > 01/24/2002 09:57 AM
> >
> > To:    Randall Stewart <randall@stewart.chicago.il.us>
> > cc:    <tsvwg@ietf.org>, <sctp-developers-list@cig.mot.com>
> > From:  Daisy Chang/Austin/IBM@IBMUS
> > Subject:    Re: question regarding the SCTP API - SCTP_INIT cmsg
> (Document
> >        link: Daisy Chang)
> >
> > Randy,
> >       Ah, connect()! I've got the impression a while back that a
> definition
> > for
> > "connect() - UDP style" is contemplated. Now that you mentioned it, I
> > somehow
> > couldn't find any reference to it in the API draft. Is this something
> that
> > we should
> > consider to add to our UDP-style support?
> >       I tend to agree that a 0-length send should not do much. I'm not
> sure
> > if a EINVAL is necessary though, given that TCP and UDP both seem to take
> > it without complaining - as far as I can tell on Linux. I could be wrong
> > though
> > because I haven't checked the library routines to see if there is any
> > specific
> > length verification there.
> >       Another thing is that, to me, SCTP_INIT seems to have slightly
> > different
> > functionality than a straight connect(). For a UDP-style socket,
> SCTP_INIT
> > could provide user with finer granularity of control of those init
> > parameters per
> > connection. While setsockopt(SCTP_INITMSG) and connect() can provide
> > the similar results, I'm wondering, rather than requiring SCTP_INIT with
> > data,
> > if we should allow my case #1 to work as "a connect() with init parms"?
> > What
> > is your insight?
> >
> > Thanks,
> > Daisy
> >
> > Daisy Chang
> > IBM Linux Technology Center
> > daisyc@us.ibm.com
> > Tel: (512)-838-4194    T/L: 678-4194
> > FAX: (512)-838-4663
> >
> > Randall Stewart<randall@stewart.chicago.il.us>@cig.mot.com on 01/24/2002
> > 04:16:34 AM
> >
> > Please respond to Randall Stewart<randall@stewart.chicago.il.us>
> >
> > Sent by:    owner-sctp-developers-list@cig.mot.com
> >
> > To:    Daisy Chang/Austin/IBM@IBMUS
> > cc:    <tsvwg@ietf.org>, <sctp-developers-list@cig.mot.com>
> > Subject:    Re: question regarding the SCTP API - SCTP_INIT cmsg
> >
> > Daisy:
> >
> > Here is my interpretation/what we did:
> >
> > 1) Any 0 length send that has no msgflg's to process is an
> >    error and return EINVAL
> > 2) A 0 length send that has msgflgs aka ABORT/TERM get the
> >    shutdown to happen.
> > 3) To start an association with no data in the UDP model we
> >    use connect (much like UDP but without the binding
> >    effect to the peer endpoint ) :>
> >
> > R
> >
> > Daisy Chang wrote:
> > >
> > > Hi Randy,
> > >       A question came up when I was implementing the SCTP_INIT cmsg in
> > > sendmsg() API on Linux (the lksctp project), I hope that I can find a
> > quick
> > > answer here.
> > >       The question is, what is the behavior of a  sendmsg() call with a
> > > 0-length
> > > message, i.e., the total length of iov_len of the msg_iov in the struct
> > > msghdr is 0, or,
> > > both the msg_iov and msg_iovlen are 0. Now, if the msg_name is
> specified,
> > > and an
> > > existing association is identified with it, I believe that no data
> should
> > > be sent out; we
> > > will process the cmsg and msg_flags if they are specified, and we are
> > done.
> > > But, if
> > > there is no existing association is found for the msg_name, should we
> go
> > > ahead
> > > establishing one?
> > >       Considering the following cases, is there any differences among
> > these
> > > 3 API's in terms of the effects:
> > > 1. sendmsg() with msg_name and 0-length data, and NULL cmsg?
> > > 2. sendmsg() with msg_name and 0-length data, and a SCTP_INIT cmsg?
> > > 3. sendmsg() with msg_name and 0-length data, and a SCTP_SNDRCVINFO
> cmsg?
> > >       The <draft-ietf-tsvwg-sctpsocket-02.txt> doesn't seem to address
> > this
> > > very
> > > specifically. Would you consider to add something to the draft to
> ensure
> > > the
> > > consistencies among implementations?
> > >
> > > Thanks,
> > > Daisy
> > >
> > > Daisy Chang
> > > IBM Linux Technology Center
> > > daisyc@us.ibm.com
> > > Tel: (512)-838-4194    T/L: 678-4194
> > > FAX: (512)-838-4663
> >
> > --
> > Randall R. Stewart
> > randall@stewart.chicago.il.us 815-342-5222 (cell phone)
> > |
> >                                      |
> > |
> >                                      |
> > ---------------------- Mailing List commands
> > --------------------------------
> > Send the following in the body of a message to
> > Majordomo@majordomo.cig.mot.com
> > SUBSCRIBE TO LIST    : subscribe list-name   | LIST OF LISTS   : lists
> > UNSUBSCRIBE TO LIST  : unsubscribe list-name | RETRIEVE HELP   : help
> > WHO ON ON THE LIST   : who list-name         | GET DESCRIPTION : info
> > list-name
> > |
> > |
> > Or you can use http://majordomo.cig.mot.com
> >
> ------------------------------------------------------------------------------
> 
> >
> > |
> |
> > |
> |
> > ---------------------- Mailing List commands
> --------------------------------
> > Send the following in the body of a message to
> Majordomo@majordomo.cig.mot.com
> > SUBSCRIBE TO LIST    : subscribe list-name   | LIST OF LISTS   : lists
> > UNSUBSCRIBE TO LIST  : unsubscribe list-name | RETRIEVE HELP   : help
> > WHO ON ON THE LIST   : who list-name         | GET DESCRIPTION : info
> list-name
> > |
> |
> > Or you can use http://majordomo.cig.mot.com
> >
> ------------------------------------------------------------------------------
> 
> --
> Randall R. Stewart
> randall@stewart.chicago.il.us 815-342-5222 (cell phone)

-- 
Randall R. Stewart
randall@stewart.chicago.il.us 815-342-5222 (cell phone)

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@ns.ietf.org  Thu Jan 24 13:31:56 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id NAA01254
	for <tsvwg-archive@odin.ietf.org>; Thu, 24 Jan 2002 13:31:55 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id NAA14145
	for tsvwg-archive@odin.ietf.org; Thu, 24 Jan 2002 13:31:57 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id NAA12698;
	Thu, 24 Jan 2002 13:02:55 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id NAA12668
	for <tsvwg@ns.ietf.org>; Thu, 24 Jan 2002 13:02:51 -0500 (EST)
Received: from e21.nc.us.ibm.com (e21.nc.us.ibm.com [32.97.136.227])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id NAA29669
	for <tsvwg@ietf.org>; Thu, 24 Jan 2002 13:02:48 -0500 (EST)
Received: from southrelay02.raleigh.ibm.com (southrelay02.raleigh.ibm.com [9.37.3.209])
	by e21.nc.us.ibm.com (8.9.3/8.9.3) with ESMTP id LAA91442;
	Thu, 24 Jan 2002 11:58:25 -0600
Received: from d04nm110.raleigh.ibm.com (d04nm110.raleigh.ibm.com [9.27.5.85])
	by southrelay02.raleigh.ibm.com (8.11.1m3/NCO v5.01) with ESMTP id g0OI2mt194792;
	Thu, 24 Jan 2002 13:02:48 -0500
Importance: Normal
To: Randall Stewart <randall@stewart.chicago.il.us>
Cc: tsvwg@ietf.org, sctp-developers-list@cig.mot.com
X-Mailer: Lotus Notes Release 5.0.5  September 22, 2000
Message-ID: <OF90F7AADB.698AC3CA-ON85256B4B.00607EC5@raleigh.ibm.com>
From: "Daisy Chang" <daisyc@us.ibm.com>
Date: Thu, 24 Jan 2002 11:34:20 -0600
X-MIMETrack: Serialize by Router on D04NM110/04/M/IBM(Release 5.0.9 |November 16, 2001) at
 01/24/2002 01:02:48 PM
MIME-Version: 1.0
Content-type: text/plain; charset=us-ascii
Subject: [Tsvwg] Re: question regarding the SCTP API - SCTP_INIT cmsg
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

Yes, Randy, I think that the same behavior exists in Linux for TCP and UDP.
Well, for TCP, appending 0 bytes to the output queue really is harmless,
nothing is affected, so it is perfectly OK. For UDP, it seems a little
wierd if it
does result into a NULL message on the wire. But for a datagram protocol,
users are taking more control than they want to in their own hands, so I
guess
the protocol has no reason to reject it.

As for SCTP, I think that it makes a lot of sense to disallow NULL user
messages. I was a bit confused by the wording in 3.3.1 Payload Data
(RFC2960) where it says "A DATA chunk with no user data field will have
Length set to 16." But Jon Grimm graciously cleared it up by pointing me to
the similar place in your book where it says "It should have a value equal
to or greater than 17, because a DATA chunk is required to have at least
one byte of user data." I gathered that this is what you refered to when
you
said that the RFC disallows 0-length send, right?

Thanks for your quick responses. I would suggest that we shall follow
your interpretation for the sendmsg and connect implementation.

Thanks,
Daisy

Daisy Chang
IBM Linux Technology Center
daisyc@us.ibm.com
Tel: (512)-838-4194    T/L: 678-4194
FAX: (512)-838-4663


Randall Stewart <randall@stewart.chicago.il.us>@cisco.com on 01/24/2002
11:19:16 AM

Sent by:    randall@cisco.com


To:    Daisy Chang/Austin/IBM@IBMUS
cc:    tsvwg@ietf.org, sctp-developers-list@cig.mot.com
Subject:    Re: question regarding the SCTP API - SCTP_INIT cmsg



Daisy:

Ok looking a bit further it appears that the BSD
stack under UDP will go ahead and send out
a packet with no user data...

On the TCP side it does append the mbuf to
the TCB for output.. not sure how it
handles this down in the stack (I did not want
to dig down that far).

So the bottom line is when you do send a message with
0 length.. in UDP it may go out :>

So one may not want to accept this since the app may
expect a packet to be generated...

R


Daisy Chang wrote:
>
> Oops, I mean "my case #2", not case #1. Sorry.
>
> Regards,
> Daisy
>
> Daisy Chang
> IBM Linux Technology Center
> daisyc@us.ibm.com
> Tel: (512)-838-4194    T/L: 678-4194
> FAX: (512)-838-4663
>
> ---------------------- Forwarded by Daisy Chang/Austin/IBM on 01/24/2002
> 10:17 AM ---------------------------
>
> Daisy Chang
> 01/24/2002 09:57 AM
>
> To:    Randall Stewart <randall@stewart.chicago.il.us>
> cc:    <tsvwg@ietf.org>, <sctp-developers-list@cig.mot.com>
> From:  Daisy Chang/Austin/IBM@IBMUS
> Subject:    Re: question regarding the SCTP API - SCTP_INIT cmsg
(Document
>        link: Daisy Chang)
>
> Randy,
>       Ah, connect()! I've got the impression a while back that a
definition
> for
> "connect() - UDP style" is contemplated. Now that you mentioned it, I
> somehow
> couldn't find any reference to it in the API draft. Is this something
that
> we should
> consider to add to our UDP-style support?
>       I tend to agree that a 0-length send should not do much. I'm not
sure
> if a EINVAL is necessary though, given that TCP and UDP both seem to take
> it without complaining - as far as I can tell on Linux. I could be wrong
> though
> because I haven't checked the library routines to see if there is any
> specific
> length verification there.
>       Another thing is that, to me, SCTP_INIT seems to have slightly
> different
> functionality than a straight connect(). For a UDP-style socket,
SCTP_INIT
> could provide user with finer granularity of control of those init
> parameters per
> connection. While setsockopt(SCTP_INITMSG) and connect() can provide
> the similar results, I'm wondering, rather than requiring SCTP_INIT with
> data,
> if we should allow my case #1 to work as "a connect() with init parms"?
> What
> is your insight?
>
> Thanks,
> Daisy
>
> Daisy Chang
> IBM Linux Technology Center
> daisyc@us.ibm.com
> Tel: (512)-838-4194    T/L: 678-4194
> FAX: (512)-838-4663
>
> Randall Stewart<randall@stewart.chicago.il.us>@cig.mot.com on 01/24/2002
> 04:16:34 AM
>
> Please respond to Randall Stewart<randall@stewart.chicago.il.us>
>
> Sent by:    owner-sctp-developers-list@cig.mot.com
>
> To:    Daisy Chang/Austin/IBM@IBMUS
> cc:    <tsvwg@ietf.org>, <sctp-developers-list@cig.mot.com>
> Subject:    Re: question regarding the SCTP API - SCTP_INIT cmsg
>
> Daisy:
>
> Here is my interpretation/what we did:
>
> 1) Any 0 length send that has no msgflg's to process is an
>    error and return EINVAL
> 2) A 0 length send that has msgflgs aka ABORT/TERM get the
>    shutdown to happen.
> 3) To start an association with no data in the UDP model we
>    use connect (much like UDP but without the binding
>    effect to the peer endpoint ) :>
>
> R
>
> Daisy Chang wrote:
> >
> > Hi Randy,
> >       A question came up when I was implementing the SCTP_INIT cmsg in
> > sendmsg() API on Linux (the lksctp project), I hope that I can find a
> quick
> > answer here.
> >       The question is, what is the behavior of a  sendmsg() call with a
> > 0-length
> > message, i.e., the total length of iov_len of the msg_iov in the struct
> > msghdr is 0, or,
> > both the msg_iov and msg_iovlen are 0. Now, if the msg_name is
specified,
> > and an
> > existing association is identified with it, I believe that no data
should
> > be sent out; we
> > will process the cmsg and msg_flags if they are specified, and we are
> done.
> > But, if
> > there is no existing association is found for the msg_name, should we
go
> > ahead
> > establishing one?
> >       Considering the following cases, is there any differences among
> these
> > 3 API's in terms of the effects:
> > 1. sendmsg() with msg_name and 0-length data, and NULL cmsg?
> > 2. sendmsg() with msg_name and 0-length data, and a SCTP_INIT cmsg?
> > 3. sendmsg() with msg_name and 0-length data, and a SCTP_SNDRCVINFO
cmsg?
> >       The <draft-ietf-tsvwg-sctpsocket-02.txt> doesn't seem to address
> this
> > very
> > specifically. Would you consider to add something to the draft to
ensure
> > the
> > consistencies among implementations?
> >
> > Thanks,
> > Daisy
> >
> > Daisy Chang
> > IBM Linux Technology Center
> > daisyc@us.ibm.com
> > Tel: (512)-838-4194    T/L: 678-4194
> > FAX: (512)-838-4663
>
> --
> Randall R. Stewart
> randall@stewart.chicago.il.us 815-342-5222 (cell phone)
> |
>                                      |
> |
>                                      |
> ---------------------- Mailing List commands
> --------------------------------
> Send the following in the body of a message to
> Majordomo@majordomo.cig.mot.com
> SUBSCRIBE TO LIST    : subscribe list-name   | LIST OF LISTS   : lists
> UNSUBSCRIBE TO LIST  : unsubscribe list-name | RETRIEVE HELP   : help
> WHO ON ON THE LIST   : who list-name         | GET DESCRIPTION : info
> list-name
> |
> |
> Or you can use http://majordomo.cig.mot.com
>
------------------------------------------------------------------------------

>
> |
|
> |
|
> ---------------------- Mailing List commands
--------------------------------
> Send the following in the body of a message to
Majordomo@majordomo.cig.mot.com
> SUBSCRIBE TO LIST    : subscribe list-name   | LIST OF LISTS   : lists
> UNSUBSCRIBE TO LIST  : unsubscribe list-name | RETRIEVE HELP   : help
> WHO ON ON THE LIST   : who list-name         | GET DESCRIPTION : info
list-name
> |
|
> Or you can use http://majordomo.cig.mot.com
>
------------------------------------------------------------------------------


--
Randall R. Stewart
randall@stewart.chicago.il.us 815-342-5222 (cell phone)




_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@ns.ietf.org  Thu Jan 24 16:49:57 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id QAA07055
	for <tsvwg-archive@odin.ietf.org>; Thu, 24 Jan 2002 16:49:57 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id QAA25883
	for tsvwg-archive@odin.ietf.org; Thu, 24 Jan 2002 16:49:49 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id QAA22986;
	Thu, 24 Jan 2002 16:04:27 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id QAA22956
	for <tsvwg@ns.ietf.org>; Thu, 24 Jan 2002 16:04:25 -0500 (EST)
Received: from prue.eim.surrey.ac.uk (IDENT:exim@prue.eim.surrey.ac.uk [131.227.76.5])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id QAA05837
	for <tsvwg@ietf.org>; Thu, 24 Jan 2002 16:04:20 -0500 (EST)
Received: from phaestos.ee.surrey.ac.uk ([131.227.88.14] ident=eep1lw)
	by prue.eim.surrey.ac.uk with esmtp (Exim 3.33 #4)
	id 16Tr2R-0005aD-00; Thu, 24 Jan 2002 21:04:03 +0000
Date: Thu, 24 Jan 2002 21:04:00 +0000 (GMT)
From: Lloyd Wood <l.wood@eim.surrey.ac.uk>
X-X-Sender: eep1lw@phaestos.ee.surrey.ac.uk
Reply-To: Lloyd Wood <L.Wood@eim.surrey.ac.uk>
To: Randall Stewart <randall@stewart.chicago.il.us>
cc: Daisy Chang <daisyc@us.ibm.com>, <tsvwg@ietf.org>,
        <sctp-developers-list@cig.mot.com>
Subject: Re: [Tsvwg] Re: question regarding the SCTP API - SCTP_INIT cmsg
In-Reply-To: <3C504214.DDF0EDC2@stewart.chicago.il.us>
Message-ID: <Pine.SOL.4.43.0201242102420.7960-100000@phaestos.ee.surrey.ac.uk>
Organization: speaking for none
X-url: http://www.ee.surrey.ac.uk/Personal/L.Wood/
X-no-archive: yes
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
X-Scanner: exiscan *16Tr2R-0005aD-00*KNCik.lKAXc* (SECM, UniS)
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

On Thu, 24 Jan 2002, Randall Stewart wrote:

> Ok looking a bit further it appears that the BSD
> stack under UDP will go ahead and send out
> a packet with no user data...

and we've got a spare length field... UDP acks?

L.

a UDP Lite for the 21st century.

<L.Wood@surrey.ac.uk>PGP<http://www.ee.surrey.ac.uk/Personal/L.Wood/>


_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Fri Jan 25 13:28:52 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id NAA10168
	for <tsvwg-archive@odin.ietf.org>; Fri, 25 Jan 2002 13:28:51 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id NAA01483
	for tsvwg-archive@odin.ietf.org; Fri, 25 Jan 2002 13:28:53 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id MAA29255;
	Fri, 25 Jan 2002 12:46:26 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id MAA29228
	for <tsvwg@optimus.ietf.org>; Fri, 25 Jan 2002 12:46:24 -0500 (EST)
Received: from mg03.austin.ibm.com (mg03.austin.ibm.com [192.35.232.20])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id MAA08961
	for <tsvwg@ietf.org>; Fri, 25 Jan 2002 12:46:22 -0500 (EST)
Received: from austin.ibm.com (netmail1.austin.ibm.com [9.3.7.138])
	by mg03.austin.ibm.com (AIX4.3/8.9.3/8.9.3) with ESMTP id LAA10752;
	Fri, 25 Jan 2002 11:43:35 -0600
Received: from austin.ibm.com (ambika.austin.ibm.com [9.53.150.77])
	by austin.ibm.com (AIX4.3/8.9.3/8.9.3) with ESMTP id LAA39384;
	Fri, 25 Jan 2002 11:46:20 -0600
Message-ID: <3C5199EC.49339A07@austin.ibm.com>
Date: Fri, 25 Jan 2002 11:46:20 -0600
From: venkat venkatsubra <venkats@austin.ibm.com>
Organization: IBM
X-Mailer: Mozilla 4.76iC-CCK-MCD  [en_US] (X11; U; AIX 5.1)
X-Accept-Language: en
MIME-Version: 1.0
To: tsvwg@ietf.org
CC: daisyc@us.ibm.com
References: <200201251700.MAA26662@optimus.ietf.org>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Transfer-Encoding: 7bit
Subject: [Tsvwg] Re: tsvwg digest, Vol 1 #486 - 4 msgs
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit

Daisy,

>
> Message: 2
> To: Randall Stewart <randall@stewart.chicago.il.us>
> Cc: tsvwg@ietf.org, sctp-developers-list@cig.mot.com
> From: "Daisy Chang" <daisyc@us.ibm.com>
> Date: Thu, 24 Jan 2002 11:34:20 -0600
> Subject: [Tsvwg] Re: question regarding the SCTP API - SCTP_INIT cmsg
>
> Yes, Randy, I think that the same behavior exists in Linux for TCP and UDP.
> Well, for TCP, appending 0 bytes to the output queue really is harmless,
> nothing is affected, so it is perfectly OK. For UDP, it seems a little
> wierd if it
> does result into a NULL message on the wire. But for a datagram protocol,
> users are taking more control than they want to in their own hands, so I
> guess
> the protocol has no reason to reject it.

Sun has added a test to the Java Compliance Kit (suite of
testcases that all Java implementations must pass
to ship) that sends a datagram of 0 bytes length.
This works on Solaris, Linux, NT,  *BSD  and Aix.

> As for SCTP, I think that it makes a lot of sense to disallow NULL user
> messages.

Is it possible that some UDP applications ported to SCTP
might want this functionality ?

Venkat


_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@ns.ietf.org  Fri Jan 25 15:18:47 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id PAA13118
	for <tsvwg-archive@odin.ietf.org>; Fri, 25 Jan 2002 15:18:47 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id PAA07499
	for tsvwg-archive@odin.ietf.org; Fri, 25 Jan 2002 15:18:50 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id OAA06193;
	Fri, 25 Jan 2002 14:53:29 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id OAA06162
	for <tsvwg@ns.ietf.org>; Fri, 25 Jan 2002 14:53:27 -0500 (EST)
Received: from sj-msg-core-4.cisco.com (sj-msg-core-4.cisco.com [171.71.163.10])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id OAA12632
	for <tsvwg@ietf.org>; Fri, 25 Jan 2002 14:53:23 -0500 (EST)
Received: from mira-sjc5-2.cisco.com (mira-sjc5-2.cisco.com [171.71.163.16])
	by sj-msg-core-4.cisco.com (8.11.3/8.9.1) with ESMTP id g0PJqrI28271;
	Fri, 25 Jan 2002 11:52:53 -0800 (PST)
Received: from stewart.chicago.il.us (ssh-sj1.cisco.com [171.68.225.134])
	by mira-sjc5-2.cisco.com (Mirapoint)
	with ESMTP id AAV95018;
	Fri, 25 Jan 2002 11:52:50 -0800 (PST)
Message-ID: <3C51B792.F9F029BC@stewart.chicago.il.us>
Date: Fri, 25 Jan 2002 13:52:50 -0600
From: Randall Stewart <randall@stewart.chicago.il.us>
X-Mailer: Mozilla 4.76 [en] (X11; U; Linux 2.2.12 i386)
X-Accept-Language: en
MIME-Version: 1.0
To: venkat venkatsubra <venkats@austin.ibm.com>
CC: tsvwg@ietf.org, daisyc@us.ibm.com
Subject: Re: [Tsvwg] Re: tsvwg digest, Vol 1 #486 - 4 msgs
References: <200201251700.MAA26662@optimus.ietf.org> <3C5199EC.49339A07@austin.ibm.com>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Transfer-Encoding: 7bit
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit

venkat venkatsubra wrote:
> 
> Daisy,
> 
> >
> > Message: 2
> > To: Randall Stewart <randall@stewart.chicago.il.us>
> > Cc: tsvwg@ietf.org, sctp-developers-list@cig.mot.com
> > From: "Daisy Chang" <daisyc@us.ibm.com>
> > Date: Thu, 24 Jan 2002 11:34:20 -0600
> > Subject: [Tsvwg] Re: question regarding the SCTP API - SCTP_INIT cmsg
> >
> > Yes, Randy, I think that the same behavior exists in Linux for TCP and UDP.
> > Well, for TCP, appending 0 bytes to the output queue really is harmless,
> > nothing is affected, so it is perfectly OK. For UDP, it seems a little
> > wierd if it
> > does result into a NULL message on the wire. But for a datagram protocol,
> > users are taking more control than they want to in their own hands, so I
> > guess
> > the protocol has no reason to reject it.
> 
> Sun has added a test to the Java Compliance Kit (suite of
> testcases that all Java implementations must pass
> to ship) that sends a datagram of 0 bytes length.
> This works on Solaris, Linux, NT,  *BSD  and Aix.
> 
> > As for SCTP, I think that it makes a lot of sense to disallow NULL user
> > messages.
> 
> Is it possible that some UDP applications ported to SCTP
> might want this functionality ?
> 
> Venkat
Venkat:

If they do then it is a problem.. since a datagram sent
on the wire with a 0 length of user data WILL get a
ABORT back if the peer implementation is compliant to
RFC2960.

R
-- 
Randall R. Stewart
randall@stewart.chicago.il.us 815-342-5222 (cell phone)

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@ns.ietf.org  Fri Jan 25 15:42:55 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id PAA13734
	for <tsvwg-archive@odin.ietf.org>; Fri, 25 Jan 2002 15:42:55 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id PAA08920
	for tsvwg-archive@odin.ietf.org; Fri, 25 Jan 2002 15:42:58 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id PAA07033;
	Fri, 25 Jan 2002 15:07:18 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id PAA07004
	for <tsvwg@ns.ietf.org>; Fri, 25 Jan 2002 15:07:16 -0500 (EST)
Received: from e21.nc.us.ibm.com (e21.nc.us.ibm.com [32.97.136.227])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id PAA12871
	for <tsvwg@ietf.org>; Fri, 25 Jan 2002 15:07:12 -0500 (EST)
Received: from southrelay02.raleigh.ibm.com (southrelay02.raleigh.ibm.com [9.37.3.209])
	by e21.nc.us.ibm.com (8.9.3/8.9.3) with ESMTP id OAA78194
	for <tsvwg@ietf.org>; Fri, 25 Jan 2002 14:02:49 -0600
Received: from d04nm103.raleigh.ibm.com (d04nm103.raleigh.ibm.com [9.67.228.98])
	by southrelay02.raleigh.ibm.com (8.11.1m3/NCO v5.01) with ESMTP id g0PK7DS107486;
	Fri, 25 Jan 2002 15:07:13 -0500
Importance: Normal
To: venkat venkatsubra <venkats@austin.ibm.com>
Cc: tsvwg@ietf.org
From: "Daisy Chang" <daisyc@us.ibm.com>
Date: Fri, 25 Jan 2002 15:07:11 -0500
Message-ID: <OF6D98DB25.13BD089C-ON85256B4C.006D9D7B@raleigh.ibm.com>
X-MIMETrack: Serialize by Router on D04NM103/04/M/IBM(Release 5.0.8 |June 18, 2001) at
 01/25/2002 03:07:13 PM
MIME-Version: 1.0
Content-type: text/plain; charset=us-ascii
Subject: [Tsvwg] Re: tsvwg digest, Vol 1 #486 - 4 msgs
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

Venkat,
     It's a good point. Do you know if there's any particular apps which
would
rely on the 0-byte datagram? I assume that they wouldn't add this to the
Java Compliance
Kit for nothing.

Thanks,
Daisy

Daisy Chang
IBM Linux Technology Center
daisyc@us.ibm.com
Tel: (512)-838-4194    T/L: 678-4194
FAX: (512)-838-4663


venkat venkatsubra <venkats@austin.ibm.com>@austin.ibm.com on 01/25/2002
11:46:20 AM

Sent by:  venkats@austin.ibm.com


To:   tsvwg@ietf.org
cc:   Daisy Chang/Austin/IBM@IBMUS
Subject:  Re: tsvwg digest, Vol 1 #486 - 4 msgs



Daisy,

>
> Message: 2
> To: Randall Stewart <randall@stewart.chicago.il.us>
> Cc: tsvwg@ietf.org, sctp-developers-list@cig.mot.com
> From: "Daisy Chang" <daisyc@us.ibm.com>
> Date: Thu, 24 Jan 2002 11:34:20 -0600
> Subject: [Tsvwg] Re: question regarding the SCTP API - SCTP_INIT cmsg
>
> Yes, Randy, I think that the same behavior exists in Linux for TCP and
UDP.
> Well, for TCP, appending 0 bytes to the output queue really is harmless,
> nothing is affected, so it is perfectly OK. For UDP, it seems a little
> wierd if it
> does result into a NULL message on the wire. But for a datagram protocol,
> users are taking more control than they want to in their own hands, so I
> guess
> the protocol has no reason to reject it.

Sun has added a test to the Java Compliance Kit (suite of
testcases that all Java implementations must pass
to ship) that sends a datagram of 0 bytes length.
This works on Solaris, Linux, NT,  *BSD  and Aix.

> As for SCTP, I think that it makes a lot of sense to disallow NULL user
> messages.

Is it possible that some UDP applications ported to SCTP
might want this functionality ?

Venkat




_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@ns.ietf.org  Fri Jan 25 16:06:27 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id QAA14344
	for <tsvwg-archive@odin.ietf.org>; Fri, 25 Jan 2002 16:06:27 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id QAA10075
	for tsvwg-archive@odin.ietf.org; Fri, 25 Jan 2002 16:06:30 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id PAA08669;
	Fri, 25 Jan 2002 15:37:13 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id PAA08642
	for <tsvwg@ns.ietf.org>; Fri, 25 Jan 2002 15:37:11 -0500 (EST)
Received: from mg03.austin.ibm.com (mg03.austin.ibm.com [192.35.232.20])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id PAA13577
	for <tsvwg@ietf.org>; Fri, 25 Jan 2002 15:37:07 -0500 (EST)
Received: from austin.ibm.com (netmail1.austin.ibm.com [9.3.7.138])
	by mg03.austin.ibm.com (AIX4.3/8.9.3/8.9.3) with ESMTP id OAA18074;
	Fri, 25 Jan 2002 14:34:21 -0600
Received: from austin.ibm.com (death.austin.ibm.com [9.53.216.109])
	by austin.ibm.com (AIX4.3/8.9.3/8.9.3) with ESMTP id OAA27006;
	Fri, 25 Jan 2002 14:37:06 -0600
Message-ID: <3C51C0B9.8CF5AE5D@austin.ibm.com>
Date: Fri, 25 Jan 2002 14:31:53 -0600
From: Jon Grimm <jgrimm@austin.ibm.com>
X-Mailer: Mozilla 4.76 [en] (X11; U; Linux 2.4.1 i686)
X-Accept-Language: en
MIME-Version: 1.0
To: Randall Stewart <randall@stewart.chicago.il.us>
CC: venkat venkatsubra <venkats@austin.ibm.com>, tsvwg@ietf.org,
        daisyc@us.ibm.com
Subject: Re: [Tsvwg] Re: tsvwg digest, Vol 1 #486 - 4 msgs
References: <200201251700.MAA26662@optimus.ietf.org> <3C5199EC.49339A07@austin.ibm.com> <3C51B792.F9F029BC@stewart.chicago.il.us>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Transfer-Encoding: 7bit
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit

Randall Stewart wrote:
> 
> venkat venkatsubra wrote:
> >
> > Daisy,
> >
> > >
> > > Message: 2
> > > To: Randall Stewart <randall@stewart.chicago.il.us>
> > > Cc: tsvwg@ietf.org, sctp-developers-list@cig.mot.com
> > > From: "Daisy Chang" <daisyc@us.ibm.com>
> > > Date: Thu, 24 Jan 2002 11:34:20 -0600
> > > Subject: [Tsvwg] Re: question regarding the SCTP API - SCTP_INIT cmsg
> > >
> > > Yes, Randy, I think that the same behavior exists in Linux for TCP and UDP.
> > > Well, for TCP, appending 0 bytes to the output queue really is harmless,
> > > nothing is affected, so it is perfectly OK. For UDP, it seems a little
> > > wierd if it
> > > does result into a NULL message on the wire. But for a datagram protocol,
> > > users are taking more control than they want to in their own hands, so I
> > > guess
> > > the protocol has no reason to reject it.
> >
> > Sun has added a test to the Java Compliance Kit (suite of
> > testcases that all Java implementations must pass
> > to ship) that sends a datagram of 0 bytes length.
> > This works on Solaris, Linux, NT,  *BSD  and Aix.
> >
> > > As for SCTP, I think that it makes a lot of sense to disallow NULL user
> > > messages.
> >
> > Is it possible that some UDP applications ported to SCTP
> > might want this functionality ?
> >
> > Venkat
> Venkat:
> 
> If they do then it is a problem.. since a datagram sent
> on the wire with a 0 length of user data WILL get a
> ABORT back if the peer implementation is compliant to
> RFC2960.
> 


There are probably two cases for the 0 length UDP sender.   One that
cares if a packet is actually put on the wire and one that doesn't, but
has been lazy and not realized they are requesting a 0 length send (and
count on a successful return value in doing so).


The first case (if it truly exists) is going to have to find some other
solution anyway as SCTP must not do 0-length payloads.   The second
doesn't care, so one could return success in the SCTP implementation (or
in the java library if that's where one only cares).       

Best Regards,
Jon

> R
> --
> Randall R. Stewart
> randall@stewart.chicago.il.us 815-342-5222 (cell phone)
> 
> _______________________________________________
> tsvwg mailing list
> tsvwg@ietf.org
> https://www1.ietf.org/mailman/listinfo/tsvwg

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@ns.ietf.org  Fri Jan 25 18:39:51 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id SAA18079
	for <tsvwg-archive@odin.ietf.org>; Fri, 25 Jan 2002 18:39:46 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id SAA17993
	for tsvwg-archive@odin.ietf.org; Fri, 25 Jan 2002 18:39:49 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id SAA16971;
	Fri, 25 Jan 2002 18:14:08 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id SAA16942
	for <tsvwg@ns.ietf.org>; Fri, 25 Jan 2002 18:14:06 -0500 (EST)
Received: from intl-bjc-cn-1.inter-touch.net ([211.101.143.98])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id SAA17705
	for <tsvwg@ietf.org>; Fri, 25 Jan 2002 18:13:58 -0500 (EST)
Received: from littlejoy ([10.5.0.17])
	by intl-bjc-cn-1.inter-touch.net (8.9.3/8.9.3) with SMTP id HAA23059;
	Sat, 26 Jan 2002 07:13:23 +0800
From: "Douglas Otis" <dotis@sanlight.net>
To: <ips@ece.cmu.edu>
Cc: "Tsvwg" <tsvwg@ietf.org>
Date: Fri, 25 Jan 2002 15:13:08 -0800
Message-ID: <NEBBJGDMMLHHCIKHGBEJEEPJCNAA.dotis@sanlight.net>
MIME-Version: 1.0
Content-Type: text/plain;
	charset="iso-8859-1"
Content-Transfer-Encoding: 7bit
X-Priority: 3 (Normal)
X-MSMail-Priority: Normal
X-Mailer: Microsoft Outlook IMO, Build 9.0.2416 (9.0.2911.0)
In-Reply-To: <277DD60FB639D511AC0400B0D068B71E029373CB@CORPMX14>
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4807.1700
Importance: Normal
Content-Transfer-Encoding: 7bit
Subject: [Tsvwg] RE: iSCSI: Framing Steps
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit

David,

Thank you for the response.  I think a further point should have been
strongly made.  Such changes to TCP are not actions of the IPS workgroup.
These matters should be discussed within the tsvwg.  Even matters of Fixed
Interval Marking brings concerns with respect to the impact even this
seemingly innocuous iSCSI option may have.

Should the interval within the FIM be a power of 2, guessing an initial
sequence value allows injection of packets within existing connections.  A
layer below TCP, as devised for iSCSI, is to place de-encapsulated iSCSI
information from isolated packets that include marker information  despite
prior packet loss.  A general goal of all the framing approaches being
sought by the IPS wg.

This would require extensive modification of TCP to facilitate this
behavior.  In addition, without the specific details of this scheme, largely
due to a desire to claim no changes, it is difficult to evaluate related
security risks.  One risk of this scheme is the requirement for application
specific code to be routinely placed below the transport.  There are no
conventions for instituting the requisite signaling between this below the
transport layer application process and the modified TCP.  Should an attack
successfully inject even null structures, it would not be clear how the
transport would behave.

FIM has been suggested by those within the IPS wg that implementers consider
implementing an otherwise obnoxious scheme requiring the injection of
markers or suffer from discarded data otherwise retained using standard
versions of TCP.  Clearly just sending these markers are not in themselves a
risk to TCP.  It is their intended use that causes concern.  This FIM has
also been indicated by several including the author of iSCSI that FIM is
just an interim measure to be followed by a full framing scheme later.  It
is clear this group wishes extensive modifications to TCP and does not wish
to share these details openly.

One suggestion was to have each iSCSI PDU become framed by a TCP segment.
Just the reduction in the average packet size within a high bandwidth stream
will stress standard versions of TCP as well as the intervening equipment.
SCTP had the foresight to include bundling methods.  This is just one aspect
of these proposals to fail consideration within the IPS wg.  In general, I
think the IETF made a wise choice in proposing SCTP as the means of
incorporating framing within IP protocols and the IPS wg was informed of
SCTP at the outset of their endeavors.  SCTP provides many advantages with
respect to pressing concerns especially the less disruptive nature
incorporation of framing using SCTP would cause over an attempt of framing
within TCP.  It is clear there is consensus within the IPS wg that SCTP is
not to be considered and that TCP framing is despite prohibitions within the
charter and the good judgement of the IETF.  The ongoing tacit approval of
these discussions regarding modification of TCP within the IPS wg is
troubling.

Doug

> -----Original Message-----
> From: owner-ips@ece.cmu.edu [mailto:owner-ips@ece.cmu.edu] On Behalf Of
> Black_David@emc.com [mailto:David_Black@emc.com]
> Sent: Wednesday, January 23, 2002 11:10 PM
> To: ips@ece.cmu.edu
> Subject: iSCSI: Framing Steps
>
>
> I want to attempt to make some steps towards resolving
> the framing issues.  In reviewing the recent discussions
> on framing, I have a couple of conclusions:
>
> (1) I do not believe that WG consensus (rough or otherwise)
> 	can be obtained for a "MUST implement" requirement
> 	for any form of framing.
> (2) The COWS mechanism has a lot of potential, but
> 	its newness, the multiple versions that
> 	have been mentioned, and the desire for some
> 	sort of alignment with new work on DDP/RDMA
> 	suggest that COWS is premature to specify as
> 	part of iSCSI.
>
> I suggest that these conclusions form the
> basis for further ips WG consideration of framing.
>
> Please think carefully before objecting to these
> conclusions on the list (I'm happy to respond to
> private questions/expressions of concern).  If the
> framing issue cannot be driven to closure in
> the next few weeks, I will be forced to conclude
> that the entire topic is experimental, and hence
> needs to be removed from the iSCSI specification
> and handled in separate drafts intended to become
> experimental RFCs.
>
> Thanks,
> --David
>
> p.s. A desire to build NICs that never behave in
> accordance with an important SHOULD in RFC 1122
> (out-of-order segments SHOULD be queued) does not
> strike me as a good reason for changing the first
> conclusion above.
>
> ---------------------------------------------------
> David L. Black, Senior Technologist
> EMC Corporation, 42 South St., Hopkinton, MA  01748
> +1 (508) 249-6449 *NEW*      FAX: +1 (508) 497-8500
> black_david@emc.com         Cell: +1 (978) 394-7754
> ---------------------------------------------------
>


_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Sat Jan 26 03:59:13 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id DAA04025
	for <tsvwg-archive@odin.ietf.org>; Sat, 26 Jan 2002 03:59:13 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id DAA13883
	for tsvwg-archive@odin.ietf.org; Sat, 26 Jan 2002 03:59:16 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id DAA13748;
	Sat, 26 Jan 2002 03:45:16 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id DAA13720
	for <tsvwg@optimus.ietf.org>; Sat, 26 Jan 2002 03:45:14 -0500 (EST)
Received: from mxic2.us.dg.com (mxic2.isus.emc.com [128.221.31.40])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id DAA03933
	for <tsvwg@ietf.org>; Sat, 26 Jan 2002 03:45:09 -0500 (EST)
From: Black_David@emc.com
Received: by mxic2.us.dg.com with Internet Mail Service (5.5.2653.19)
	id <ZRAVRRG2>; Sat, 26 Jan 2002 03:39:46 -0500
Message-ID: <277DD60FB639D511AC0400B0D068B71E039C7EAB@CORPMX14>
To: dotis@sanlight.net, ips@ece.cmu.edu
Cc: tsvwg@ietf.org
Date: Sat, 26 Jan 2002 03:31:25 -0500
MIME-Version: 1.0
X-Mailer: Internet Mail Service (5.5.2653.19)
Content-Type: text/plain;
	charset="iso-8859-1"
Subject: [Tsvwg] RE: iSCSI: Framing Steps
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

Attempting to kill a few red herrings ...

TCP is vulnerable to packet injection based on guessing or observing
the initial sequence number in the absence of fixed-interval-markers
(FIM).  The well-known connection hijack attack is an example.  An
attacker can inject arbitrary data into a connection without FIM and
cause TCP to misbehave as a result.  FIM does not make this any
easier or change the ways in which TCP would misbehave when attacked
in this fashion.

FIM does not entail any changes to TCP.  TCP does not care about
where incoming data is placed in memory - it cares about the order
in which data is processed by TCP and delivered to the application.
The intended usage of FIM affects only the placement of incoming
data in memory and hence does not change TCP.

Meanwhile, Doug appears to have resumed his smear campaign against
the IPS wg.  Here are three examples:

> FIM has been suggested by those within the IPS wg that implementers
consider
> implementing an otherwise obnoxious scheme requiring the injection of
> markers or suffer from discarded data otherwise retained using standard
> versions of TCP.

Indeed it is obnoxious, and Doug conveniently omitted the fact that the
original message in this thread contains a strong suggestion from this
IPS wg chair that such an implementation approach is a bad idea because
it violates an important SHOULD in RFC 1122 (SHOULD queue out of order
data).
Scroll to the bottom of this email - it's still there ...

> It is clear this group wishes extensive modifications to TCP and 
> does not wish to share these details openly.

Extensive is in the eye of the beholder, but ALL of the proposed framing
schemes have been openly described on the IPS and/or TSVWG lists and/or
in Internet-Drafts.

> One suggestion was to have each iSCSI PDU become framed by a TCP segment.
> Just the reduction in the average packet size within a high bandwidth
stream
> will stress standard versions of TCP as well as the intervening equipment.
> SCTP had the foresight to include bundling methods.  This is just one
aspect
> of these proposals to fail consideration within the IPS wg.

This refers to draft-ietf-tsvwg-tcp-ulp-frame-01.txt, a tsvwg that
proposes changing TCP.  Despite Doug's proclamations that the IPS wg
should not be discussing changes to TCP, he now criticizes the IPS
wg for failing to discuss an aspect of a change to TCP???  There's
just no pleasing some people ... especially as I don't recall this
specific issue being raised in the past.

Thanks,
--David

---------------------------------------------------
David L. Black, Senior Technologist
EMC Corporation, 42 South St., Hopkinton, MA  01748
+1 (508) 249-6449 *NEW*      FAX: +1 (508) 497-8500
black_david@emc.com         Cell: +1 (978) 394-7754
---------------------------------------------------

> -----Original Message-----
> From: Douglas Otis [mailto:dotis@sanlight.net]
> Sent: Friday, January 25, 2002 6:13 PM
> To: ips@ece.cmu.edu
> Cc: Tsvwg
> Subject: RE: iSCSI: Framing Steps
> 
> 
> David,
> 
> Thank you for the response.  I think a further point should have been
> strongly made.  Such changes to TCP are not actions of the IPS workgroup.
> These matters should be discussed within the tsvwg.  Even matters of Fixed
> Interval Marking brings concerns with respect to the impact even this
> seemingly innocuous iSCSI option may have.
> 
> Should the interval within the FIM be a power of 2, guessing an initial
> sequence value allows injection of packets within existing connections.  A
> layer below TCP, as devised for iSCSI, is to place de-encapsulated iSCSI
> information from isolated packets that include marker information  despite
> prior packet loss.  A general goal of all the framing approaches being
> sought by the IPS wg.
> 
> This would require extensive modification of TCP to facilitate this
> behavior.  In addition, without the specific details of this scheme,
largely
> due to a desire to claim no changes, it is difficult to evaluate related
> security risks.  One risk of this scheme is the requirement for
application
> specific code to be routinely placed below the transport.  There are no
> conventions for instituting the requisite signaling between this below the
> transport layer application process and the modified TCP.  Should an
attack
> successfully inject even null structures, it would not be clear how the
> transport would behave.
> 
> FIM has been suggested by those within the IPS wg that implementers
consider
> implementing an otherwise obnoxious scheme requiring the injection of
> markers or suffer from discarded data otherwise retained using standard
> versions of TCP.  Clearly just sending these markers are not in themselves
a
> risk to TCP.  It is their intended use that causes concern.  This FIM has
> also been indicated by several including the author of iSCSI that FIM is
> just an interim measure to be followed by a full framing scheme later.  It
> is clear this group wishes extensive modifications to TCP and does not
wish
> to share these details openly.
> 
> One suggestion was to have each iSCSI PDU become framed by a TCP segment.
> Just the reduction in the average packet size within a high bandwidth
stream
> will stress standard versions of TCP as well as the intervening equipment.
> SCTP had the foresight to include bundling methods.  This is just one
aspect
> of these proposals to fail consideration within the IPS wg.  In general, I
> think the IETF made a wise choice in proposing SCTP as the means of
> incorporating framing within IP protocols and the IPS wg was informed of
> SCTP at the outset of their endeavors.  SCTP provides many advantages with
> respect to pressing concerns especially the less disruptive nature
> incorporation of framing using SCTP would cause over an attempt of framing
> within TCP.  It is clear there is consensus within the IPS wg that SCTP is
> not to be considered and that TCP framing is despite prohibitions within
the
> charter and the good judgement of the IETF.  The ongoing tacit approval of
> these discussions regarding modification of TCP within the IPS wg is
troubling.
> 
> Doug
> 
> > -----Original Message-----
> > From: owner-ips@ece.cmu.edu [mailto:owner-ips@ece.cmu.edu] 
> On Behalf Of
> > Black_David@emc.com [mailto:David_Black@emc.com]
> > Sent: Wednesday, January 23, 2002 11:10 PM
> > To: ips@ece.cmu.edu
> > Subject: iSCSI: Framing Steps
> >
> >
> > I want to attempt to make some steps towards resolving
> > the framing issues.  In reviewing the recent discussions
> > on framing, I have a couple of conclusions:
> >
> > (1) I do not believe that WG consensus (rough or otherwise)
> > 	can be obtained for a "MUST implement" requirement
> > 	for any form of framing.
> > (2) The COWS mechanism has a lot of potential, but
> > 	its newness, the multiple versions that
> > 	have been mentioned, and the desire for some
> > 	sort of alignment with new work on DDP/RDMA
> > 	suggest that COWS is premature to specify as
> > 	part of iSCSI.
> >
> > I suggest that these conclusions form the
> > basis for further ips WG consideration of framing.
> >
> > Please think carefully before objecting to these
> > conclusions on the list (I'm happy to respond to
> > private questions/expressions of concern).  If the
> > framing issue cannot be driven to closure in
> > the next few weeks, I will be forced to conclude
> > that the entire topic is experimental, and hence
> > needs to be removed from the iSCSI specification
> > and handled in separate drafts intended to become
> > experimental RFCs.
> >
> > Thanks,
> > --David
> >
> > p.s. A desire to build NICs that never behave in
> > accordance with an important SHOULD in RFC 1122
> > (out-of-order segments SHOULD be queued) does not
> > strike me as a good reason for changing the first
> > conclusion above.
> >
> > ---------------------------------------------------
> > David L. Black, Senior Technologist
> > EMC Corporation, 42 South St., Hopkinton, MA  01748
> > +1 (508) 249-6449 *NEW*      FAX: +1 (508) 497-8500
> > black_david@emc.com         Cell: +1 (978) 394-7754
> > ---------------------------------------------------
> >
> 

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Sat Jan 26 12:09:20 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id MAA08950
	for <tsvwg-archive@odin.ietf.org>; Sat, 26 Jan 2002 12:09:19 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id MAA29349
	for tsvwg-archive@odin.ietf.org; Sat, 26 Jan 2002 12:09:24 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id LAA27989;
	Sat, 26 Jan 2002 11:54:45 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id LAA27958
	for <tsvwg@optimus.ietf.org>; Sat, 26 Jan 2002 11:54:43 -0500 (EST)
Received: from localhost ([211.117.232.146])
	by ietf.org (8.9.1a/8.9.1a) with SMTP id LAA08782
	for <tsvwg@ietf.org>; Sat, 26 Jan 2002 11:54:36 -0500 (EST)
Message-Id: <200201261654.LAA08782@ietf.org>
Reply-To: office7000@yahoo.co.kr
From: 창업정보<office7000@yahoo.co.kr>
To: tsvwg@ietf.org
Mime-Version: 1.0
Content-Type: text/html; charset="ks_c_5601-1987"
Date: Sun, 27 Jan 2002 01:55:01 +0900
Subject: [Tsvwg] [광고] 돈 되는 사업정보입니다
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

<html>

<head>
<meta http-equiv="content-type" content="text/html; charset=euc-kr">
<title>귀하의 창업을 돕겠습니다.  </title>
<meta name="generator" content="Namo WebEditor v5.0">
<meta name="description" content="한 문단으로 된 본문이 있는 문서 요소입니다.">
</head>

<body bgcolor="white" text="black" link="blue" vlink="purple" alink="red">
                <table border="0" width="621" align="center" bgcolor="white">
                    <tr>
                        <td colspan="2" width="617">
            <p align="center">                            <a href="http://700office.com"><img src="http://www.700office.com/board/icon/gu.gif" width="295" height="43" border="0"></a></p>
                        </td>
                    </tr>
                    <tr>
                        <td width="19">
                            <p>&nbsp;</p>
                        </td>
                        <td width="596" bgcolor="#F6F6F6">
                            <p>&nbsp;<font face="돋움" size="2">귀하의 창업을 
                            돕겠습니다. &nbsp;</font></p>
            <p><font face="돋움" size="2">&nbsp;저희 (주)나래정보에서는 전화정보 
                            창업자를 모십니다. </font></p>
            <p><font face="돋움" size="2">&nbsp;본 &nbsp;사업소에서 보유한 
                            아이템을 드릴수도 있습니다. </font></p>
            <p><font face="돋움" size="2">귀하께서 계획하신 아이템을 
                            수익성 있는 사업으로 만들어 드리겠습니다. </font></p>
            <p><font face="돋움" size="2">수년간의 
                            기술과 노하우로 최선을 다하며, 고객의 이익과 사업의 
                            발전을 위해 노력하는 기업으로</font></p>
            <p> <font face="돋움" size="2">귀하의 곁에 끝까지 
                            남겠습니다.</font></p>
            <p style="line-height:20px;" align="left"><font size="4" face="돋움" color="blue"><b>전화정보(700, 
                            060, 800, 0600)특징</b></font></p>
                            <p style="line-height:5px;" align="left"><font size="2" face="굴림">&nbsp;&nbsp;</font><font size="2" face="돋움" color="#0066FF">1. 
                            특별한 자격을 요하지 않습니다.(사업자 등록증 필요)</font></p>
                            <p style="line-height:5px;" align="left"><font size="2" face="돋움" color="#0066FF">&nbsp;&nbsp;2. 
                            수금의 필요성이 없습니다.(각 통신사 수금대행)</font></p>
                            <p style="line-height:5px;" align="left"><font size="2" face="돋움" color="#0066FF">&nbsp;&nbsp;3. 
                            타 사업보다 운영과 관리가 편리합니다.(전화로 관리가능)</font></p>
                            <p style="line-height:5px;" align="left"><font size="2" face="돋움" color="#0066FF">&nbsp;&nbsp;4. 
                            귀하만의 노하우로 정보를 제공합니다.(아이템제공)</font></p>
                            <p style="line-height:5px;" align="left"><font size="2" face="돋움" color="#0066FF">&nbsp;&nbsp;5. 
                            사업성 타당여부 조사.(시장성 조사 대행)</font></p>
                            <p style="line-height:5px;" align="left"><font size="2" face="돋움" color="#0066FF">&nbsp;&nbsp;6. 
                            창업자금지원.(창업자금의 50%지원, 수익금 창출후에 
                            수금)</font></p>
                            <p style="line-height:5px;" align="left"><font size="2" face="돋움" color="#0066FF">&nbsp;&nbsp;7. 
                            사업의 지속성.(시스템 본사 관리. 사업자는 기획과 
                            광고)</font></p>
                            <p style="line-height:5px;" align="left"><font size="2" face="돋움" color="#0066FF">&nbsp;&nbsp;8. 
                            사업의 안전성.(한국통신, 하나로통신, 데이콤등과 
                            연계하여 사업영위)</font></p>
                            <p style="line-height:5px;" align="left"><font size="2" face="돋움" color="#0066FF">&nbsp;&nbsp;9. 
                            본사의 지원.((주)나래에서 사업자등록신청, 시스템제작, 
                            운영, 관리)</font></p>
                            <p style="line-height:5px;" align="left"><font size="2" face="돋움" color="#0066FF">&nbsp;10. 
                            장비임대.(금3,100,000원으로 장비일체와 프로그램을 
                            제공합니다.) </font></p>
            <p style="line-height:5px;" align="left">&nbsp;</p>
<TABLE cellSpacing=0 cellPadding=0 width="539" align=center border=0 bordercolor="#CCCCCC">
<TBODY>
<TR>
<TD width="237" bgColor=#eeeef7 height="65">
<P style="MARGIN-TOP: 0px; MARGIN-BOTTOM: 0px; LINE-HEIGHT: 150%"><img src="http://www.700office.com/neww.gif" width="50" height="15" border="0"> 
                                        <a href="mailto:skylove0028@dreamwiz.com"><font color="blue">&quot;</font><font color="blue" size="2">광고메일대행</font></a><span style="font-size:9pt;"><a href="mailto:skylove0028@dreamwiz.com"><font face="돋움" color="blue">&quot; &nbsp;링크해주세요 
                                        </font></a></span><FONT 
color=#0033ff><SPAN style="FONT-SIZE: 9pt">&nbsp;</SPAN></FONT> 
</P></TD>
<TD width="302" height="65">
<SPAN style="FONT-SIZE: 9pt"><FONT face=돋움>가장 빠른 
광고효과입니다. 빠른 시간안에 확실한 효과를 보실수 있습니다. 이보다 &nbsp;뛰어난 
광고효과를 기대할 수 없습니다. 저희 회원에 한하여 후불로 결재 받습니다.</FONT></SPAN></TD></TR>
<TR>
<TD width="237">
&nbsp;</TD>
<TD width="302">
<TABLE cellSpacing=0 cellPadding=0 width=288 border=0>
<TBODY>
<TR>
<TD width=10 height="11">&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; </TD>
<TD width=278 height="11">&nbsp;</TD></TR></TBODY></TABLE>
<P></P></TD></TR></TBODY></TABLE>
<TABLE width="100%" bgColor=#ffffff text="#000000">
<TBODY>
<TR>
<TD vAlign=top width="588">
<TABLE cellSpacing=0 width=590 border=0>
<TBODY>
<TR>
<TD bgColor=#999999>
<TABLE cellSpacing=0 width="100%" border="1" bordercolordark="black" bordercolorlight="black">
<TBODY>
<TR>
<TD vAlign=top align=middle bgColor=#ffffff>                            <p style="line-height:20px;" align="center"><SPAN style="FONT-SIZE: 9pt">&nbsp;귀하의 메일주소는 웹서핑중, <B>http://www.xxxxxxxx.com/</B> 
<BR>에서 알게 된것이며, E-Mail 주소 외에, 다른 정보는 갖고 있지 않습니다.<BR>정통부 권고사항에 의거 제목에 
</SPAN><B><SPAN style="FONT-SIZE: 9pt">[광고]</SPAN></B><SPAN style="FONT-SIZE: 9pt">라고 표기한 메일입니다. 원치 않으면 </SPAN><A 
href="mailto:joki@borahome.net?subject=수신거부" target=new><B><SPAN 
style="FONT-SIZE: 9pt">,</SPAN></B></A><A 
href="mailto:office7000@yahoo.co.kr?subject=수신거부" target=new><B><SPAN 
style="FONT-SIZE: 9pt"><FONT color=red>수신거부</FONT></SPAN></B></A><SPAN style="FONT-SIZE: 9pt">를 눌러주세요</SPAN> 
</p>
</TD></TR></TBODY></TABLE></TD></TR></TBODY></TABLE></TD></TR></TBODY></TABLE>
            <p style="line-height:20px;" align="center"><SPAN style="FONT-SIZE: 9pt">&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;</SPAN></p>
                        </td>
                    </tr>
                </table>
<p>&nbsp;</p>
</body>




</html>

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Sun Jan 27 19:48:37 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id TAA05209
	for <tsvwg-archive@odin.ietf.org>; Sun, 27 Jan 2002 19:48:37 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id TAA02331
	for tsvwg-archive@odin.ietf.org; Sun, 27 Jan 2002 19:48:39 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id TAA01373;
	Sun, 27 Jan 2002 19:22:31 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id TAA01345
	for <tsvwg@optimus.ietf.org>; Sun, 27 Jan 2002 19:22:29 -0500 (EST)
Received: from c003.snv.cp.net (c003-h000.c003.snv.cp.net [209.228.32.214])
	by ietf.org (8.9.1a/8.9.1a) with SMTP id TAA05071
	for <tsvwg@ietf.org>; Sun, 27 Jan 2002 19:22:27 -0500 (EST)
Received: (cpmta 13006 invoked from network); 27 Jan 2002 16:21:57 -0800
Received: from 64.130.130.105 (HELO dougrmt)
  by smtp.telocity.com (209.228.32.214) with SMTP; 27 Jan 2002 16:21:57 -0800
X-Sent: 28 Jan 2002 00:21:57 GMT
From: "Douglas Otis" <dotis@sanlight.net>
To: <Black_David@emc.com>, <ips@ece.cmu.edu>
Cc: <tsvwg@ietf.org>
Date: Sun, 27 Jan 2002 16:23:14 -0800
Message-ID: <NEBBJGDMMLHHCIKHGBEJOEDGCPAA.dotis@sanlight.net>
MIME-Version: 1.0
Content-Type: text/plain;
	charset="iso-8859-1"
Content-Transfer-Encoding: 7bit
X-Priority: 3 (Normal)
X-MSMail-Priority: Normal
X-Mailer: Microsoft Outlook IMO, Build 9.0.2416 (9.0.2910.0)
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4807.1700
Importance: Normal
In-Reply-To: <277DD60FB639D511AC0400B0D068B71E039C7EAB@CORPMX14>
Content-Transfer-Encoding: 7bit
Subject: [Tsvwg] RE: iSCSI: Framing Steps
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit

David,

> Attempting to kill a few red herrings ...

A "red herring" is irrelevance offered to distract.  For red herrings, your
statements are examples.

> TCP is vulnerable to packet injection based on guessing or observing
> the initial sequence number in the absence of fixed-interval-markers
> (FIM).  The well-known connection hijack attack is an example.  An
> attacker can inject arbitrary data into a connection without FIM and
> cause TCP to misbehave as a result.  FIM does not make this any
> easier or change the ways in which TCP would misbehave when attacked
> in this fashion.

FIM enables *processing* of packets out of sequence from the initial send.
The intent of FIM is to make such processing possible and that *does* enable
a greater opportunity for attack especially if the measure is predictable
within future headers.

> FIM does not entail any changes to TCP.  TCP does not care about
> where incoming data is placed in memory - it cares about the order
> in which data is processed by TCP and delivered to the application.

A process that examines application specific structures at the packet (IP)
level, to then move portions of the expected TCP byte stream into user space
based on application references matching these structures, is *not* a
modification to TCP provided there is a *sequential* order of delivery?
Clearly placing a portion of the application below TCP is a modification to
TCP and network layering.  Are you referring to a means at the TCP API to
hide this activity?  Are you saying that the TCP API does not change?  How
is this done without changing TCP?

The application specific processing below this modified TCP transport is to
de-encapsulate the iSCSI payloads into user space. iSCSI user space may not
be satisfied in the sequence they were established.  The complexity involved
will require modification of TCP to support this buffer handling and
application specific structure processing.  In the case of iSCSI, the
content of the byte stream is not known before hand.  This activity below
TCP places and splices sections of the TCP byte stream into designated iSCSI
user space.  Clearly there are many questions one may have with respect to
what and when TCP validates and delivers.  TCP could easily reject a packet
based on the byte sequence within headers that was not rejected by this
"below the transport process" that would be unable to see the header
continuum.  This added state below the TCP transport provides opportunity
for attack or in some implementations, poor behavior.

> The intended usage of FIM affects only the placement of incoming
> data in memory and hence does not change TCP.

Placement of application specific payloads into regions matched against
application specific structure content prior to being handled by the TCP
layer is a change to TCP.  TCP is not able to manipulate buffers in this
manner without extensive modification.  Such modification is quite possible
however, and all the details should be explored before the use of this
scheme is placed into a standards draft however.  My concern remains that
the IPS wg is NOT the place to discuss these concerns as that is not their
expertise nor their charter.

> Meanwhile, Doug appears to have resumed his smear campaign against
> the IPS wg.  Here are three examples:
>
> > FIM has been suggested by those within the IPS wg that implementers
> > consider implementing an otherwise obnoxious scheme requiring the
> > injection of markers or suffer from discarded data otherwise
> > retained using standard versions of TCP.
>
> Indeed it is obnoxious, and Doug conveniently omitted the fact that the
> original message in this thread contains a strong suggestion from this
> IPS wg chair that such an implementation approach is a bad idea because
> it violates an important SHOULD in RFC 1122 (SHOULD queue out of order
> data).
> Scroll to the bottom of this email - it's still there ...

As you point out, I did include such a statement made by you.  You have also
asked the group to drive the framing issue to conclusion but you failed to
indicate that such discussion should not take place within the IPS.  Ensure
that an error is not made where everyone expects such modifications to TCP
even if done unofficially by the IPS.

> > It is clear this group wishes extensive modifications to TCP and
> > does not wish to share these details openly.
>
> Extensive is in the eye of the beholder, but ALL of the proposed framing
> schemes have been openly described on the IPS and/or TSVWG lists and/or
> in Internet-Drafts.

That is a valid statement.  In this case, these details should be extensive
to ensure some problem is not overlooked.

> > One suggestion was to have each iSCSI PDU become framed by a
> > TCP segment. Just the reduction in the average packet size within
> > a high bandwidth stream will stress standard versions of TCP as
> > well as the intervening equipment. SCTP had the foresight to
> > include bundling methods.  This is just one aspect of these
> > proposals to fail consideration within the IPS wg.
>
> This refers to draft-ietf-tsvwg-tcp-ulp-frame-01.txt, a tsvwg that
> proposes changing TCP.  Despite Doug's proclamations that the IPS wg
> should not be discussing changes to TCP, he now criticizes the IPS
> wg for failing to discuss an aspect of a change to TCP???  There's
> just no pleasing some people ... especially as I don't recall this
> specific issue being raised in the past.

I think your using the word smear and this cynical statement are intended to
divert attention from concerns expressed rather than addressing them.  My
critique was with respect to expertise within the IPS in regard to transport
related issues.  A discussion within the IPS should not be expected to fully
vent transport related concerns nor should the IPS be able to declare
closure on this subject.

Doug

>
> Thanks,
> --David
>
> ---------------------------------------------------
> David L. Black, Senior Technologist
> EMC Corporation, 42 South St., Hopkinton, MA  01748
> +1 (508) 249-6449 *NEW*      FAX: +1 (508) 497-8500
> black_david@emc.com         Cell: +1 (978) 394-7754
> ---------------------------------------------------
>
> > -----Original Message-----
> > From: Douglas Otis [mailto:dotis@sanlight.net]
> > Sent: Friday, January 25, 2002 6:13 PM
> > To: ips@ece.cmu.edu
> > Cc: Tsvwg
> > Subject: RE: iSCSI: Framing Steps
> >
> >
> > David,
> >
> > Thank you for the response.  I think a further point should have been
> > strongly made.  Such changes to TCP are not actions of the IPS
> > workgroup. These matters should be discussed within the tsvwg.  Even
> > matters of Fixed Interval Marking brings concerns with respect to the
> > impact even this seemingly innocuous iSCSI option may have.
> >
> > Should the interval within the FIM be a power of 2, guessing an initial
> > sequence value allows injection of packets within existing connections.
> > A layer below TCP, as devised for iSCSI, is to place de-encapsulated
> > iSCSI information from isolated packets that include marker information
> > despite prior packet loss.  A general goal of all the framing
> > approaches being sought by the IPS wg.
> >
> > This would require extensive modification of TCP to facilitate this
> > behavior.  In addition, without the specific details of this scheme,
> > largely due to a desire to claim no changes, it is difficult to evaluate
> > related security risks.  One risk of this scheme is the requirement for
> > application specific code to be routinely placed below the transport.
> > There are no conventions for instituting the requisite signaling between
> > this below the transport layer application process and the modified TCP.
> > Should an attack successfully inject even null structures, it would not
> > be clear how the transport would behave.
> >
> > FIM has been suggested by those within the IPS wg that implementers
> > consider implementing an otherwise obnoxious scheme requiring the
> > injection of markers or suffer from discarded data otherwise retained
> > using standard versions of TCP.  Clearly just sending these markers are
> > not in themselves a risk to TCP.  It is their intended use that causes
> > concern.  This FIM has also been indicated by several including the
> > author of iSCSI that FIM is just an interim measure to be followed by a
> > full framing scheme later.  It is clear this group wishes extensive
> > modifications to TCP and does not wish to share these details openly.
> >
> > One suggestion was to have each iSCSI PDU become framed by a
> > TCP segment. Just the reduction in the average packet size within
> > a high bandwidth stream will stress standard versions of TCP as
> > well as the intervening equipment. SCTP had the foresight to
> > include bundling methods.  This is just one aspect of these
> > proposals to fail consideration within the IPS wg.  In general,
> > I think the IETF made a wise choice in proposing SCTP as the means
> > of incorporating framing within IP protocols and the IPS wg was
> > informed of SCTP at the outset of their endeavors.  SCTP provides
> > many advantages with respect to pressing concerns especially the
> > less disruptive nature incorporation of framing using SCTP would
> > cause over an attempt of framing within TCP.  It is clear there is
> > consensus within the IPS wg that SCTP is not to be considered and
> > that TCP framing is despite prohibitions within the charter and
> > the good judgment of the IETF.  The ongoing tacit approval of
> > these discussions regarding modification of TCP within the IPS wg
> > is troubling.
> >
> > Doug
> >
> > > -----Original Message-----
> > > From: owner-ips@ece.cmu.edu [mailto:owner-ips@ece.cmu.edu]
> > On Behalf Of
> > > Black_David@emc.com [mailto:David_Black@emc.com]
> > > Sent: Wednesday, January 23, 2002 11:10 PM
> > > To: ips@ece.cmu.edu
> > > Subject: iSCSI: Framing Steps
> > >
> > >
> > > I want to attempt to make some steps towards resolving
> > > the framing issues.  In reviewing the recent discussions
> > > on framing, I have a couple of conclusions:
> > >
> > > (1) I do not believe that WG consensus (rough or otherwise)
> > > 	can be obtained for a "MUST implement" requirement
> > > 	for any form of framing.
> > > (2) The COWS mechanism has a lot of potential, but
> > > 	its newness, the multiple versions that
> > > 	have been mentioned, and the desire for some
> > > 	sort of alignment with new work on DDP/RDMA
> > > 	suggest that COWS is premature to specify as
> > > 	part of iSCSI.
> > >
> > > I suggest that these conclusions form the
> > > basis for further ips WG consideration of framing.
> > >
> > > Please think carefully before objecting to these
> > > conclusions on the list (I'm happy to respond to
> > > private questions/expressions of concern).  If the
> > > framing issue cannot be driven to closure in
> > > the next few weeks, I will be forced to conclude
> > > that the entire topic is experimental, and hence
> > > needs to be removed from the iSCSI specification
> > > and handled in separate drafts intended to become
> > > experimental RFCs.
> > >
> > > Thanks,
> > > --David
> > >
> > > p.s. A desire to build NICs that never behave in
> > > accordance with an important SHOULD in RFC 1122
> > > (out-of-order segments SHOULD be queued) does not
> > > strike me as a good reason for changing the first
> > > conclusion above.
> > >
> > > ---------------------------------------------------
> > > David L. Black, Senior Technologist
> > > EMC Corporation, 42 South St., Hopkinton, MA  01748
> > > +1 (508) 249-6449 *NEW*      FAX: +1 (508) 497-8500
> > > black_david@emc.com         Cell: +1 (978) 394-7754
> > > ---------------------------------------------------
> > >
> >
>


_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Sun Jan 27 23:23:43 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id XAA09105
	for <tsvwg-archive@odin.ietf.org>; Sun, 27 Jan 2002 23:23:43 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id XAA09239
	for tsvwg-archive@odin.ietf.org; Sun, 27 Jan 2002 23:23:46 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id XAA08863;
	Sun, 27 Jan 2002 23:06:12 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id XAA08833
	for <tsvwg@optimus.ietf.org>; Sun, 27 Jan 2002 23:06:09 -0500 (EST)
Received: from imo-m06.mx.aol.com (imo-m06.mx.aol.com [64.12.136.161])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id XAA08943
	for <tsvwg@ietf.org>; Sun, 27 Jan 2002 23:06:04 -0500 (EST)
From: DGPickett@aol.com
Received: from DGPickett@aol.com
	by imo-m06.mx.aol.com (mail_out_v31_r1.26.) id l.168.7d4404c (3846);
	Sun, 27 Jan 2002 23:05:30 -0500 (EST)
Message-ID: <168.7d4404c.29862809@aol.com>
Date: Sun, 27 Jan 2002 23:05:29 EST
To: Randall Stewart <randall@stewart.chicago.il.us>
CC: tsvwg@ietf.org
MIME-Version: 1.0
Content-Type: multipart/alternative; boundary="part1_168.7d4404c.29862809_boundary"
X-Mailer: AOL 6.0 for Windows US sub 10556
Subject: [Tsvwg] Re: zero length send
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org


--part1_168.7d4404c.29862809_boundary
Content-Type: text/plain; charset="US-ASCII"
Content-Transfer-Encoding: 7bit

Even if TCP did send an empty packet for a zero length write, wouldn't it be 
essentially identical to a 'keep alive' packet, and so of little signaling 
value?  If it did affect the receiving end, a zero length read might be 
mistaken for EOF.  I once coded this on a UNIX pipe, to test if the reading 
process was alive, but all I did was trick the recipient into closing the 
pipe.

It seems proper that UDP sends an empty datagram, as that is just a very 
trivial case of a datagram.  It's most obvious use would be as an application 
level 'keep alive' to support a virtual connection.

If a protocol sends and receives messages, an empty message is valid, but if 
it is a stream of primitives (words, bytes, bits), then an empty write should 
put nothing on the stream.  Historically, such a write has either been 
ignored, illegal, or impossible (assigning the length 0 to mean 2^n, thereby 
gaining an enticingly round but tiny bit of range at the cost of a bunch of 
counter-intuitiveness).  I like 'ignored,' as it generates less code.  It's 
nice to decide, and tell users explicitly what to expect.  It hardly seems 
worth making the behavior explicitly implementation dependent or implicitly 
laissez-faire.

--part1_168.7d4404c.29862809_boundary
Content-Type: text/html; charset="US-ASCII"
Content-Transfer-Encoding: 7bit

<HTML><FONT FACE=arial,helvetica><FONT  SIZE=2>Even if TCP did send an empty packet for a zero length write, wouldn't it be essentially identical to a 'keep alive' packet, and so of little signaling value? &nbsp;If it did affect the receiving end, a zero length read might be mistaken for EOF. &nbsp;I once coded this on a UNIX pipe, to test if the reading process was alive, but all I did was trick the recipient into closing the pipe.
<BR>
<BR>It seems proper that UDP sends an empty datagram, as that is just a very trivial case of a datagram. &nbsp;It's most obvious use would be as an application level 'keep alive' to support a virtual connection.
<BR>
<BR>If a protocol sends and receives messages, an empty message is valid, but if it is a stream of primitives (words, bytes, bits), then an empty write should put nothing on the stream. &nbsp;Historically, such a write has either been ignored, illegal, or impossible (assigning the length 0 to mean 2^n, thereby gaining an enticingly round but tiny bit of range at the cost of a bunch of counter-intuitiveness). &nbsp;I like 'ignored,' as it generates less code. &nbsp;It's nice to decide, and tell users explicitly what to expect. &nbsp;It hardly seems worth making the behavior explicitly implementation dependent or implicitly laissez-faire.</FONT></HTML>

--part1_168.7d4404c.29862809_boundary--

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Mon Jan 28 03:25:39 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id DAA20367
	for <tsvwg-archive@odin.ietf.org>; Mon, 28 Jan 2002 03:25:39 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id DAA26971
	for tsvwg-archive@odin.ietf.org; Mon, 28 Jan 2002 03:25:41 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id DAA26527;
	Mon, 28 Jan 2002 03:07:24 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id DAA26485
	for <tsvwg@optimus.ietf.org>; Mon, 28 Jan 2002 03:07:19 -0500 (EST)
Received: from mxic2.us.dg.com (mxic2.isus.emc.com [128.221.31.40])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id DAA20133
	for <tsvwg@ietf.org>; Mon, 28 Jan 2002 03:07:17 -0500 (EST)
From: Black_David@emc.com
Received: by mxic2.us.dg.com with Internet Mail Service (5.5.2653.19)
	id <ZRAVSPW6>; Mon, 28 Jan 2002 03:01:46 -0500
Message-ID: <277DD60FB639D511AC0400B0D068B71E029373D7@CORPMX14>
To: dotis@sanlight.net, ips@ece.cmu.edu
Cc: tsvwg@ietf.org
Date: Mon, 28 Jan 2002 02:53:20 -0500
MIME-Version: 1.0
X-Mailer: Internet Mail Service (5.5.2653.19)
Content-Type: text/plain;
	charset="iso-8859-1"
Subject: [Tsvwg] RE: iSCSI: Framing Steps
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

> > TCP is vulnerable to packet injection based on guessing or observing
> > the initial sequence number in the absence of fixed-interval-markers
> > (FIM).  The well-known connection hijack attack is an example.  An
> > attacker can inject arbitrary data into a connection without FIM and
> > cause TCP to misbehave as a result.  FIM does not make this any
> > easier or change the ways in which TCP would misbehave when attacked
> > in this fashion.
> 
> FIM enables *processing* of packets out of sequence from the initial send.
> The intent of FIM is to make such processing possible and that *does*
enable
> a greater opportunity for attack especially if the measure is predictable
> within future headers.

Nonsense.  TCP processes packets in the same order (i.e., the order
received)
with or without FIM.  TCP would not need or use FIM to find TCP headers,
hence FIM provides no additional opportunity to inject a fake TCP header.
Every TCP receiver already has to perform processing on packets "out of
sequence from the initial send" due to drops and reordering that may
happen in the network; this processing may include queuing to wait for
missing segments to arrive.  FIM changes only the memory locations of
the packets that are processed, not the order of their processing.

> > FIM does not entail any changes to TCP.  TCP does not care about
> > where incoming data is placed in memory - it cares about the order
> > in which data is processed by TCP and delivered to the application.
> 
> A process that examines application specific structures at the packet (IP)
> level, to then move portions of the expected TCP byte stream into user
space
> based on application references matching these structures, is *not* a
> modification to TCP provided there is a *sequential* order of delivery?

That is correct.  RFC 793 specifically allows sharing of storage between
TCP and the user provided that delivery order is maintained at the
interface,
and even gives an example of a data structure that could be used (see
below).

> Clearly placing a portion of the application below TCP is a modification
to
> TCP and network layering.  Are you referring to a means at the TCP API to
> hide this activity? Are you saying that the TCP API does not change?
> How is this done without changing TCP?

The only change TCP sees is that the location of packets in memory is
different.  As far as TCP is concerned, this is equivalent to replacing
the operating system's memory allocator, including possible modifications
to TCP code to use the new allocator (e.g., the BSD and streams
implementations
of TCP use rather different memory allocators).  As for the "TCP API" ...

> The application specific processing below this modified TCP transport is
to
> de-encapsulate the iSCSI payloads into user space. iSCSI user space may
not
> be satisfied in the sequence they were established.

Both the de-encapsulation into user space and the satisfaction of receive
operations out of order are allowed by RFC 793.  Here's the text from
pp. 48-49 (definition of Receive in the nominal TCP/user interface):

	  To distinguish among several outstanding RECEIVEs and to take
        care of the case that a buffer is not completely filled, the
        return code is accompanied by both a buffer pointer and a byte
        count indicating the actual length of the data received.

        Alternative implementations of RECEIVE might have the TCP
        allocate buffer storage, or the TCP might share a ring buffer
        with the user.

The first paragraph appears to allow completion of RECEIVEs out of order
and the second definitely allows sharing of TCP's data storage with
"the user".  In practice, most TCP implementations complete RECEIVEs in
order for simplicity and to avoid surprises at the socket API ...

Picking up the "TCP API" issue from above, no modifications are needed
to the nominal API defined in RFC 793.  The intended use of FIM is
not possible through the socket API (not an IETF standard, FWIW), but
there is no proposal to expose FIM through that API - it's intended
for combined implementations of TCP, FIM, and iSCSI, and hence would
only be specified as part of iSCSI.

> The complexity involved
> will require modification of TCP to support this buffer handling and
> application specific structure processing.  In the case of iSCSI, the
> content of the byte stream is not known before hand.  This activity below
> TCP places and splices sections of the TCP byte stream into designated
iSCSI
> user space.  Clearly there are many questions one may have with respect to
> what and when TCP validates and delivers.

Modifications to TCP implementations are definitely involved.  The
answers to the questions can generally be found in RFC 793 and 1122
because FIM makes no changes to "what and when TCP validates and delivers".
For example ...

> TCP could easily reject a packet
> based on the byte sequence within headers that was not rejected by this
> "below the transport process" that would be unable to see the header
> continuum.

Since FIM performs no protocol processing (it only allocates memory for the
incoming packet), TCP rejects the packet in the usual fashion, causing
it never to be delivered to iSCSI.  Even with FIM, data is not delivered
to iSCSI until TCP delivers it in sequence, hence the rejected packet
has no effect on iSCSI.

> This added state below the TCP transport provides opportunity
> for attack or in some implementations, poor behavior.

Any work on implementations has the potential to introduce bugs; in
this respect FIM is no different from any other implementation work.

> > The intended usage of FIM affects only the placement of incoming
> > data in memory and hence does not change TCP.
> 
> Placement of application specific payloads into regions matched against
> application specific structure content prior to being handled by the TCP
> layer is a change to TCP.  TCP is not able to manipulate buffers in this
> manner without extensive modification.  Such modification is quite
possible
> however, and all the details should be explored before the use of this
> scheme is placed into a standards draft however.

I would agree insofar as changes to TCP implementation code are
required to use FIM in the fashion intended.  I maintain that FIM does
not change the TCP protocol as specified in RFC 793 and 1122.  The memory
and buffer allocation aspects of TCP have not been standardized by the IETF
for very good reasons.

> My concern remains that the IPS wg is NOT the place to discuss these
> concerns as that is not their expertise nor their charter.

I have no problem with bringing this sort of issue to tsvwg, as long
as it's done in an honest fashion.  Specifically, given Doug's past
insistence that the IPS wg not discuss TCP changes, he should not
complain that the IPS wg has failed to adequately discuss TCP changes
in the TSVWG TCP ULP framing draft.

Finally, on requirements:

> You have also
> asked the group to drive the framing issue to conclusion but you failed to
> indicate that such discussion should not take place within the IPS.

iSCSI requirements for use of framing mechanisms are proper for discussion
in IPS.  That requirements discussion will be driven to a conclusion in
IPS because that conclusion will be reflected in the iSCSI specification.

> Ensure that an error is not made where everyone expects such modifications
> to TCP even if done unofficially by the IPS.

Just in case anyone else is confused - TCP ULP framing is a tsvwg draft,
and FIM is currently in the iSCSI draft because it does not modify the
TCP protocol.

Thanks,
--David

---------------------------------------------------
David L. Black, Senior Technologist
EMC Corporation, 42 South St., Hopkinton, MA  01748
+1 (508) 249-6449 *NEW*      FAX: +1 (508) 497-8500
black_david@emc.com         Cell: +1 (978) 394-7754
---------------------------------------------------

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Mon Jan 28 05:44:51 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id FAA21883
	for <tsvwg-archive@odin.ietf.org>; Mon, 28 Jan 2002 05:44:51 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id FAA03067
	for tsvwg-archive@odin.ietf.org; Mon, 28 Jan 2002 05:44:54 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id FAA01996;
	Mon, 28 Jan 2002 05:15:51 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id FAA01931
	for <tsvwg@optimus.ietf.org>; Mon, 28 Jan 2002 05:15:48 -0500 (EST)
Received: from c003.snv.cp.net (c003-h004.c003.snv.cp.net [209.228.32.218])
	by ietf.org (8.9.1a/8.9.1a) with SMTP id FAA21647
	for <tsvwg@ietf.org>; Mon, 28 Jan 2002 05:15:45 -0500 (EST)
Received: (cpmta 10842 invoked from network); 28 Jan 2002 02:15:15 -0800
Received: from 64.130.130.105 (HELO dougrmt)
  by smtp.telocity.com (209.228.32.218) with SMTP; 28 Jan 2002 02:15:15 -0800
X-Sent: 28 Jan 2002 10:15:15 GMT
From: "Douglas Otis" <dotis@sanlight.net>
To: <Black_David@emc.com>, <ips@ece.cmu.edu>
Cc: <tsvwg@ietf.org>
Date: Mon, 28 Jan 2002 02:16:34 -0800
Message-ID: <NEBBJGDMMLHHCIKHGBEJOEDJCPAA.dotis@sanlight.net>
MIME-Version: 1.0
Content-Type: text/plain;
	charset="iso-8859-1"
Content-Transfer-Encoding: 7bit
X-Priority: 3 (Normal)
X-MSMail-Priority: Normal
X-Mailer: Microsoft Outlook IMO, Build 9.0.2416 (9.0.2910.0)
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4807.1700
Importance: Normal
In-Reply-To: <277DD60FB639D511AC0400B0D068B71E029373D7@CORPMX14>
Content-Transfer-Encoding: 7bit
Subject: [Tsvwg] RE: iSCSI: Framing Steps
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit

David,

> > > TCP is vulnerable to packet injection based on guessing or observing
> > > the initial sequence number in the absence of fixed-interval-markers
> > > (FIM).  The well-known connection hijack attack is an example.  An
> > > attacker can inject arbitrary data into a connection without FIM and
> > > cause TCP to misbehave as a result.  FIM does not make this any
> > > easier or change the ways in which TCP would misbehave when attacked
> > > in this fashion.
> >
> > FIM enables *processing* of packets out of sequence from the
> > initial send. The intent of FIM is to make such processing
> > possible and that *does* enable a greater opportunity for
> > attack especially if the measure is predictable within future
> > headers.
>
> Nonsense.  TCP processes packets in the same order (i.e., the order
> received) with or without FIM.  TCP would not need or use FIM to
> find TCP headers, hence FIM provides no additional opportunity to
> inject a fake TCP header. Every TCP receiver already has to perform
> processing on packets "out of sequence from the initial send" due to
> drops and reordering that may happen in the network; this processing
> may include queuing to wait for missing segments to arrive.  FIM
> changes only the memory locations of the packets that are processed,
> not the order of their processing.

The concerns are not with TCP but with this process running below TCP.  This
would be a stateful process that could indeed be confused by an
inappropriate reliance on a marker.  This change in location of content
arrival would be a complex process that includes application specific
structure handling.  This "only changes memory locations of" is an
application specific process.  In addition, it is not the packet location
that is being changed, but the content of the packet or it would be of
little value.  There are already zero copy schemes that do not rely on
markers to achieve appropriate packet placement.  The iSCSI appendix clearly
indicates the value of these markers is to locate iSCSI specific structures
within the TCP byte stream should there be dropped packets.  FIM is to allow
processing of these structures in the face of previously dropped packets.

> > > FIM does not entail any changes to TCP.  TCP does not care about
> > > where incoming data is placed in memory - it cares about the order
> > > in which data is processed by TCP and delivered to the application.
> >
> > A process that examines application specific structures at the
> > packet (IP) level, to then move portions of the expected TCP byte
> > stream into user space based on application references matching
> > these structures, is *not* a modification to TCP provided there
> > is a *sequential* order of delivery?
>
> That is correct.  RFC 793 specifically allows sharing of storage between
> TCP and the user provided that delivery order is maintained at the
> interface, and even gives an example of a data structure that could be
> used (see below).
>
> > Clearly placing a portion of the application below TCP is a modification
> > to TCP and network layering.  Are you referring to a means at the
> > TCP API to hide this activity? Are you saying that the TCP API does not
> > change? How is this done without changing TCP?
>
> The only change TCP sees is that the location of packets in memory is
> different.  As far as TCP is concerned, this is equivalent to replacing
> the operating system's memory allocator, including possible modifications
> to TCP code to use the new allocator (e.g., the BSD and streams
> implementations of TCP use rather different memory allocators).  As for
> the "TCP API" ...
>
> > The application specific processing below this modified TCP transport is
> > to de-encapsulate the iSCSI payloads into user space. iSCSI user space
> > may not be satisfied in the sequence they were established.
>
> Both the de-encapsulation into user space and the satisfaction of receive
> operations out of order are allowed by RFC 793.  Here's the text from
> pp. 48-49 (definition of Receive in the nominal TCP/user interface):
>
>  To distinguish among several outstanding RECEIVEs and to take
>  care of the case that a buffer is not completely filled, the
>  return code is accompanied by both a buffer pointer and a byte
>  count indicating the actual length of the data received.
>
>  Alternative implementations of RECEIVE might have the TCP
>  allocate buffer storage, or the TCP might share a ring buffer
>  with the user.
>
> The first paragraph appears to allow completion of RECEIVEs out of order
> and the second definitely allows sharing of TCP's data storage with
> "the user".  In practice, most TCP implementations complete RECEIVEs in
> order for simplicity and to avoid surprises at the socket API ...
>
> Picking up the "TCP API" issue from above, no modifications are needed
> to the nominal API defined in RFC 793.  The intended use of FIM is
> not possible through the socket API (not an IETF standard, FWIW), but
> there is no proposal to expose FIM through that API - it's intended
> for combined implementations of TCP, FIM, and iSCSI, and hence would
> only be specified as part of iSCSI.

TCP buffer treatment does not begin to compare to the complex process that
would be employed to handle the dispatching of the TCP byte stream
anticipated by iSCSI.  The TCP byte stream is commonly stored out of
sequence from the send but this process does not depend on the data.  This
iSCSI structure marker process goes well beyond a simple buffer assignment,
is highly stateful, and prone to errors where recovery would be difficult.
Indeed, TCP allows flexibility within the API with respect to buffer
handling, but I think it would be a mistake to assume this flexibility
allows modes of operation dependent upon the recognition and handling of
application specific structures.  iSCSI wishes to run a portion of the
application in front of TCP and to do this on a byte by byte basis.  TCP can
not "deliver" data to the application out of sequence so logically there
should be no need for such markers if TCP were allowed to run first.  iSCSI
structure recognition, synchronization, validation would be run prior to the
normal TCP delivery and is the reasoning for the use of these PDU markers.

> > The complexity involved will require modification of TCP to support
> > this buffer handling and application specific structure processing.
> > In the case of iSCSI, the content of the byte stream is not known
> > before hand.  This activity below TCP places and splices sections
> > of the TCP byte stream into designated iSCSI user space.  Clearly
> > there are many questions one may have with respect to what and when
> > TCP validates and delivers.
>
> Modifications to TCP implementations are definitely involved.  The
> answers to the questions can generally be found in RFC 793 and 1122
> because FIM makes no changes to "what and when TCP validates and
> delivers".
> For example ...
>
> > TCP could easily reject a packet based on the byte sequence within
> > headers that was not rejected by this "below the transport process"
> > that would be unable to see the header continuum.
>
> Since FIM performs no protocol processing (it only allocates
> memory for the incoming packet), TCP rejects the packet in the usual
> fashion, causing it never to be delivered to iSCSI.  Even with FIM,
> data is not delivered to iSCSI until TCP delivers it in sequence,
> hence the rejected packet has no effect on iSCSI.

The goal of the marker has nothing to do with what may be described as zero
copy TCP.  FIM is to locate iSCSI PDUs as defined in the iSCSI draft.  The
driving motivation behind this effort is to allow a direct placement of the
data payload portion of the iSCSI structures into the associated user space
referenced by the tags within these structures.  In other words, a type of
direct data placement.  The "deliveries" to TCP from this process would be
an interesting topic.  Deliveries in turn from TCP back into the application
would need to know if this "below the transport" process had finished
successfully or had even been run.  This is not TCP.

> > This added state below the TCP transport provides opportunity
> > for attack or in some implementations, poor behavior.
>
> Any work on implementations has the potential to introduce bugs; in
> this respect FIM is no different from any other implementation work.

This will impact many applications that will also be using the same stack.
Ensuring sound concepts would be to the benefit of IPS as well as other
working groups.  Should the implementation of FIM create substantial
security risks by nature of its design, then such design should be stopped.

> > > The intended usage of FIM affects only the placement of incoming
> > > data in memory and hence does not change TCP.
> >
> > Placement of application specific payloads into regions matched against
> > application specific structure content prior to being handled by the TCP
> > layer is a change to TCP.  TCP is not able to manipulate buffers in this
> > manner without extensive modification.  Such modification is quite
> > possible however, and all the details should be explored before the use
> > of this scheme is placed into a standards draft however.
>
> I would agree insofar as changes to TCP implementation code are
> required to use FIM in the fashion intended.  I maintain that FIM does
> not change the TCP protocol as specified in RFC 793 and 1122.  The memory
> and buffer allocation aspects of TCP have not been standardized
> by the IETF for very good reasons.

Please refer to my prior statement.

> > My concern remains that the IPS wg is NOT the place to discuss these
> > concerns as that is not their expertise nor their charter.
>
> I have no problem with bringing this sort of issue to tsvwg, as long
> as it's done in an honest fashion.  Specifically, given Doug's past
> insistence that the IPS wg not discuss TCP changes, he should not
> complain that the IPS wg has failed to adequately discuss TCP changes
> in the TSVWG TCP ULP framing draft.
>
> Finally, on requirements:
>
> > You have also asked the group to drive the framing issue to conclusion
> > but you failed to indicate that such discussion should not take place
> > within the IPS.
>
> iSCSI requirements for use of framing mechanisms are proper for discussion
> in IPS.  That requirements discussion will be driven to a conclusion in
> IPS because that conclusion will be reflected in the iSCSI specification.
>
> > Ensure that an error is not made where everyone expects such
> > modifications to TCP even if done unofficially by the IPS.
>
> Just in case anyone else is confused - TCP ULP framing is a tsvwg draft,
> and FIM is currently in the iSCSI draft because it does not modify the
> TCP protocol.

I think it would be a reasonable statement that I do not agree with your
assessment on the impact to TCP created by the optional markers used in
iSCSI.  The IPS group represents talented people highly skilled in the field
of storage but could benefit from the advice of the community at large with
respect to this topic.  As I do feel an implementation as intended of the
iSCSI marker scheme would be a modification to TCP, I also felt strongly
that this topic get raised within the tsvwg work group.  I would like to
thank you for your response.

Doug

> Thanks,
> --David
>
> ---------------------------------------------------
> David L. Black, Senior Technologist
> EMC Corporation, 42 South St., Hopkinton, MA  01748
> +1 (508) 249-6449 *NEW*      FAX: +1 (508) 497-8500
> black_david@emc.com         Cell: +1 (978) 394-7754
> ---------------------------------------------------
>


_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Mon Jan 28 08:20:00 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id IAA24123
	for <tsvwg-archive@odin.ietf.org>; Mon, 28 Jan 2002 08:19:56 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id IAA09105
	for tsvwg-archive@odin.ietf.org; Mon, 28 Jan 2002 08:19:58 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id IAA08617;
	Mon, 28 Jan 2002 08:03:04 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id IAA08588
	for <tsvwg@optimus.ietf.org>; Mon, 28 Jan 2002 08:03:02 -0500 (EST)
Received: from CNRI.Reston.VA.US (localhost [127.0.0.1])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id IAA23637;
	Mon, 28 Jan 2002 08:02:58 -0500 (EST)
Message-Id: <200201281302.IAA23637@ietf.org>
Mime-Version: 1.0
Content-Type: Multipart/Mixed; Boundary="NextPart"
To: IETF-Announce: ;
Cc: tsvwg@ietf.org
From: Internet-Drafts@ietf.org
Reply-to: Internet-Drafts@ietf.org
Date: Mon, 28 Jan 2002 08:02:58 -0500
Subject: [Tsvwg] I-D ACTION:draft-ietf-tsvwg-udp-lite-00.txt
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

--NextPart

A New Internet-Draft is available from the on-line Internet-Drafts directories.
This draft is a work item of the Transport Area Working Group Working Group of the IETF.

	Title		: The UDP Lite Protocol
	Author(s)	: L. Larzon, M. Degermark
	Filename	: draft-ietf-tsvwg-udp-lite-00.txt
	Pages		: 8
	Date		: 25-Jan-02
	
This document describes the UDP Lite protocol, which is similar to
classic UDP [RFC-768], but can also serve applications which in lossy
network environments prefer to have partially damaged payloads
delivered rather than discarded.  If this feature is not used, UDP
Lite is semantically identical to classic UDP.

A URL for this Internet-Draft is:
http://www.ietf.org/internet-drafts/draft-ietf-tsvwg-udp-lite-00.txt

To remove yourself from the IETF Announcement list, send a message to 
ietf-announce-request with the word unsubscribe in the body of the message.

Internet-Drafts are also available by anonymous FTP. Login with the username
"anonymous" and a password of your e-mail address. After logging in,
type "cd internet-drafts" and then
	"get draft-ietf-tsvwg-udp-lite-00.txt".

A list of Internet-Drafts directories can be found in
http://www.ietf.org/shadow.html 
or ftp://ftp.ietf.org/ietf/1shadow-sites.txt


Internet-Drafts can also be obtained by e-mail.

Send a message to:
	mailserv@ietf.org.
In the body type:
	"FILE /internet-drafts/draft-ietf-tsvwg-udp-lite-00.txt".
	
NOTE:	The mail server at ietf.org can return the document in
	MIME-encoded form by using the "mpack" utility.  To use this
	feature, insert the command "ENCODING mime" before the "FILE"
	command.  To decode the response(s), you will need "munpack" or
	a MIME-compliant mail reader.  Different MIME-compliant mail readers
	exhibit different behavior, especially when dealing with
	"multipart" MIME messages (i.e. documents which have been split
	up into multiple messages), so check your local documentation on
	how to manipulate these messages.
		
		
Below is the data which will enable a MIME compliant mail reader
implementation to automatically retrieve the ASCII version of the
Internet-Draft.

--NextPart
Content-Type: Multipart/Alternative; Boundary="OtherAccess"

--OtherAccess
Content-Type: Message/External-body;
	access-type="mail-server";
	server="mailserv@ietf.org"

Content-Type: text/plain
Content-ID:	<20020125134428.I-D@ietf.org>

ENCODING mime
FILE /internet-drafts/draft-ietf-tsvwg-udp-lite-00.txt

--OtherAccess
Content-Type: Message/External-body;
	name="draft-ietf-tsvwg-udp-lite-00.txt";
	site="ftp.ietf.org";
	access-type="anon-ftp";
	directory="internet-drafts"

Content-Type: text/plain
Content-ID:	<20020125134428.I-D@ietf.org>

--OtherAccess--

--NextPart--



_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Mon Jan 28 09:01:15 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id JAA25963
	for <tsvwg-archive@odin.ietf.org>; Mon, 28 Jan 2002 09:01:15 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id JAA12131
	for tsvwg-archive@odin.ietf.org; Mon, 28 Jan 2002 09:01:18 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id IAA10168;
	Mon, 28 Jan 2002 08:41:14 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id IAA10139
	for <tsvwg@optimus.ietf.org>; Mon, 28 Jan 2002 08:41:12 -0500 (EST)
Received: from localhost ([203.228.21.33])
	by ietf.org (8.9.1a/8.9.1a) with SMTP id IAA24891
	for <tsvwg@ietf.org>; Mon, 28 Jan 2002 08:41:05 -0500 (EST)
Message-Id: <200201281341.IAA24891@ietf.org>
Reply-To: lie_mail@hanmail.net
From: lie_mail<lie_mail@hanmail.net>
To: tsvwg@ietf.org
Mime-Version: 1.0
Content-Type: text/html; charset="ks_c_5601-1987"
Date: Mon, 28 Jan 2002 22:17:27 +0900
Subject: [Tsvwg] [성인광고]다있다!!
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

<HTML>
  <HEAD>
      <TITLE></TITLE>
  </HEAD>
  <BODY>
<IMG alt="" hspace=0 src="d:\mail\7747\7747.gif" align=left border=0>
<pre>

<h1><font color = red>
귀하의 품안에 원하는 애인을 
              쏘옥~~안겨드립니다!! </font>
<h4>
1. 먼저 전화기를 들고(핸드폰 일반전화가능) 

   <font color = blue size = 18>060-700-7747</font>

   를 꾹꾹 누릅니다.

2. 이쁜언냐의 안내멘트가 끝나고 *를 눌러줍니다.

3. 귀하가 원하는 회원번호 5자리를 눌러줍니다.

4. 귀하가 원하는 비밀번호 4자리를 눌러줍니다.

5. 메세지가 도착했을때나 추후에 회원번호 비밀번호

   확인을 위한 귀하의 핸드폰번호를 입력한후 *를 

   눌러줍니다.(필확인요!!)

6. 축하합니다~란 멘트와함께 1번 2번 코너 안내멘트가

   나오면 회원등록성공!!

7. 1번메세지청취코너나 3번 공개게시판코너에서

   맘에드는 이쁜언냐들의 메세지를 듣고 5번

   눌러 멋진음성으로 답장을 보내거나 
  
   귀하의 개인프로필을 남깁니다~~

** 이케하면 오빠야 품으로 열나섹쒸 쭉쭉빵빵 언냐들이 호박넝쿨처럼 우루룩!! **

<h5>
본 메일은 정보통신망 이용촉진법 규정에 따라 광고메일임을 표시하였으며 수신거부장치를 마련하고 있습니다. 
원치 않은 정보였다면 정중히 사과 드리며, 수신거부를 해 주시면 다음부터는 메일이 발송되지 않을 것입니다
<a href="mailto:mail_lie@hanmail.net?subject=수신거부">수신거부</a>
                                                                     유료정보 (30/\100)

  </BODY>
</HTML>

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Mon Jan 28 09:40:14 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id JAA27391
	for <tsvwg-archive@odin.ietf.org>; Mon, 28 Jan 2002 09:40:09 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id JAA14684
	for tsvwg-archive@odin.ietf.org; Mon, 28 Jan 2002 09:40:13 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id JAA13336;
	Mon, 28 Jan 2002 09:17:14 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id JAA13296
	for <tsvwg@optimus.ietf.org>; Mon, 28 Jan 2002 09:17:09 -0500 (EST)
Received: from sj-msg-core-1.cisco.com (sj-msg-core-1.cisco.com [171.71.163.11])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id JAA26650
	for <tsvwg@ietf.org>; Mon, 28 Jan 2002 09:17:05 -0500 (EST)
Received: from mira-sjc5-2.cisco.com (mira-sjc5-2.cisco.com [171.71.163.16])
	by sj-msg-core-1.cisco.com (8.11.3/8.9.1) with ESMTP id g0SEGbL23087;
	Mon, 28 Jan 2002 06:16:37 -0800 (PST)
Received: from stewart.chicago.il.us (ssh-sj1.cisco.com [171.68.225.134])
	by mira-sjc5-2.cisco.com (Mirapoint)
	with ESMTP id AAX21376;
	Mon, 28 Jan 2002 06:16:36 -0800 (PST)
Message-ID: <3C555D43.3C57E78E@stewart.chicago.il.us>
Date: Mon, 28 Jan 2002 08:16:35 -0600
From: Randall Stewart <randall@stewart.chicago.il.us>
X-Mailer: Mozilla 4.76 [en] (X11; U; Linux 2.2.12 i386)
X-Accept-Language: en
MIME-Version: 1.0
To: DGPickett@aol.com
CC: tsvwg@ietf.org
Subject: Re: [Tsvwg] Re: zero length send
References: <168.7d4404c.29862809@aol.com>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Transfer-Encoding: 7bit
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit

Comments below..

DGPickett@aol.com wrote:
> 
> Even if TCP did send an empty packet for a zero length write, wouldn't
> it be essentially identical to a 'keep alive' packet, and so of little
> signaling value?  If it did affect the receiving end, a zero length
> read might be mistaken for EOF.  I once coded this on a UNIX pipe, to
> test if the reading process was alive, but all I did was trick the
> recipient into closing the pipe.

Yes, this is (if I remember right) why the IESG had us
make it illegal to send a 0 length chunk... i.e. it might
be mistaken for a EOF...

> 
> It seems proper that UDP sends an empty datagram, as that is just a
> very trivial case of a datagram.  It's most obvious use would be as an
> application level 'keep alive' to support a virtual connection.
> 
> If a protocol sends and receives messages, an empty message is valid,
> but if it is a stream of primitives (words, bytes, bits), then an
> empty write should put nothing on the stream.  Historically, such a
> write has either been ignored, illegal, or impossible (assigning the
> length 0 to mean 2^n, thereby gaining an enticingly round but tiny bit
> of range at the cost of a bunch of counter-intuitiveness).  I like
> 'ignored,' as it generates less code.  It's nice to decide, and tell
> users explicitly what to expect.  It hardly seems worth making the
> behavior explicitly implementation dependent or implicitly
> laissez-faire.

True, we probably should add something that says either
that an error should be returned or it should be accepted
and discarded .. I don't really care which.. 

R

-- 
Randall R. Stewart
randall@stewart.chicago.il.us 815-342-5222 (cell phone)

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Mon Jan 28 22:03:51 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id WAA15009
	for <tsvwg-archive@odin.ietf.org>; Mon, 28 Jan 2002 22:03:51 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id WAA19398
	for tsvwg-archive@odin.ietf.org; Mon, 28 Jan 2002 22:03:54 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id VAA18715;
	Mon, 28 Jan 2002 21:46:26 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id VAA18684
	for <tsvwg@optimus.ietf.org>; Mon, 28 Jan 2002 21:46:24 -0500 (EST)
Received: from imo-d03.mx.aol.com (imo-d03.mx.aol.com [205.188.157.35])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id VAA14857
	for <tsvwg@ietf.org>; Mon, 28 Jan 2002 21:46:21 -0500 (EST)
From: DGPickett@aol.com
Received: from DGPickett@aol.com
	by imo-d03.mx.aol.com (mail_out_v31_r1.26.) id l.9f.21cc88bd (3314);
	Mon, 28 Jan 2002 21:45:44 -0500 (EST)
Message-ID: <9f.21cc88bd.298766d7@aol.com>
Date: Mon, 28 Jan 2002 21:45:43 EST
Subject: Re: [Tsvwg] Re: zero length send
To: randall@stewart.chicago.il.us
CC: tsvwg@ietf.org
MIME-Version: 1.0
Content-Type: multipart/alternative; boundary="part1_9f.21cc88bd.298766d7_boundary"
X-Mailer: AOL 6.0 for Windows US sub 10556
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org


--part1_9f.21cc88bd.298766d7_boundary
Content-Type: text/plain; charset="US-ASCII"
Content-Transfer-Encoding: 7bit

Well, like I said, I like 'ignore.'  By specifying that it MUST be ignored in 
a primitive stream situation, we get all the value there is in mentioning it: 
the opportunity to prevent code (KISS) on both sides of the protocol!

--part1_9f.21cc88bd.298766d7_boundary
Content-Type: text/html; charset="US-ASCII"
Content-Transfer-Encoding: 7bit

<HTML><FONT FACE=arial,helvetica><FONT  SIZE=2>Well, like I said, I like 'ignore.' &nbsp;By specifying that it MUST be ignored in a primitive stream situation, we get all the value there is in mentioning it: the opportunity to prevent code (KISS) on both sides of the protocol!</FONT></HTML>

--part1_9f.21cc88bd.298766d7_boundary--

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Mon Jan 28 22:04:10 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id WAA15026
	for <tsvwg-archive@odin.ietf.org>; Mon, 28 Jan 2002 22:04:06 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id WAA19412
	for tsvwg-archive@odin.ietf.org; Mon, 28 Jan 2002 22:04:09 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id VAA18620;
	Mon, 28 Jan 2002 21:44:43 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id VAA18589
	for <tsvwg@optimus.ietf.org>; Mon, 28 Jan 2002 21:44:41 -0500 (EST)
Received: from magic.adaptec.com (magic.adaptec.com [208.236.45.80])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id VAA14841
	for <tsvwg@ietf.org>; Mon, 28 Jan 2002 21:44:38 -0500 (EST)
Received: from redfish.adaptec.com (redfish.adaptec.com [162.62.50.11])
	by magic.adaptec.com (8.8.8+Sun/8.8.8) with ESMTP id SAA28647;
	Mon, 28 Jan 2002 18:44:06 -0800 (PST)
Received: from aimexc07.corp.adaptec.com (aimexc07.corp.adaptec.com [162.62.62.47])
	by redfish.adaptec.com (8.8.8+Sun/8.8.8) with ESMTP id SAA12375;
	Mon, 28 Jan 2002 18:27:25 -0800 (PST)
Received: by aimexc07.corp.adaptec.com with Internet Mail Service (5.5.2653.19)
	id <DNC4ZY25>; Mon, 28 Jan 2002 18:42:59 -0800
Message-ID: <E156A23F1885D4119ED800B0D0498A9F0165FF38@aimexc07.corp.adaptec.com>
From: "Mukund, Shridhar" <Shridhar_Mukund@adaptec.com>
To: "'Douglas Otis'" <dotis@sanlight.net>, Black_David@emc.com,
        ips@ece.cmu.edu
Cc: tsvwg@ietf.org
Date: Mon, 28 Jan 2002 18:42:54 -0800
MIME-Version: 1.0
X-Mailer: Internet Mail Service (5.5.2653.19)
Content-Type: text/plain;
	charset="iso-8859-1"
Subject: [Tsvwg] RE: iSCSI: Framing Steps
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org


Dear colleagues,

1. Doug : David made the point, "it(FIM) does not modify 
   the TCP protocol", as lucid as it gets. It will be nice 
   to hear some opinions from the TCP gurus out there
   without mixing it up with SCTP and TUF.

2. It will also be nice if other iSCSI folks on the reflector 
   cast their votes per John's request.
   Putting on my politician hat :-), pl. vote for option 6.
   >>>> 6. FIM now, and some kind of Framing later

3. I do agree with David that the implication, "Senders that 
   do not insert markers on send should be willing to accept 
   up to RTT*BW drops" in my email was too strong as it goes 
   against a RFC1122-SHOULD. ( See reference below ). I
   stand corrected.

   The current iSCSI draft addresses my point anyway:
   "In certain environments, a sender not willing to supply 
   markers to a receiver willing to accept markers MAY 
   suffer from a considerable performance degradation."

   In my interpretation, the spirit behind the RFC1122-SHOULD 
   is to save IP networks from having to transport 
   same data all over again in the event of packet drop. 
   In fact, FIM directly addresses this very spirit.

   We may want to add a note in the iSCSI draft that 
   implementations should try to honor the RFC1122-SHOULD
   when the sender is not willing to supply markers.

Thanks.

-Shridhar Mukund
   

------------------------------------------------------------------------
Reference : David's response dtd 14 Jan. to my email:

>    b. I strongly recommend SHOULD implement FIM on the send side. 
>       Implication -> Senders that do not insert markers should be 
>       willing to accept up to RTT*BW data drops! Headers being 
>       "reasonably" out-of-order is OK. Of course, senders that do not 
>       insert markers but are willing to pay big $$$ to the SSP will 
>       get their buffer/BW allocation as usual and customary :-)

I think the Implication is too strong, as it goes against the following
SHOULD from RFC 1122 (which modifies RFC 793):

         4.2.2.20  Event Processing: RFC-793 Section 3.9
 
            While it is not strictly required, a TCP SHOULD be capable
            of queueing out-of-order TCP segments. 

A drop causes the segments up to the retransmission of the drop to 
be out-of-order.  This does not preclude "SHOULD implement FIM", but
the above Implication is overdone in my view as it appears to condone
drops of all out-of-order segments.



_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Tue Jan 29 05:05:54 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id FAA29140
	for <tsvwg-archive@odin.ietf.org>; Tue, 29 Jan 2002 05:05:54 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id FAA20406
	for tsvwg-archive@odin.ietf.org; Tue, 29 Jan 2002 05:05:57 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id EAA19424;
	Tue, 29 Jan 2002 04:46:57 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id EAA19396
	for <tsvwg@optimus.ietf.org>; Tue, 29 Jan 2002 04:46:55 -0500 (EST)
Received: from mxic1.isus.emc.com ([168.159.129.100])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id EAA28945
	for <tsvwg@ietf.org>; Tue, 29 Jan 2002 04:46:52 -0500 (EST)
From: Black_David@emc.com
Received: by mxic1.isus.emc.com with Internet Mail Service (5.5.2653.19)
	id <DA2YHFC3>; Tue, 29 Jan 2002 04:49:11 -0500
Message-ID: <277DD60FB639D511AC0400B0D068B71E029373E3@CORPMX14>
To: dotis@sanlight.net, ips@ece.cmu.edu
Cc: tsvwg@ietf.org
Date: Tue, 29 Jan 2002 04:32:58 -0500
MIME-Version: 1.0
X-Mailer: Internet Mail Service (5.5.2653.19)
Content-Type: text/plain;
	charset="iso-8859-1"
Subject: [Tsvwg] RE: iSCSI: Framing Steps
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

A few more clarifications:

(1) If the data-dependent memory allocator below TCP gets confused,
	that's its problem, and iSCSI's problem, not TCP's.  As long
	as the inbound data is presented to TCP in the order it came
	off the wire, TCP's processing behavior is unchanged.  If the
	data is delivered by TCP in locations other than those iSCSI
	was expecting, it can be copied to the right locations with
	suitable safeguards.

(2) FIM processing makes no changes to packet contents, period.
	I have no idea where Doug got the impression "it is not the packet
	location that is being changed, but the content of the packet
	or it would be of little value".  That is simply not true -
	the FIM markers can't be removed below TCP because they were
	sent in the TCP bytestream and have to be presented to TCP.	
	The contents of each packet may not be contiguous in
	memory, but that has been common in TCP implementations
	from the beginning (e.g., see the original BSD mbuf code).

(3) Zero-copy TCP is irrelevant to this discussion.  In an integrated
	TCP/FIM/iSCSI implementation, there would be no address boundaries
	between any of the components and hence no need to avoid the copy
	across address boundaries that is the motivation for zero-copy TCP.
	Zero copy TCP schemes cannot achieve "appropriate packet placement"
	for iSCSI by themselves because they're not capable of ensuring that
	a 4k SCSI data payload lands in 4k of contiguous memory aligned to
	a 4k boundary, and hence will cause copies to be made.

Beyond this, Doug continues to claim that there are new security
issues in FIM but has failed to describe any significant new risks,
aside from a general concern that implementation work could introduce
security bugs - that is the case for most IETF protocols, and is not
a reason to stop working on protocols.

Thanks,
--David

---------------------------------------------------
David L. Black, Senior Technologist
EMC Corporation, 42 South St., Hopkinton, MA  01748
+1 (508) 249-6449 *NEW*      FAX: +1 (508) 497-8500
black_david@emc.com         Cell: +1 (978) 394-7754
---------------------------------------------------

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@ns.ietf.org  Tue Jan 29 17:02:55 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id RAA22332
	for <tsvwg-archive@odin.ietf.org>; Tue, 29 Jan 2002 17:02:55 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id RAA03241
	for tsvwg-archive@odin.ietf.org; Tue, 29 Jan 2002 17:02:59 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id QAA01460;
	Tue, 29 Jan 2002 16:32:31 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id QAA01429
	for <tsvwg@ns.ietf.org>; Tue, 29 Jan 2002 16:32:29 -0500 (EST)
Received: from c003.snv.cp.net (c003-h000.c003.snv.cp.net [209.228.32.214])
	by ietf.org (8.9.1a/8.9.1a) with SMTP id QAA21572
	for <tsvwg@ietf.org>; Tue, 29 Jan 2002 16:32:24 -0500 (EST)
Received: (cpmta 4346 invoked from network); 29 Jan 2002 13:31:55 -0800
Received: from 64.130.130.105 (HELO dougrmt)
  by smtp.telocity.com (209.228.32.214) with SMTP; 29 Jan 2002 13:31:55 -0800
X-Sent: 29 Jan 2002 21:31:55 GMT
From: "Douglas Otis" <dotis@sanlight.net>
To: <ips@ece.cmu.edu>
Cc: <tsvwg@ietf.org>
Subject: RE: [Tsvwg] RE: iSCSI: Framing Steps
Date: Tue, 29 Jan 2002 13:33:18 -0800
Message-ID: <NEBBJGDMMLHHCIKHGBEJCEEKCPAA.dotis@sanlight.net>
MIME-Version: 1.0
Content-Type: text/plain;
	charset="iso-8859-1"
Content-Transfer-Encoding: 7bit
X-Priority: 3 (Normal)
X-MSMail-Priority: Normal
X-Mailer: Microsoft Outlook IMO, Build 9.0.2416 (9.0.2910.0)
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4807.1700
Importance: Normal
In-Reply-To: <277DD60FB639D511AC0400B0D068B71E029373E3@CORPMX14>
Content-Transfer-Encoding: 7bit
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit

Shridhar, David,

>Shridhar Mukund wrote:
>>A few more clarifications:
>>
>>(1) If the data-dependent memory allocator below TCP gets confused,
>>    that's its problem, and iSCSI's problem, not TCP's.  As long
>> 	as the inbound data is presented to TCP in the order it came
>> 	off the wire, TCP's processing behavior is unchanged.  If the
>> 	data is delivered by TCP in locations other than those iSCSI
>> 	was expecting, it can be copied to the right locations with
>>    suitable safeguards.

Let's say the packet received off of the wire is not in contagious memory
(as typical) but is a memory array.  The layer below TCP constructs this
array to present information in the correct location as per the iSCSI
structures found within the content of the expected TCP byte stream.  This
assignment process will require prior state to be considered in performing
this assignment, in addition to sharing the identification of these
locations with the iSCSI application.  Should there be an error in this
process, the normal expectations of correct placement is no longer valid as
a result.  As there must already be communication between the application
and this "below TCP layer process", posting error information seems like a
small detail.  Indeed, one could assume the application would track each
piece of information in this array to detect an error, but this would not
likely the solution used as the intent is to minimize overhead.  I would
hope that if this scheme is to be used, a definition of the information
relayed between the application and the application specific "below TCP
layer process" would be included within the description of this technique.

The safeguards would need to include that this layer acts to prevent
duplications to avoid overwritting information that may be already found in
a prior array sent to TCP.  Should TCP processing of these packets not match
exactly with this "below TCP layer process", there would be potential for
these differences to create problems.  In addition to matching the view of
TCP within this "below TCP layer process", errors found in the processing of
the iSCSI structures within the "below TCP layer process" will also need
special handling.  The simple act of placing the iSCSI data into the correct
location based on the content of the TCP byte stream implies that the "below
TCP layer process" understand beforehand this byte stream.

>>(2) FIM processing makes no changes to packet contents, period.
>>	I have no idea where Doug got the impression "it is not the packet
>> 	location that is being changed, but the content of the packet
>> 	or it would be of little value".  That is simply not true -
>> 	the FIM markers can't be removed below TCP because they were
>> 	sent in the TCP bytestream and have to be presented to TCP.
>> 	The contents of each packet may not be contiguous in
>> 	memory, but that has been common in TCP implementations
>> 	from the beginning (e.g., see the original BSD mbuf code).

It is not a reassignment of the packet, but detailed placement of the
content within the packet that I was referencing.  This operation is not
done on the packet as a whole, but to many portions of information found
within the packet.  This operation is not a simple, linear, sequential
placement of information.  An array could make such appear as a contagious
packet following this operation.  I agree with that point.  I was attempting
to make clear the function that was being performed and the level of
complexity involved.

>>(3) Zero-copy TCP is irrelevant to this discussion.  In an integrated
>> 	TCP/FIM/iSCSI implementation, there would be no address boundaries
>> 	between any of the components and hence no need to avoid the copy
>> 	across address boundaries that is the motivation for zero-copy TCP.
>>	Zero copy TCP schemes cannot achieve "appropriate packet placement"
>> 	for iSCSI by themselves because they're not capable of ensuring that
>> 	a 4k SCSI data payload lands in 4k of contiguous memory aligned to
>> 	a 4k boundary, and hence will cause copies to be made.

The description that David was using sounded as if he was describing Zero
Copy TCP.  I felt there was detail lacking in this description.  iSCSI
structures would not ensure page alignment, and the process of placing and
splicing together prior information would be part of this "below TCP layer
process."  In addition, there would be a requirement to hold some packets
should information arrive not aligned within the TCP segment or before the
marker.  This additional requirement together with a more complex state
becomes motivation for transitioning this scheme toward creating a framed
version of TCP.  My argument from the very beginning was that we should not
consider changing TCP and that SCTP provided the needed framing as well as
removed the need for placing the application below the transport.  This
debate has evolved into semantics as to what constitutes a change to TCP.
Placing the application below TCP seems like a change.

David Black wrote:
>
> Beyond this, Doug continues to claim that there are new security
> issues in FIM but has failed to describe any significant new risks,
> aside from a general concern that implementation work could introduce
> security bugs - that is the case for most IETF protocols, and is not
> a reason to stop working on protocols.

Perhaps a full description of the "below TCP layer process" would be
helpful.  Examining the state created, the types of commutations relayed
between this layer and the application would be needed for assessment.  If
the goal is to allow applications a common service or API as was once
possible with TCP, then these definitions are important and to allow risks
to be assessed.  My added concern was that using an interval that was a
power of 2 would allow a prediction of a where the mark would be found in
the future.  I think it would be incumbent to present full details of this
scheme to allow a complete review and for a standardize interface to emerge.
The state found in this "below TCP layer process" in conjunction with the
types of application communications are potential areas where security may
become problematic.

Doug


_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@ns.ietf.org  Tue Jan 29 18:12:59 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id SAA23893
	for <tsvwg-archive@odin.ietf.org>; Tue, 29 Jan 2002 18:12:54 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id SAA06691
	for tsvwg-archive@odin.ietf.org; Tue, 29 Jan 2002 18:12:57 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id RAA05337;
	Tue, 29 Jan 2002 17:53:06 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id RAA05306
	for <tsvwg@ns.ietf.org>; Tue, 29 Jan 2002 17:53:04 -0500 (EST)
Received: from mg03.austin.ibm.com (mg03.austin.ibm.com [192.35.232.20])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id RAA23632
	for <tsvwg@ietf.org>; Tue, 29 Jan 2002 17:53:00 -0500 (EST)
Received: from austin.ibm.com (netmail1.austin.ibm.com [9.3.7.138])
	by mg03.austin.ibm.com (AIX4.3/8.9.3/8.9.3) with ESMTP id QAA21704;
	Tue, 29 Jan 2002 16:50:12 -0600
Received: from austin.ibm.com (ambika.austin.ibm.com [9.53.150.77])
	by austin.ibm.com (AIX4.3/8.9.3/8.9.3) with ESMTP id QAA32560;
	Tue, 29 Jan 2002 16:53:01 -0600
Message-ID: <3C5727CE.61244CF7@austin.ibm.com>
Date: Tue, 29 Jan 2002 16:53:02 -0600
From: venkat venkatsubra <venkats@austin.ibm.com>
Organization: IBM
X-Mailer: Mozilla 4.76iC-CCK-MCD  [en_US] (X11; U; AIX 5.1)
X-Accept-Language: en
MIME-Version: 1.0
To: tsvwg@ietf.org
CC: daisyc@us.ibm.com
References: <200201251700.MAA26662@optimus.ietf.org>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Transfer-Encoding: 7bit
Subject: [Tsvwg] Re: tsvwg digest, Vol 1 #486 - 4 msgs
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit

Daisy,

> To: Randall Stewart <randall@stewart.chicago.il.us>
> Cc: tsvwg@ietf.org, sctp-developers-list@cig.mot.com
> From: "Daisy Chang" <daisyc@us.ibm.com>
> Date: Thu, 24 Jan 2002 11:34:20 -0600
> Subject: [Tsvwg] Re: question regarding the SCTP API - SCTP_INIT cmsg
>
> Yes, Randy, I think that the same behavior exists in Linux for TCP and UDP.
> Well, for TCP, appending 0 bytes to the output queue really is harmless,
> nothing is affected, so it is perfectly OK. For UDP, it seems a little
> wierd if it
> does result into a NULL message on the wire. But for a datagram protocol,
> users are taking more control than they want to in their own hands, so I
> guess
> the protocol has no reason to reject it.
>
> As for SCTP, I think that it makes a lot of sense to disallow NULL user
> messages. I was a bit confused by the wording in 3.3.1 Payload Data
> (RFC2960) where it says "A DATA chunk with no user data field will have
> Length set to 16." But Jon Grimm graciously cleared it up by pointing me to
> the similar place in your book where it says "It should have a value equal
> to or greater than 17, because a DATA chunk is required to have at least
> one byte of user data." I gathered that this is what you refered to when
> you
> said that the RFC disallows 0-length send, right?

The last few lines of section 6.2 from rfc2960:

   If an endpoint receives a DATA chunk with no user data (i.e., the
   Length field is set to 16) it MUST send an ABORT with error cause set
   to "No User Data".

   An endpoint SHOULD NOT send a DATA chunk with no user data part.

This must be what Randall must have referred to.

> Thanks for your quick responses. I would suggest that we shall follow
> your interpretation for the sendmsg and connect implementation.
>
> Thanks,
> Daisy
>
> Daisy Chang
> IBM Linux Technology Center
> daisyc@us.ibm.com
> Tel: (512)-838-4194    T/L: 678-4194
> FAX: (512)-838-4663

Venkat


_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@ns.ietf.org  Tue Jan 29 18:39:33 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id SAA24563
	for <tsvwg-archive@odin.ietf.org>; Tue, 29 Jan 2002 18:39:33 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id SAA08025
	for tsvwg-archive@odin.ietf.org; Tue, 29 Jan 2002 18:39:36 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id SAA06729;
	Tue, 29 Jan 2002 18:13:00 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id SAA06703
	for <tsvwg@ns.ietf.org>; Tue, 29 Jan 2002 18:12:59 -0500 (EST)
Received: from e21.nc.us.ibm.com (e21.nc.us.ibm.com [32.97.136.227])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id SAA23897
	for <tsvwg@ietf.org>; Tue, 29 Jan 2002 18:12:55 -0500 (EST)
Received: from southrelay02.raleigh.ibm.com (southrelay02.raleigh.ibm.com [9.37.3.209])
	by e21.nc.us.ibm.com (8.9.3/8.9.3) with ESMTP id RAA62380
	for <tsvwg@ietf.org>; Tue, 29 Jan 2002 17:08:27 -0600
Received: from d04nm110.raleigh.ibm.com (d04nm110.raleigh.ibm.com [9.27.5.85])
	by southrelay02.raleigh.ibm.com (8.11.1m3/NCO v5.01) with ESMTP id g0TNCvo228614;
	Tue, 29 Jan 2002 18:12:57 -0500
Importance: Normal
Subject: Re: [Tsvwg] Re: tsvwg digest, Vol 1 #486 - 4 msgs
To: venkat venkatsubra <venkats@austin.ibm.com>
Cc: tsvwg@ietf.org
From: "Daisy Chang" <daisyc@us.ibm.com>
Date: Tue, 29 Jan 2002 18:12:56 -0500
Message-ID: <OFBE2E41A0.99280165-ON85256B50.007E695C@raleigh.ibm.com>
X-MIMETrack: Serialize by Router on D04NM110/04/M/IBM(Release 5.0.9 |November 16, 2001) at
 01/29/2002 06:12:57 PM
MIME-Version: 1.0
Content-type: text/plain; charset=us-ascii
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

Venkat,
Yes, thanks for pointing it out. I mainly wanted to make sure that there
are no inconsistencies
in the RFC regarding to the "no user data field" rule. So, to stick with
the RFC, looks like we
have to
1. check the message size in API to prevent from sending a 0-length message
as a sender;
2. check the message size in data reception as a receiver and send ABORT to
violators
    of this rule.


Thanks,
Daisy

Daisy Chang
IBM Linux Technology Center
daisyc@us.ibm.com
Tel: (512)-838-4194    T/L: 678-4194
FAX: (512)-838-4663


venkat venkatsubra <venkats@austin.ibm.com>@ietf.org on 01/29/2002 04:53:02
PM

Sent by:    tsvwg-admin@ietf.org


To:    tsvwg@ietf.org
cc:    Daisy Chang/Austin/IBM@IBMUS
Subject:    [Tsvwg] Re: tsvwg digest, Vol 1 #486 - 4 msgs



Daisy,

> To: Randall Stewart <randall@stewart.chicago.il.us>
> Cc: tsvwg@ietf.org, sctp-developers-list@cig.mot.com
> From: "Daisy Chang" <daisyc@us.ibm.com>
> Date: Thu, 24 Jan 2002 11:34:20 -0600
> Subject: [Tsvwg] Re: question regarding the SCTP API - SCTP_INIT cmsg
>
> Yes, Randy, I think that the same behavior exists in Linux for TCP and
UDP.
> Well, for TCP, appending 0 bytes to the output queue really is harmless,
> nothing is affected, so it is perfectly OK. For UDP, it seems a little
> wierd if it
> does result into a NULL message on the wire. But for a datagram protocol,
> users are taking more control than they want to in their own hands, so I
> guess
> the protocol has no reason to reject it.
>
> As for SCTP, I think that it makes a lot of sense to disallow NULL user
> messages. I was a bit confused by the wording in 3.3.1 Payload Data
> (RFC2960) where it says "A DATA chunk with no user data field will have
> Length set to 16." But Jon Grimm graciously cleared it up by pointing me
to
> the similar place in your book where it says "It should have a value
equal
> to or greater than 17, because a DATA chunk is required to have at least
> one byte of user data." I gathered that this is what you refered to when
> you
> said that the RFC disallows 0-length send, right?

The last few lines of section 6.2 from rfc2960:

   If an endpoint receives a DATA chunk with no user data (i.e., the
   Length field is set to 16) it MUST send an ABORT with error cause set
   to "No User Data".

   An endpoint SHOULD NOT send a DATA chunk with no user data part.

This must be what Randall must have referred to.

> Thanks for your quick responses. I would suggest that we shall follow
> your interpretation for the sendmsg and connect implementation.
>
> Thanks,
> Daisy
>
> Daisy Chang
> IBM Linux Technology Center
> daisyc@us.ibm.com
> Tel: (512)-838-4194    T/L: 678-4194
> FAX: (512)-838-4663

Venkat


_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@ns.ietf.org  Tue Jan 29 19:10:51 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id TAA25124
	for <tsvwg-archive@odin.ietf.org>; Tue, 29 Jan 2002 19:10:42 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id TAA09086
	for tsvwg-archive@odin.ietf.org; Tue, 29 Jan 2002 19:10:43 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id SAA07896;
	Tue, 29 Jan 2002 18:33:52 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id SAA07865
	for <tsvwg@ns.ietf.org>; Tue, 29 Jan 2002 18:33:50 -0500 (EST)
Received: from sj-msg-core-1.cisco.com (sj-msg-core-1.cisco.com [171.71.163.11])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id SAA24422
	for <tsvwg@ietf.org>; Tue, 29 Jan 2002 18:33:46 -0500 (EST)
Received: from mira-sjc5-2.cisco.com (mira-sjc5-2.cisco.com [171.71.163.16])
	by sj-msg-core-1.cisco.com (8.11.3/8.9.1) with ESMTP id g0TNXJL07844;
	Tue, 29 Jan 2002 15:33:19 -0800 (PST)
Received: from stewart.chicago.il.us (ssh-sj1.cisco.com [171.68.225.134])
	by mira-sjc5-2.cisco.com (Mirapoint)
	with ESMTP id AAY37282;
	Tue, 29 Jan 2002 15:33:18 -0800 (PST)
Message-ID: <3C57313D.8C221557@stewart.chicago.il.us>
Date: Tue, 29 Jan 2002 17:33:18 -0600
From: Randall Stewart <randall@stewart.chicago.il.us>
X-Mailer: Mozilla 4.76 [en] (X11; U; Linux 2.2.12 i386)
X-Accept-Language: en
MIME-Version: 1.0
To: Daisy Chang <daisyc@us.ibm.com>
CC: venkat venkatsubra <venkats@austin.ibm.com>, tsvwg@ietf.org
Subject: Re: [Tsvwg] Re: tsvwg digest, Vol 1 #486 - 4 msgs
References: <OFBE2E41A0.99280165-ON85256B50.007E695C@raleigh.ibm.com>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Transfer-Encoding: 7bit
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit


Yep..

You must check both.. for now my implementation
gives you an EINVAL if you try to send a 0
lenght message and have no msgcontrol structure
that either causes an ABORT or a SHUTDOWN...

Maybe this discussion should move off of tsvwg and over to
sctp-developers-list@cig.mot.com... I don't know if 
this is of general interest any more :>

And by the way I am only 700 msg's or so behind in that
list so make sure and copy me directly if you want
me to chime in :=0

R

Daisy Chang wrote:
> 
> Venkat,
> Yes, thanks for pointing it out. I mainly wanted to make sure that there
> are no inconsistencies
> in the RFC regarding to the "no user data field" rule. So, to stick with
> the RFC, looks like we
> have to
> 1. check the message size in API to prevent from sending a 0-length message
> as a sender;
> 2. check the message size in data reception as a receiver and send ABORT to
> violators
>     of this rule.
> 
> Thanks,
> Daisy
> 
> Daisy Chang
> IBM Linux Technology Center
> daisyc@us.ibm.com
> Tel: (512)-838-4194    T/L: 678-4194
> FAX: (512)-838-4663
> 
> venkat venkatsubra <venkats@austin.ibm.com>@ietf.org on 01/29/2002 04:53:02
> PM
> 
> Sent by:    tsvwg-admin@ietf.org
> 
> To:    tsvwg@ietf.org
> cc:    Daisy Chang/Austin/IBM@IBMUS
> Subject:    [Tsvwg] Re: tsvwg digest, Vol 1 #486 - 4 msgs
> 
> Daisy,
> 
> > To: Randall Stewart <randall@stewart.chicago.il.us>
> > Cc: tsvwg@ietf.org, sctp-developers-list@cig.mot.com
> > From: "Daisy Chang" <daisyc@us.ibm.com>
> > Date: Thu, 24 Jan 2002 11:34:20 -0600
> > Subject: [Tsvwg] Re: question regarding the SCTP API - SCTP_INIT cmsg
> >
> > Yes, Randy, I think that the same behavior exists in Linux for TCP and
> UDP.
> > Well, for TCP, appending 0 bytes to the output queue really is harmless,
> > nothing is affected, so it is perfectly OK. For UDP, it seems a little
> > wierd if it
> > does result into a NULL message on the wire. But for a datagram protocol,
> > users are taking more control than they want to in their own hands, so I
> > guess
> > the protocol has no reason to reject it.
> >
> > As for SCTP, I think that it makes a lot of sense to disallow NULL user
> > messages. I was a bit confused by the wording in 3.3.1 Payload Data
> > (RFC2960) where it says "A DATA chunk with no user data field will have
> > Length set to 16." But Jon Grimm graciously cleared it up by pointing me
> to
> > the similar place in your book where it says "It should have a value
> equal
> > to or greater than 17, because a DATA chunk is required to have at least
> > one byte of user data." I gathered that this is what you refered to when
> > you
> > said that the RFC disallows 0-length send, right?
> 
> The last few lines of section 6.2 from rfc2960:
> 
>    If an endpoint receives a DATA chunk with no user data (i.e., the
>    Length field is set to 16) it MUST send an ABORT with error cause set
>    to "No User Data".
> 
>    An endpoint SHOULD NOT send a DATA chunk with no user data part.
> 
> This must be what Randall must have referred to.
> 
> > Thanks for your quick responses. I would suggest that we shall follow
> > your interpretation for the sendmsg and connect implementation.
> >
> > Thanks,
> > Daisy
> >
> > Daisy Chang
> > IBM Linux Technology Center
> > daisyc@us.ibm.com
> > Tel: (512)-838-4194    T/L: 678-4194
> > FAX: (512)-838-4663
> 
> Venkat
> 
> _______________________________________________
> tsvwg mailing list
> tsvwg@ietf.org
> https://www1.ietf.org/mailman/listinfo/tsvwg
> 
> _______________________________________________
> tsvwg mailing list
> tsvwg@ietf.org
> https://www1.ietf.org/mailman/listinfo/tsvwg

-- 
Randall R. Stewart
randall@stewart.chicago.il.us 815-342-5222 (cell phone)

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Wed Jan 30 07:22:40 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id HAA16019
	for <tsvwg-archive@odin.ietf.org>; Wed, 30 Jan 2002 07:22:39 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id HAA21308
	for tsvwg-archive@odin.ietf.org; Wed, 30 Jan 2002 07:22:40 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id HAA20590;
	Wed, 30 Jan 2002 07:02:28 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id HAA20559
	for <tsvwg@optimus.ietf.org>; Wed, 30 Jan 2002 07:02:26 -0500 (EST)
Received: from CNRI.Reston.VA.US (localhost [127.0.0.1])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id HAA15612;
	Wed, 30 Jan 2002 07:02:25 -0500 (EST)
Message-Id: <200201301202.HAA15612@ietf.org>
Mime-Version: 1.0
Content-Type: Multipart/Mixed; Boundary="NextPart"
To: IETF-Announce: ;
Cc: tsvwg@ietf.org
From: Internet-Drafts@ietf.org
Reply-to: Internet-Drafts@ietf.org
Date: Wed, 30 Jan 2002 07:02:24 -0500
Subject: [Tsvwg] I-D ACTION:draft-ietf-tsvwg-addip-sctp-04.txt
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

--NextPart

A New Internet-Draft is available from the on-line Internet-Drafts directories.
This draft is a work item of the Transport Area Working Group Working Group of the IETF.

	Title		: SCTP Extensions for Dynamic Reconfiguration of IP 
                          Addresses
	Author(s)	: R. Stewart et al.
	Filename	: draft-ietf-tsvwg-addip-sctp-04.txt
	Pages		: 21
	Date		: 29-Jan-02
	
This document describes extensions to the Stream Control
Transmission Protocol (SCTP) [RFC2960] that provide methods to (1)
reconfigure IP address information on an existing association and
(2) request that a peer set flow limits in units of bytes or
messages, either on a per-stream or per-association basis.

A URL for this Internet-Draft is:
http://www.ietf.org/internet-drafts/draft-ietf-tsvwg-addip-sctp-04.txt

To remove yourself from the IETF Announcement list, send a message to 
ietf-announce-request with the word unsubscribe in the body of the message.

Internet-Drafts are also available by anonymous FTP. Login with the username
"anonymous" and a password of your e-mail address. After logging in,
type "cd internet-drafts" and then
	"get draft-ietf-tsvwg-addip-sctp-04.txt".

A list of Internet-Drafts directories can be found in
http://www.ietf.org/shadow.html 
or ftp://ftp.ietf.org/ietf/1shadow-sites.txt


Internet-Drafts can also be obtained by e-mail.

Send a message to:
	mailserv@ietf.org.
In the body type:
	"FILE /internet-drafts/draft-ietf-tsvwg-addip-sctp-04.txt".
	
NOTE:	The mail server at ietf.org can return the document in
	MIME-encoded form by using the "mpack" utility.  To use this
	feature, insert the command "ENCODING mime" before the "FILE"
	command.  To decode the response(s), you will need "munpack" or
	a MIME-compliant mail reader.  Different MIME-compliant mail readers
	exhibit different behavior, especially when dealing with
	"multipart" MIME messages (i.e. documents which have been split
	up into multiple messages), so check your local documentation on
	how to manipulate these messages.
		
		
Below is the data which will enable a MIME compliant mail reader
implementation to automatically retrieve the ASCII version of the
Internet-Draft.

--NextPart
Content-Type: Multipart/Alternative; Boundary="OtherAccess"

--OtherAccess
Content-Type: Message/External-body;
	access-type="mail-server";
	server="mailserv@ietf.org"

Content-Type: text/plain
Content-ID:	<20020129135941.I-D@ietf.org>

ENCODING mime
FILE /internet-drafts/draft-ietf-tsvwg-addip-sctp-04.txt

--OtherAccess
Content-Type: Message/External-body;
	name="draft-ietf-tsvwg-addip-sctp-04.txt";
	site="ftp.ietf.org";
	access-type="anon-ftp";
	directory="internet-drafts"

Content-Type: text/plain
Content-ID:	<20020129135941.I-D@ietf.org>

--OtherAccess--

--NextPart--



_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Wed Jan 30 07:22:49 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id HAA16042
	for <tsvwg-archive@odin.ietf.org>; Wed, 30 Jan 2002 07:22:49 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id HAA21332
	for tsvwg-archive@odin.ietf.org; Wed, 30 Jan 2002 07:22:50 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id HAA20666;
	Wed, 30 Jan 2002 07:02:43 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id HAA20637
	for <tsvwg@optimus.ietf.org>; Wed, 30 Jan 2002 07:02:41 -0500 (EST)
Received: from CNRI.Reston.VA.US (localhost [127.0.0.1])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id HAA15656;
	Wed, 30 Jan 2002 07:02:39 -0500 (EST)
Message-Id: <200201301202.HAA15656@ietf.org>
Mime-Version: 1.0
Content-Type: Multipart/Mixed; Boundary="NextPart"
To: IETF-Announce: ;
Cc: tsvwg@ietf.org
From: Internet-Drafts@ietf.org
Reply-to: Internet-Drafts@ietf.org
Date: Wed, 30 Jan 2002 07:02:39 -0500
Subject: [Tsvwg] I-D ACTION:draft-ietf-tsvwg-sctpsocket-03.txt
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

--NextPart

A New Internet-Draft is available from the on-line Internet-Drafts directories.
This draft is a work item of the Transport Area Working Group Working Group of the IETF.

	Title		: Sockets API Extensions for SCTP
	Author(s)	: R. Stewart et al.
	Filename	: draft-ietf-tsvwg-sctpsocket-03.txt
	Pages		: 56
	Date		: 29-Jan-02
	
This document describes a mapping of the Stream Control Transmission
Protocol [SCTP] into a sockets API. The benefits of this mapping
include compatibility for TCP applications, access to new SCTP
features and a consolidated error and event notification scheme.

A URL for this Internet-Draft is:
http://www.ietf.org/internet-drafts/draft-ietf-tsvwg-sctpsocket-03.txt

To remove yourself from the IETF Announcement list, send a message to 
ietf-announce-request with the word unsubscribe in the body of the message.

Internet-Drafts are also available by anonymous FTP. Login with the username
"anonymous" and a password of your e-mail address. After logging in,
type "cd internet-drafts" and then
	"get draft-ietf-tsvwg-sctpsocket-03.txt".

A list of Internet-Drafts directories can be found in
http://www.ietf.org/shadow.html 
or ftp://ftp.ietf.org/ietf/1shadow-sites.txt


Internet-Drafts can also be obtained by e-mail.

Send a message to:
	mailserv@ietf.org.
In the body type:
	"FILE /internet-drafts/draft-ietf-tsvwg-sctpsocket-03.txt".
	
NOTE:	The mail server at ietf.org can return the document in
	MIME-encoded form by using the "mpack" utility.  To use this
	feature, insert the command "ENCODING mime" before the "FILE"
	command.  To decode the response(s), you will need "munpack" or
	a MIME-compliant mail reader.  Different MIME-compliant mail readers
	exhibit different behavior, especially when dealing with
	"multipart" MIME messages (i.e. documents which have been split
	up into multiple messages), so check your local documentation on
	how to manipulate these messages.
		
		
Below is the data which will enable a MIME compliant mail reader
implementation to automatically retrieve the ASCII version of the
Internet-Draft.

--NextPart
Content-Type: Multipart/Alternative; Boundary="OtherAccess"

--OtherAccess
Content-Type: Message/External-body;
	access-type="mail-server";
	server="mailserv@ietf.org"

Content-Type: text/plain
Content-ID:	<20020129140006.I-D@ietf.org>

ENCODING mime
FILE /internet-drafts/draft-ietf-tsvwg-sctpsocket-03.txt

--OtherAccess
Content-Type: Message/External-body;
	name="draft-ietf-tsvwg-sctpsocket-03.txt";
	site="ftp.ietf.org";
	access-type="anon-ftp";
	directory="internet-drafts"

Content-Type: text/plain
Content-ID:	<20020129140006.I-D@ietf.org>

--OtherAccess--

--NextPart--



_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Wed Jan 30 08:16:55 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id HAA16017
	for <tsvwg-archive@odin.ietf.org>; Wed, 30 Jan 2002 07:22:39 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id HAA21312
	for tsvwg-archive@odin.ietf.org; Wed, 30 Jan 2002 07:22:40 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id HAA20629;
	Wed, 30 Jan 2002 07:02:38 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id HAA20598
	for <tsvwg@optimus.ietf.org>; Wed, 30 Jan 2002 07:02:35 -0500 (EST)
Received: from CNRI.Reston.VA.US (localhost [127.0.0.1])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id HAA15640;
	Wed, 30 Jan 2002 07:02:34 -0500 (EST)
Message-Id: <200201301202.HAA15640@ietf.org>
Mime-Version: 1.0
Content-Type: Multipart/Mixed; Boundary="NextPart"
To: IETF-Announce: ;
Cc: tsvwg@ietf.org
From: Internet-Drafts@ietf.org
Reply-to: Internet-Drafts@ietf.org
Date: Wed, 30 Jan 2002 07:02:33 -0500
Subject: [Tsvwg] I-D ACTION:draft-ietf-tsvwg-sctpimpguide-03.txt
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

--NextPart

A New Internet-Draft is available from the on-line Internet-Drafts directories.
This draft is a work item of the Transport Area Working Group Working Group of the IETF.

	Title		: SCTP Implementors Guide
	Author(s)	: R. Stewart, L. Ong, I. Rodriguez, K. Poon
	Filename	: draft-ietf-tsvwg-sctpimpguide-03.txt
	Pages		: 30
	Date		: 29-Jan-02
	
This document contains a compilation of all defects found up until
June 2001 for the Stream Control Transmission Protocol (SCTP)
[RFC2960]. These defects may be of an editorial or technical nature.
This document may be thought of as a companion document to be
used in the implementation of SCTP to clearify errors in the
original SCTP document.
This document updates RFC2960 and text within this document
supersedes the text found in RFC2960

A URL for this Internet-Draft is:
http://www.ietf.org/internet-drafts/draft-ietf-tsvwg-sctpimpguide-03.txt

To remove yourself from the IETF Announcement list, send a message to 
ietf-announce-request with the word unsubscribe in the body of the message.

Internet-Drafts are also available by anonymous FTP. Login with the username
"anonymous" and a password of your e-mail address. After logging in,
type "cd internet-drafts" and then
	"get draft-ietf-tsvwg-sctpimpguide-03.txt".

A list of Internet-Drafts directories can be found in
http://www.ietf.org/shadow.html 
or ftp://ftp.ietf.org/ietf/1shadow-sites.txt


Internet-Drafts can also be obtained by e-mail.

Send a message to:
	mailserv@ietf.org.
In the body type:
	"FILE /internet-drafts/draft-ietf-tsvwg-sctpimpguide-03.txt".
	
NOTE:	The mail server at ietf.org can return the document in
	MIME-encoded form by using the "mpack" utility.  To use this
	feature, insert the command "ENCODING mime" before the "FILE"
	command.  To decode the response(s), you will need "munpack" or
	a MIME-compliant mail reader.  Different MIME-compliant mail readers
	exhibit different behavior, especially when dealing with
	"multipart" MIME messages (i.e. documents which have been split
	up into multiple messages), so check your local documentation on
	how to manipulate these messages.
		
		
Below is the data which will enable a MIME compliant mail reader
implementation to automatically retrieve the ASCII version of the
Internet-Draft.

--NextPart
Content-Type: Multipart/Alternative; Boundary="OtherAccess"

--OtherAccess
Content-Type: Message/External-body;
	access-type="mail-server";
	server="mailserv@ietf.org"

Content-Type: text/plain
Content-ID:	<20020129135954.I-D@ietf.org>

ENCODING mime
FILE /internet-drafts/draft-ietf-tsvwg-sctpimpguide-03.txt

--OtherAccess
Content-Type: Message/External-body;
	name="draft-ietf-tsvwg-sctpimpguide-03.txt";
	site="ftp.ietf.org";
	access-type="anon-ftp";
	directory="internet-drafts"

Content-Type: text/plain
Content-ID:	<20020129135954.I-D@ietf.org>

--OtherAccess--

--NextPart--



_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Wed Jan 30 08:40:01 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id IAA17972
	for <tsvwg-archive@odin.ietf.org>; Wed, 30 Jan 2002 08:39:57 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id IAA25617
	for tsvwg-archive@odin.ietf.org; Wed, 30 Jan 2002 08:39:58 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id IAA23891;
	Wed, 30 Jan 2002 08:15:45 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id IAA23864
	for <tsvwg@optimus.ietf.org>; Wed, 30 Jan 2002 08:15:43 -0500 (EST)
Received: from mgw-x2.nokia.com (mgw-x2.nokia.com [131.228.20.22])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id IAA17274
	for <tsvwg@ietf.org>; Wed, 30 Jan 2002 08:15:40 -0500 (EST)
From: john.loughney@nokia.com
Received: from esvir02nok.ntc.nokia.com (esvir02nokt.ntc.nokia.com [172.21.143.34])
	by mgw-x2.nokia.com (Switch-2.1.0/Switch-2.1.0) with ESMTP id g0UDFiZ09974
	for <tsvwg@ietf.org>; Wed, 30 Jan 2002 15:15:44 +0200 (EET)
Received: from esebh002.NOE.Nokia.com (unverified) by esvir02nok.ntc.nokia.com
 (Content Technologies SMTPRS 4.2.5) with ESMTP id <T58c45fe0b7ac158f22077@esvir02nok.ntc.nokia.com>;
 Wed, 30 Jan 2002 15:15:39 +0200
Received: from esebe004.NOE.Nokia.com ([172.21.138.44]) by esebh002.NOE.Nokia.com with Microsoft SMTPSVC(5.0.2195.3779);
	 Wed, 30 Jan 2002 15:15:39 +0200
X-MimeOLE: Produced By Microsoft Exchange V6.0.5762.3
content-class: urn:content-classes:message
MIME-Version: 1.0
Content-Type: text/plain;
	charset="iso-8859-1"
Date: Wed, 30 Jan 2002 15:15:39 +0200
Message-ID: <0C1353ABB1DEB74DB067ADFF749C4EEF09BC84@esebe004.NOE.Nokia.com>
Thread-Topic: [Tsvwg] Comment on:I-D ACTION:draft-ietf-tsvwg-sctpcsum-02.txt
Thread-Index: AcGiulbLaB1D/nMvQjiTjiSfOBgeygAcuYGwAZiFycA=
To: <tsvwg@ietf.org>
Cc: <lln@cdt.luth.se>, <micke@cs.arizona.edu>, <steve@cs.arizona.edu>
X-OriginalArrivalTime: 30 Jan 2002 13:15:39.0844 (UTC) FILETIME=[36895840:01C1A990]
Content-Transfer-Encoding: 8bit
X-MIME-Autoconverted: from quoted-printable to 8bit by optimus.ietf.org id IAA23865
Subject: [Tsvwg] Comment on: draft-ietf-tsvwg-udp-lite-00.txt
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 8bit

Hi all,

First off, my understanding from Salt Lake City is that UDP-lite would
be ready for WG last call. If this is true, consider my comment
applicable for last call.

My comment regards the applicability of UDP-lite for header compression.

RFC3095 ROHC header compression cannot be used for UDP Lite. The reason is the following: The classic UDP length field is replaced in UDP Lite by the checkusm coverage. ROHC compressor assumes that the UDP length field is redundant, i.e., it is not sent over the link, and can be inferred from other information in the decompressor (see references in the end of my e-mail). Naturally, checksum coverage of UDP Lite is not redundant, and cannot be built the same way as UDP lenght field by decompressor. This means that UDP Lite cannot be handled by current ROHC, so a new ROHC profile is needed.

This means that it must be possible to distinguish between classic UDP and UDP Lite, so that in the presence of Header Compression, UDP Lite will be sent as uncompressed and classic UDP as compressed by ROHC. Or later, when there is a new profile for UDP Lite, and it is implemented, the profile could be selected accordingly, based on the protocol identifier. 

If the protocol identifier is shared between classic UDP and UDP Lite, I wonder, how the above-mentioned situation could be handled. Seemingly, there is no other way to distinguish between UDP Lite and classic UDP, so the new protocol identifier is needed. I suggest that the draft's recommendation of a new protocol ID be required and a new IP protocol ID be applied for.

cheers,
John

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Wed Jan 30 08:40:41 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id IAA18003
	for <tsvwg-archive@odin.ietf.org>; Wed, 30 Jan 2002 08:40:41 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id IAA25672
	for tsvwg-archive@odin.ietf.org; Wed, 30 Jan 2002 08:40:42 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id IAA24230;
	Wed, 30 Jan 2002 08:22:31 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id IAA24203
	for <tsvwg@optimus.ietf.org>; Wed, 30 Jan 2002 08:22:30 -0500 (EST)
Received: from mgw-x2.nokia.com (mgw-x2.nokia.com [131.228.20.22])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id IAA17455
	for <tsvwg@ietf.org>; Wed, 30 Jan 2002 08:22:22 -0500 (EST)
From: john.loughney@nokia.com
Received: from esvir04nok.ntc.nokia.com (esvir04nokt.ntc.nokia.com [172.21.143.36])
	by mgw-x2.nokia.com (Switch-2.1.0/Switch-2.1.0) with ESMTP id g0UDMOZ14707
	for <tsvwg@ietf.org>; Wed, 30 Jan 2002 15:22:24 +0200 (EET)
Received: from esebh004.NOE.Nokia.com (unverified) by esvir04nok.ntc.nokia.com
 (Content Technologies SMTPRS 4.2.5) with ESMTP id <T58c465fc93ac158f24077@esvir04nok.ntc.nokia.com>;
 Wed, 30 Jan 2002 15:22:20 +0200
Received: from esebe004.NOE.Nokia.com ([172.21.138.44]) by esebh004.NOE.Nokia.com with Microsoft SMTPSVC(5.0.2195.3779);
	 Wed, 30 Jan 2002 15:22:20 +0200
X-MimeOLE: Produced By Microsoft Exchange V6.0.5762.3
content-class: urn:content-classes:message
MIME-Version: 1.0
Content-Type: text/plain;
	charset="iso-8859-1"
Subject: [Tsvwg] Second Comment on: draft-ietf-tsvwg-udp-lite-00.txt 
Date: Wed, 30 Jan 2002 15:22:19 +0200
Message-ID: <0C1353ABB1DEB74DB067ADFF749C4EEF5D9505@esebe004.NOE.Nokia.com>
Thread-Topic: [Tsvwg] Comment on:I-D ACTION:draft-ietf-tsvwg-sctpcsum-02.txt
Thread-Index: AcGiulbLaB1D/nMvQjiTjiSfOBgeygAcuYGwAZiFycAAAGdYEA==
To: <tsvwg@ietf.org>
Cc: <lln@cdt.luth.se>, <micke@cs.arizona.edu>, <steve@cs.arizona.edu>
X-OriginalArrivalTime: 30 Jan 2002 13:22:20.0204 (UTC) FILETIME=[252B6EC0:01C1A991]
Content-Transfer-Encoding: 8bit
X-MIME-Autoconverted: from quoted-printable to 8bit by optimus.ietf.org id IAA24204
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 8bit

Hi all,

One more comment:

Should this text:

  IP Interface

     As for classic UDP, the UDP Lite module must pass the pseudo header
     to the UDP Lite module.  

actually be:

  IP Interface

     As for classic UDP, the UDP Lite module must pass the pseudo header
     to the IP module.  
            ^^

Thanks!
John

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Wed Jan 30 10:18:32 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id KAA21956
	for <tsvwg-archive@odin.ietf.org>; Wed, 30 Jan 2002 10:18:28 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id KAA02445
	for tsvwg-archive@odin.ietf.org; Wed, 30 Jan 2002 10:18:29 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id JAA01062;
	Wed, 30 Jan 2002 09:59:42 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id JAA00981
	for <tsvwg@optimus.ietf.org>; Wed, 30 Jan 2002 09:59:35 -0500 (EST)
Received: from albatross.wise.edt.ericsson.se (albatross-ext.wise.edt.ericsson.se [194.237.142.116])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id JAA21196
	for <tsvwg@ietf.org>; Wed, 30 Jan 2002 09:59:31 -0500 (EST)
Received: from esealnt462.al.sw.ericsson.se (ESEALNT462.al.sw.ericsson.se [153.88.251.62])
	by albatross.wise.edt.ericsson.se (8.12.1/8.12.1/WIREfire-1.4) with SMTP id g0UExLX4017383
	for <tsvwg@ietf.org>; Wed, 30 Jan 2002 15:59:21 +0100 (MET)
Received: FROM esealnt742.al.sw.ericsson.se BY esealnt462.al.sw.ericsson.se ; Wed Jan 30 15:59:02 2002 +0100
Received: by esealnt742.al.sw.ericsson.se with Internet Mail Service (5.5.2653.19)
	id <YHKC9D66>; Wed, 30 Jan 2002 15:50:05 +0100
Message-ID: <A943FD84BD9ED41193460008C7918050023BC3DD@ESEALNT419.al.sw.ericsson.se>
From: "Lars-Erik Jonsson (EPL)" <Lars-Erik.Jonsson@epl.ericsson.se>
To: "'john.loughney@nokia.com'" <john.loughney@nokia.com>, tsvwg@ietf.org
Cc: lln@cdt.luth.se, micke@cs.arizona.edu, steve@cs.arizona.edu
Subject: RE: [Tsvwg] Comment on: draft-ietf-tsvwg-udp-lite-00.txt
Date: Wed, 30 Jan 2002 15:58:56 +0100
MIME-Version: 1.0
X-Mailer: Internet Mail Service (5.5.2653.19)
Content-Type: text/plain;
	charset="iso-8859-1"
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

John,

I think you may have misread the UDP Lite draft. Also, the
potential header compression problem you mention is not as
difficult as one may believe. See details below...


> RFC3095 ROHC header compression cannot be used for UDP Lite.

It would just not compress if the Coverage fields is not equal 
to Length.
 
> ...so a new ROHC profile is needed.

Yes, to be able to compress it.

> This means that it must be possible to distinguish between 
> classic UDP and UDP Lite, so that in the presence of Header 
> Compression, UDP Lite will be sent as uncompressed and 
> classic UDP as compressed by ROHC. 

An explicit indication is not really needed, se below.

> Or later, when there is a new profile for UDP Lite, and it is
> implemented, the profile could be selected accordingly, based
> on the protocol identifier. 

One profile could support both, se below.
 
> If the protocol identifier is shared between classic UDP and 
> UDP Lite, I wonder, how the above-mentioned situation could 
> be handled. 

According to the UDP Lite draft, it is not the intention to share
protocol identifier with UDP.

   "Therefore, this draft proposes to allocate a new protocol
    identifier for UDP Lite."

   "We request that a new ip protocol identifier is allocated
    for UDP Lite."

For more than a year now, we have been discussing the two alternatives
of either having a separate identifier or share identifier with UDP. For
the reasons described in the draft, the latter alternative has been 
given up. Anyway, I agree with you that the draft could probably be 
clearer in this regard, i.e. avoid wording such as "propose to".

By the way, the header compression aspects you refer to have been
discussed (in Minneapolis last year, I think), and we agreed that for
HC purposes both approaches would work. The reason is that the current
RFC3095 profiles would automatically not compress UDP Lite if the 
coverage field is not equal to length, and new UDP Lite-aware profiles
could be made able to handle (compress) both UDP variants. However, as
mentioned above, this is not an issue any longer since we have agreed
not to share protocol id with UDP.

Cheers,
/Lars-Erik


_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Wed Jan 30 10:39:47 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id KAA22774
	for <tsvwg-archive@odin.ietf.org>; Wed, 30 Jan 2002 10:39:47 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id KAA03697
	for tsvwg-archive@odin.ietf.org; Wed, 30 Jan 2002 10:39:48 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id KAA02217;
	Wed, 30 Jan 2002 10:13:22 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id KAA02173
	for <tsvwg@optimus.ietf.org>; Wed, 30 Jan 2002 10:13:19 -0500 (EST)
Received: from sm.luth.se (ny.sm.luth.se [130.240.3.1])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id KAA21695
	for <tsvwg@ietf.org>; Wed, 30 Jan 2002 10:13:14 -0500 (EST)
Received: from plupp (plupp.cdt.luth.se [130.240.64.16])
	by sm.luth.se (8.10.2+Sun/8.10.0) with SMTP id g0UFD0L04414;
	Wed, 30 Jan 2002 16:13:00 +0100 (MET)
Reply-To: <lln@cdt.luth.se>
From: =?iso-8859-1?Q?Lars-=C5ke_Larzon?= <lln@cdt.luth.se>
To: <john.loughney@nokia.com>, <tsvwg@ietf.org>
Cc: <lln@cdt.luth.se>, <micke@cs.arizona.edu>, <steve@cs.arizona.edu>
Subject: RE: [Tsvwg] Second Comment on: draft-ietf-tsvwg-udp-lite-00.txt 
Date: Wed, 30 Jan 2002 16:13:29 +0100
Message-ID: <JDEOLNENPBGIMMOIDBBCAEJPCEAA.lln@sm.luth.se>
MIME-Version: 1.0
Content-Type: text/plain;
	charset="iso-8859-1"
Content-Transfer-Encoding: 8bit
X-Priority: 3 (Normal)
X-MSMail-Priority: Normal
X-Mailer: Microsoft Outlook IMO, Build 9.0.2416 (9.0.2911.0)
Importance: Normal
In-Reply-To: <0C1353ABB1DEB74DB067ADFF749C4EEF5D9505@esebe004.NOE.Nokia.com>
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4807.1700
Content-Transfer-Encoding: 8bit
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 8bit



> Should this text:
>
>   IP Interface
>
>      As for classic UDP, the UDP Lite module must pass the pseudo header
>      to the UDP Lite module.
>
> actually be:
>
>   IP Interface
>
>      As for classic UDP, the UDP Lite module must pass the pseudo header
>      to the IP module.
>             ^^

You are right.

/Lars-흆e


_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Wed Jan 30 13:52:48 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id NAA28954
	for <tsvwg-archive@odin.ietf.org>; Wed, 30 Jan 2002 13:52:47 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id NAA15209
	for tsvwg-archive@odin.ietf.org; Wed, 30 Jan 2002 13:52:49 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id NAA14456;
	Wed, 30 Jan 2002 13:32:30 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id NAA14427
	for <tsvwg@optimus.ietf.org>; Wed, 30 Jan 2002 13:32:28 -0500 (EST)
Received: from mx10.radisys.com (mx10.radisys.com [206.102.10.36])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id NAA28360
	for <tsvwg@ietf.org>; Wed, 30 Jan 2002 13:32:26 -0500 (EST)
From: Heinz.Prantner@radisys.com
Received: by mx10.radisys.com (Postfix, from userid 5)
	id 0A04B142F0E; Wed, 30 Jan 2002 10:31:55 -0800 (PST)
Received: from UNKNOWN(206.103.52.220), claiming to be "notes.radisys.com"
 via SMTP by mx10, id smtpdAAAum3SN_; Wed Jan 30 10:31:46 2002
To: tsvwg@ietf.org
Cc: Dave.Sclarsky@radisys.com, Simon.Zhuang@radisys.com,
        OrgSPDCSCTP@RADISYS_CORPORATION.radisys.com
Subject: Re: [Tsvwg] draft-ietf-tsvwg-sctpimpguide-03.txt: Questions
MIME-Version: 1.0
X-Mailer: Lotus Notes Release 5.0.8  June 18, 2001
Message-ID: <OF819184A2.ED43B41A-ON85256B51.005EAD38@radisys.com>
Date: Wed, 30 Jan 2002 13:21:18 -0500
X-MIMETrack: Serialize by Router on HQ_ACS_1/Radisys_Corporation/US(Release 5.0.8 |June
 18, 2001) at 01/30/2002 10:26:37 AM,
	Serialize complete at 01/30/2002 10:26:37 AM
Content-Type: multipart/alternative; boundary="=_alternative 0065D10B85256B51_="
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

This is a multipart message in MIME format.
--=_alternative 0065D10B85256B51_=
Content-Type: text/plain; charset="us-ascii"

Hello Sctp,

In sctpimpguide chapter *2.8 Issues with Fast Retransmit* in *New text: 
(Section 7.2.4)* paragraph 5) it reads:

"...However,  as they are marked for retransmission they will be 
retransmitted
      later on as soon as cwnd allows."

This refers to the TSN's marked for fast retransmission, but could not 
been fast retransmitted due to 
restriction in paragraph 3 (only one single packet allowed).

Question: Do we wait for T3 for retransmission, or do we *delay* the fast 
retransmission until consecutive SACKs
will open the CWND. We feel that the latter is applicable and the impguide 
needs clarification. In our implementation,
we do not reset the TSN.Missing.Report counter, if we can not FR 
immediatly, the next SACK which opens CWND,
will kick off the *delayed fast retransmission* for those TSNs.

paragraph 3) says, the cwnd SHOULD be ignored for fast retransmission. 
What about rwnd? If we receive a SACK
which kicks off fast retransmission, where the SACK advertises rwnd=0, we 
are probably blocked. The impguide
says 1 single packet according to PMTU is allowed, cwnd SHOULD be ignored, 
but does not mention rwnd.

The impguide defines max burst rules for FR. Why is cwnd/rwnd not 
sufficient? What makes a SACK acknowledging a FR different to apply
that rule?

BTW: we ran into the same problem addressed in *2.8 Issues with Fast 
Retransmit*, having high traffic, delays and losses.
 We realized, that one packet loss will cause the CWND to crumble, due to 
repeated FR's.
 We *fixed* the problem (before reading the implguide) by saving the 
number of packets in flight for that TSN
at the first FR, and decrementing the outstanding packets counter (for 
that TSN) at receiption
of each consecutive SACK. While the counter is greater zero, we did not 
apply another FR. I have to confess though, that keeping
track of the number of packets in flight is not an easy task.

Another question regarding RFC2960:

where it says:
C5) Karn's algorithm: RTT measurements MUST NOT be made using packets
       that were retransmitted (and thus for which it is ambiguous
       whether the reply was for the first instance of the packet or a
       later instance).

In case we have continous retransmissions we do not update the RTT 
measurement.
Also, above rule implies, that RTT measuremt (for DATA) is performed only 
on the *current* destination, where
current is the primary, if active. The RtoPending flag is per destination, 
where the SACK __MAY__ come
back on a different path. Shouldnt the RtoPending flag relate to the 
assoc?


Heinz Prantner

--=_alternative 0065D10B85256B51_=
Content-Type: text/html; charset="us-ascii"


<br><font size=2 face="sans-serif">Hello Sctp,</font>
<br>
<br><font size=2 face="sans-serif">In sctpimpguide chapter *2.8 Issues with Fast Retransmit* in *New text: (Section 7.2.4)* paragraph 5) it reads:</font>
<br>
<br><font size=2 face="sans-serif">&quot;...However, &nbsp;as they are marked for retransmission they will be retransmitted</font>
<br><font size=2 face="sans-serif">&nbsp; &nbsp; &nbsp; later on as soon as cwnd allows.&quot;</font>
<br>
<br><font size=2 face="sans-serif">This refers to the TSN's marked for fast retransmission, but could not been fast retransmitted due to </font>
<br><font size=2 face="sans-serif">restriction in paragraph 3 (only one single packet allowed).</font>
<br>
<br><font size=2 face="sans-serif">Question: Do we wait for T3 for retransmission, or do we *delay* the fast retransmission until consecutive SACKs</font>
<br><font size=2 face="sans-serif">will open the CWND. We feel that the latter is applicable and the impguide needs clarification. In our implementation,</font>
<br><font size=2 face="sans-serif">we do not reset the TSN.Missing.Report counter, if we can not FR immediatly, the next SACK which opens CWND,</font>
<br><font size=2 face="sans-serif">will kick off the *delayed fast retransmission* for those TSNs.</font>
<br>
<br><font size=2 face="sans-serif">paragraph 3) says, the cwnd SHOULD be ignored for fast retransmission. What about rwnd? If we receive a SACK</font>
<br><font size=2 face="sans-serif">which kicks off fast retransmission, where the SACK advertises rwnd=0, we are probably blocked. The impguide</font>
<br><font size=2 face="sans-serif">says 1 single packet according to PMTU is allowed, cwnd SHOULD be ignored, but does not mention rwnd.</font>
<br>
<br><font size=2 face="sans-serif">The impguide defines max burst rules for FR. Why is cwnd/rwnd not sufficient? What makes a SACK acknowledging a FR different to apply</font>
<br><font size=2 face="sans-serif">that rule?</font>
<br>
<br><font size=2 face="sans-serif">BTW: we ran into the same problem addressed in *2.8 Issues with Fast Retransmit*, having high traffic, delays and losses.</font>
<br><font size=2 face="sans-serif">&nbsp;We realized, that one packet loss will cause the CWND to crumble, due to repeated FR's.</font>
<br><font size=2 face="sans-serif">&nbsp;We *fixed* the problem (before reading the implguide) by saving the number of packets in flight for that TSN</font>
<br><font size=2 face="sans-serif">at the first FR, and decrementing the outstanding packets counter (for that TSN) at receiption</font>
<br><font size=2 face="sans-serif">of each consecutive SACK. While the counter is greater zero, we did not apply another FR. I have to confess though, that keeping</font>
<br><font size=2 face="sans-serif">track of the number of packets in flight is not an easy task.</font>
<br>
<br><font size=2 face="sans-serif">Another question regarding RFC2960:</font>
<br>
<br><font size=2 face="sans-serif">where it says:</font>
<br><font size=2 face="sans-serif">C5) Karn's algorithm: RTT measurements MUST NOT be made using packets</font>
<br><font size=2 face="sans-serif">&nbsp; &nbsp; &nbsp; &nbsp;that were retransmitted (and thus for which it is ambiguous</font>
<br><font size=2 face="sans-serif">&nbsp; &nbsp; &nbsp; &nbsp;whether the reply was for the first instance of the packet or a</font>
<br><font size=2 face="sans-serif">&nbsp; &nbsp; &nbsp; &nbsp;later instance).</font>
<br>
<br><font size=2 face="sans-serif">In case we have continous retransmissions we do not update the RTT measurement.</font>
<br><font size=2 face="sans-serif">Also, above rule implies, that RTT measuremt (for DATA) is performed only on the *current* destination, where</font>
<br><font size=2 face="sans-serif">current is the primary, if active. The RtoPending flag is per destination, where the SACK __MAY__ come</font>
<br><font size=2 face="sans-serif">back on a different path. Shouldnt the RtoPending flag relate to the assoc?</font>
<br>
<br>
<br><font size=2 face="sans-serif">Heinz Prantner</font>
<br>
--=_alternative 0065D10B85256B51_=--

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Wed Jan 30 14:42:53 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id OAA00525
	for <tsvwg-archive@odin.ietf.org>; Wed, 30 Jan 2002 14:42:53 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id OAA19112
	for tsvwg-archive@odin.ietf.org; Wed, 30 Jan 2002 14:42:55 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id OAA17130;
	Wed, 30 Jan 2002 14:10:24 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id OAA17099
	for <tsvwg@optimus.ietf.org>; Wed, 30 Jan 2002 14:10:22 -0500 (EST)
Received: from mgw-x3.nokia.com (mgw-x3.nokia.com [131.228.20.26])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id OAA29681
	for <tsvwg@ietf.org>; Wed, 30 Jan 2002 14:10:19 -0500 (EST)
From: Ivan.Arias-Rodriguez@nokia.com
Received: from esvir01nok.ntc.nokia.com (esvir01nokt.ntc.nokia.com [172.21.143.33])
	by mgw-x3.nokia.com (Switch-2.1.0/Switch-2.1.0) with ESMTP id g0UJAVi16143
	for <tsvwg@ietf.org>; Wed, 30 Jan 2002 21:10:32 +0200 (EET)
Received: from esebh002.NOE.Nokia.com (unverified) by esvir01nok.ntc.nokia.com
 (Content Technologies SMTPRS 4.2.5) with ESMTP id <T58c5a47d96ac158f21081@esvir01nok.ntc.nokia.com>;
 Wed, 30 Jan 2002 21:10:13 +0200
Received: from esebe017.NOE.Nokia.com ([172.21.138.56]) by esebh002.NOE.Nokia.com with Microsoft SMTPSVC(5.0.2195.3779);
	 Wed, 30 Jan 2002 21:10:15 +0200
content-class: urn:content-classes:message
MIME-Version: 1.0
Subject: RE: [Tsvwg] draft-ietf-tsvwg-sctpimpguide-03.txt: Questions
Content-Type: text/plain;
	charset="iso-8859-1"
X-MimeOLE: Produced By Microsoft Exchange V6.0.5762.3
Date: Wed, 30 Jan 2002 21:10:15 +0200
Message-ID: <58AEDE27D48F884285007ABF827FF26308EE5A@esebe017.NOE.Nokia.com>
Thread-Topic: [Tsvwg] draft-ietf-tsvwg-sctpimpguide-03.txt: Questions
Thread-Index: AcGpvSjKKpeCjQCvRLO9AoS3C1B+IQAABgrg
To: <Heinz.Prantner@radisys.com>, <tsvwg@ietf.org>
Cc: <Dave.Sclarsky@radisys.com>, <Simon.Zhuang@radisys.com>,
        <OrgSPDCSCTP@RADISYS_CORPORATION.radisys.com>
X-OriginalArrivalTime: 30 Jan 2002 19:10:15.0739 (UTC) FILETIME=[BFF548B0:01C1A9C1]
Content-Transfer-Encoding: 8bit
X-MIME-Autoconverted: from quoted-printable to 8bit by optimus.ietf.org id OAA17100
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 8bit

	Hi!

	See some comments inline...

-----Original Message-----
From: ext Heinz.Prantner@radisys.com [mailto:Heinz.Prantner@radisys.com]
Sent: 30. January 2002 20:21
To: tsvwg@ietf.org
Cc: Dave.Sclarsky@radisys.com; Simon.Zhuang@radisys.com; OrgSPDCSCTP@RADISYS_CORPORATION.radisys.com
Subject: Re: [Tsvwg] draft-ietf-tsvwg-sctpimpguide-03.txt: Questions



Hello Sctp, 

In sctpimpguide chapter *2.8 Issues with Fast Retransmit* in *New text: (Section 7.2.4)* paragraph 5) it reads: 

"...However,  as they are marked for retransmission they will be retransmitted 
      later on as soon as cwnd allows." 

This refers to the TSN's marked for fast retransmission, but could not been fast retransmitted due to 
restriction in paragraph 3 (only one single packet allowed). 


---------------------
	Exactly.
---------------------


Question: Do we wait for T3 for retransmission, or do we *delay* the fast retransmission until consecutive SACKs 
will open the CWND. We feel that the latter is applicable and the impguide needs clarification. In our implementation, 


---------------------
	The idea is the second one. Take into account that once you have marked one packet for retransmission, it is not supposed to be on flight any more, and thus no T3 timer should be running for that TSN. The problem with a Fast retransmission is that it shrinks the cwnd. Then, if several TSNs must be fast retransmitted and their total size is larger than one MTU, it is highly probable that you won't be able to send all of them...

	When the cwnd grows later on thanks to the received SACK chunks, then you will be able to retransmit it...
---------------------


we do not reset the TSN.Missing.Report counter, if we can not FR immediatly, the next SACK which opens CWND, 
will kick off the *delayed fast retransmission* for those TSNs. 


---------------------
	Uhm... :-/

	Do you mean that if the next SACK does not acknowledge those TSNs not already fast retransmitted, you make another fast retransmission?

	But the problem with this is that you will receive many SACKs. They are on the wire comming back... If you do that, you will apply the Fast Retransmission algorithm several times, and you will lower the cwnd a lot. If you make a fast retransmission you will normaly have suddenly more outstanding data than the value of cwnd. Then you will have to wait until cwnd grows (and the number of MTUs on flight lowers) before you can send anything (the chunks marked for retransmission).
---------------------


paragraph 3) says, the cwnd SHOULD be ignored for fast retransmission. What about rwnd? If we receive a SACK 


---------------------
	Well, if you have to retransmit something, the rwnd will never limit you, as you consider those TSNs to be retransmitted as not being on flight any more, and thus they do not occupy any space at the receiver... You decrease rwnd by the size of all the TSNs to be fast retransmitted, and then you add only some of them... you can't have more than before (though some politicians would say that it is possible :->)
---------------------


which kicks off fast retransmission, where the SACK advertises rwnd=0, we are probably blocked. The impguide 


---------------------
	Uhm... I get your point... So the receiver is effectively shrinking its buffers... bad buy... :-0

	Maybe it could be mentioned...
---------------------


says 1 single packet according to PMTU is allowed, cwnd SHOULD be ignored, but does not mention rwnd. 


---------------------
	Yeah... You probably have already that packet on flight...
---------------------


The impguide defines max burst rules for FR. Why is cwnd/rwnd not sufficient? What makes a SACK acknowledging a FR different to apply 
that rule? 


---------------------
	Imagine that if there is one TSN missing and the subsequent TSNs have already arrive. In the normal case, when the TSNs are delivered in order to the upper user, then the SCTP receiver will have part of its buffer full due to all those received TSNs that can't be delivered. So, once the missing TSN arrives, all those TSNs will be delivered to the upper user and rwnd will grow considerably. If rwnd was limiting the sender, and not cwnd, suddenly the data sender could be allowed to send a lot of TSNs...

	That's what Max.Burst is for...
---------------------


BTW: we ran into the same problem addressed in *2.8 Issues with Fast Retransmit*, having high traffic, delays and losses. 
 We realized, that one packet loss will cause the CWND to crumble, due to repeated FR's. 
 We *fixed* the problem (before reading the implguide) by saving the number of packets in flight for that TSN 
at the first FR, and decrementing the outstanding packets counter (for that TSN) at receiption 
of each consecutive SACK. While the counter is greater zero, we did not apply another FR. I have to confess though, that keeping 
track of the number of packets in flight is not an easy task. 

Another question regarding RFC2960: 

where it says: 
C5) Karn's algorithm: RTT measurements MUST NOT be made using packets 
       that were retransmitted (and thus for which it is ambiguous 
       whether the reply was for the first instance of the packet or a 
       later instance). 

In case we have continous retransmissions we do not update the RTT measurement. 


---------------------
	In any case, after a retransmission you duplicate RTO, and thus the main use of RTT is somehow lost... :-/
---------------------

Also, above rule implies, that RTT measuremt (for DATA) is performed only on the *current* destination, where 
current is the primary, if active. The RtoPending flag is per destination, where the SACK __MAY__ come 
back on a different path. Shouldnt the RtoPending flag relate to the assoc? 


---------------------
	I'm not sure if I understand your question... :-/

	But what you test is the RTT of a specific address. The SACK are supposed to be sent to the source address of the DATA that originates it (note that due to the delayed SACK algorithm this is not always possible). If the SACK comes to another different address, then the RTT is not that exact, but this should be a corner case.

	In any case, you measure the RTT of the rest of addresses (not the primary one) using the heartbeats... At the end, you are measuring the RTT of the primary address once every RTO, and you measure the RTT of the other addresses once every heartbeating interval (only *one* of those addresses at a time)...

	I hope these comments are of any help...

	Bye!

	BR Iv�n Arias Rodr�guez

---------------------


Heinz Prantner 

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Wed Jan 30 16:44:30 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id QAA02887
	for <tsvwg-archive@odin.ietf.org>; Wed, 30 Jan 2002 16:44:30 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id QAA24923
	for tsvwg-archive@odin.ietf.org; Wed, 30 Jan 2002 16:44:33 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id QAA23681;
	Wed, 30 Jan 2002 16:19:07 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id QAA23614
	for <tsvwg@optimus.ietf.org>; Wed, 30 Jan 2002 16:19:04 -0500 (EST)
Received: from company.mail (host096103.arnet.net.ar [200.45.96.103] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id QAA02363
	for <tsvwg@ietf.org>; Wed, 30 Jan 2002 16:18:50 -0500 (EST)
Message-Id: <200201302118.QAA02363@ietf.org>
Received: from localhost [127.0.0.1]
	by company.mail [169.254.120.177]
	with SMTP (MDaemon.PRO.v5.0.1.R)
	for <tsvwg@ietf.org>; Wed, 30 Jan 2002 13:46:47 -0300
X-Sender: noreply@server.com
From: DAP Consulting <noreply@server.com>
To: tsvwg@ietf.org
Date: Wed, 30 Jan 2002 13:46:45 -0300
Reply-To: noreply@server.com
MIME-Version: 1.0
Content-Type: multipart/mixed;
	boundary="----=_NextPart_001__8923473_49605,9"
X-Priority: 1
X-MDRemoteIP: 127.0.0.1
X-Return-Path: noreply@server.com
X-MDaemon-Deliver-To: tsvwg@ietf.org
Subject: [Tsvwg] =?iso-8859-1?Q?Propuesta_de_Consultor=EDa?=
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

This is a Multipart MIME message.

------=_NextPart_001__8923473_49605,9
Content-Type: text/plain; charset=iso-8859-1
Content-Transfer-Encoding: 8bit

Buenos Aires, enero de 2002.- 

Atenci�n Sr. Gerente General 

Molestamos su atenci�n a efectos de adjuntarle nuestra propuesta de servicios 
de consultor�a en planeamiento estrat�gico, la cual en momentos como los actuales, 
creemos que ser� de su inter�s. 

Estamos informandole que proximamente le enviarios nuestro bolet�n "Reflexiones", 
en el cual abordamos cuestiones de inter�s estrat�gico y econ�mico, para la alta 
gerencia empresaria. 

Le saludo atentamente 

Dr. Obdulio Dur�n 

Para contactarse con nosotros por favor NO RESPONDA A ESTE MENSAJE, utilice la 
direcci�n dap_consulting@yahoo.com o ll�menos al 15-4414-2008


Por secci�n 301, p�rrafo (a)(2)(C) de S.1618. Bajo el decreto S.1618 
titulo 3ro. Aprobado por el 105 Congreso Base de las Normativas 
Internacionales sobre SPAM, un E-mail no podr� ser considerado SPAM 
mientras incluya una forma de ser removido.Si usted desea ser removido de 
nuestra base de datos en forma definitiva por favor env�e un e-mail a la direcci�n 
dap_consulting@yahoo.com indicando "Remover" en el subject.






------=_NextPart_001__8923473_49605,9
Content-Type: application/octet-stream
Content-Disposition: attachment; filename="resumen de propuesta.doc"
Content-Length: 39994
Content-Transfer-Encoding: base64
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------=_NextPart_001__8923473_49605,9--



_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Wed Jan 30 17:20:07 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id RAA03601
	for <tsvwg-archive@odin.ietf.org>; Wed, 30 Jan 2002 17:20:07 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id RAA26535
	for tsvwg-archive@odin.ietf.org; Wed, 30 Jan 2002 17:20:10 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id QAA25286;
	Wed, 30 Jan 2002 16:59:46 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id QAA25256
	for <tsvwg@optimus.ietf.org>; Wed, 30 Jan 2002 16:59:44 -0500 (EST)
Received: from mx10.radisys.com (mx10.radisys.com [206.102.10.36])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id QAA03206
	for <tsvwg@ietf.org>; Wed, 30 Jan 2002 16:59:40 -0500 (EST)
From: Heinz.Prantner@radisys.com
Received: by mx10.radisys.com (Postfix, from userid 5)
	id C8172142E2A; Wed, 30 Jan 2002 13:59:11 -0800 (PST)
Received: from UNKNOWN(206.103.52.220), claiming to be "notes.radisys.com"
 via SMTP by mx10, id smtpdAAA0JR88_; Wed Jan 30 13:59:03 2002
To: <Ivan.Arias-Rodriguez@nokia.com>, tsvwg@ietf.org
Cc: Dave.Sclarsky@radisys.com, OrgSPDCSCTP@RADISYS_CORPORATION.radisys.com,
        Simon.Zhuang@radisys.com
Subject: RE: [Tsvwg] draft-ietf-tsvwg-sctpimpguide-03.txt: Questions
MIME-Version: 1.0
X-Mailer: Lotus Notes Release 5.0.8  June 18, 2001
Message-ID: <OFBA4BD893.9EA6886C-ON85256B51.0072804D@radisys.com>
Date: Wed, 30 Jan 2002 16:48:37 -0500
X-MIMETrack: Serialize by Router on HQ_ACS_1/Radisys_Corporation/US(Release 5.0.8 |June
 18, 2001) at 01/30/2002 01:53:54 PM,
	Serialize complete at 01/30/2002 01:53:54 PM
Content-Type: multipart/alternative; boundary="=_alternative 0078CBDD85256B51_="
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

This is a multipart message in MIME format.
--=_alternative 0078CBDD85256B51_=
Content-Type: text/plain; charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable

Hi Ivan,

thank you for your explanations, yes, your comments help :-)
please see my few more comments inline...

Heinz





<Ivan.Arias-Rodriguez@nokia.com>
01/30/2002 02:10 PM

=20
        To:     <Heinz.Prantner@radisys.com>, <tsvwg@ietf.org>
        cc:     <Dave.Sclarsky@radisys.com>, <Simon.Zhuang@radisys.com>,=20
<OrgSPDCSCTP@RADISYS=5FCORPORATION.radisys.com>
        Subject:        RE: [Tsvwg] draft-ietf-tsvwg-sctpimpguide-03.txt: Q=
uestions


                 Hi!

                 See some comments inline...

-----Original Message-----
From: ext Heinz.Prantner@radisys.com [mailto:Heinz.Prantner@radisys.com]
Sent: 30. January 2002 20:21
To: tsvwg@ietf.org
Cc: Dave.Sclarsky@radisys.com; Simon.Zhuang@radisys.com;=20
OrgSPDCSCTP@RADISYS=5FCORPORATION.radisys.com
Subject: Re: [Tsvwg] draft-ietf-tsvwg-sctpimpguide-03.txt: Questions



Hello Sctp,=20

In sctpimpguide chapter *2.8 Issues with Fast Retransmit* in *New text:=20
(Section 7.2.4)* paragraph 5) it reads:=20

"...However,  as they are marked for retransmission they will be=20
retransmitted=20
      later on as soon as cwnd allows."=20

This refers to the TSN's marked for fast retransmission, but could not=20
been fast retransmitted due to=20
restriction in paragraph 3 (only one single packet allowed).=20


---------------------
                 Exactly.
---------------------


Question: Do we wait for T3 for retransmission, or do we *delay* the fast=20
retransmission until consecutive SACKs=20
will open the CWND. We feel that the latter is applicable and the impguide =

needs clarification. In our implementation,=20


---------------------
                 The idea is the second one. Take into account that once=20
you have marked one packet for retransmission, it is not supposed to be on =

flight any more, and thus no T3 timer should be running for that TSN. The=20
problem with a Fast retransmission is that it shrinks the cwnd. Then, if=20
several TSNs must be fast retransmitted and their total size is larger=20
than one MTU, it is highly probable that you won't be able to send all of=20
them...

                 When the cwnd grows later on thanks to the received SACK=20
chunks, then you will be able to retransmit it...
---------------------


we do not reset the TSN.Missing.Report counter, if we can not FR=20
immediatly, the next SACK which opens CWND,=20
will kick off the *delayed fast retransmission* for those TSNs.=20


---------------------
                 Uhm... :-/

                 Do you mean that if the next SACK does not acknowledge=20
those TSNs not already fast retransmitted, you make another fast=20
retransmission?

                 But the problem with this is that you will receive many=20
SACKs. They are on the wire comming back... If you do that, you will apply =

the Fast Retransmission algorithm several times, and you will lower the=20
cwnd a lot. If you make a fast retransmission you will normaly have=20
suddenly more outstanding data than the value of cwnd. Then you will have=20
to wait until cwnd grows (and the number of MTUs on flight lowers) before=20
you can send anything (the chunks marked for retransmission).

[HP] I was not precise: we basically do what you describe above in *The=20
idea is the second one*. We do not apply FR algorithm several times.=20
It would shrink the cwnd a lot, as you say.
---------------------


paragraph 3) says, the cwnd SHOULD be ignored for fast retransmission.=20
What about rwnd? If we receive a SACK=20


---------------------
                 Well, if you have to retransmit something, the rwnd will=20
never limit you, as you consider those TSNs to be retransmitted as not=20
being on flight any more, and thus they do not occupy any space at the=20
receiver... You decrease rwnd by the size of all the TSNs to be fast=20
retransmitted, and then you add only some of them... you can't have more=20
than before (though some politicians would say that it is possible :->)

[HP] "...have to retransmit something, the rwnd will never limit you...",=20
can we have that in the impguide?
The RFC says the opposite:

6.1  Transmission of DATA Chunks

...

The following general rules MUST be applied by the data sender for
   transmission and/or retransmission of outbound DATA chunks:
                ^^^^^^^^^^^^^^^^^^^^^
=20
   A) At any given time, the data sender MUST NOT transmit new data to
      any destination transport address if its peer's rwnd indicates
      that the peer has no buffer space=20

---------------------


which kicks off fast retransmission, where the SACK advertises rwnd=3D0, we=
=20
are probably blocked. The impguide=20


---------------------
                 Uhm... I get your point... So the receiver is effectively =

shrinking its buffers... bad buy... :-0

                 Maybe it could be mentioned...

[HP] according to above statement, no rwnd limitation at retrans, it will=20
work ... and might be the only way to help the receiver
 (the one who advertised rwnd=3D0) to free his buffers, since he needs that=
=20
TSN, otherwise he is blocked.
---------------------


says 1 single packet according to PMTU is allowed, cwnd SHOULD be ignored, =

but does not mention rwnd.=20


---------------------
                 Yeah... You probably have already that packet on=20
flight...
---------------------


The impguide defines max burst rules for FR. Why is cwnd/rwnd not=20
sufficient? What makes a SACK acknowledging a FR different to apply=20
that rule?=20


---------------------
                 Imagine that if there is one TSN missing and the=20
subsequent TSNs have already arrive. In the normal case, when the TSNs are =

delivered in order to the upper user, then the SCTP receiver will have=20
part of its buffer full due to all those received TSNs that can't be=20
delivered. So, once the missing TSN arrives, all those TSNs will be=20
delivered to the upper user and rwnd will grow considerably. If rwnd was=20
limiting the sender, and not cwnd, suddenly the data sender could be=20
allowed to send a lot of TSNs...

                 That's what Max.Burst is for...
---------------------


BTW: we ran into the same problem addressed in *2.8 Issues with Fast=20
Retransmit*, having high traffic, delays and losses.=20
 We realized, that one packet loss will cause the CWND to crumble, due to=20
repeated FR's.=20
 We *fixed* the problem (before reading the implguide) by saving the=20
number of packets in flight for that TSN=20
at the first FR, and decrementing the outstanding packets counter (for=20
that TSN) at receiption=20
of each consecutive SACK. While the counter is greater zero, we did not=20
apply another FR. I have to confess though, that keeping=20
track of the number of packets in flight is not an easy task.=20

Another question regarding RFC2960:=20

where it says:=20
C5) Karn's algorithm: RTT measurements MUST NOT be made using packets=20
       that were retransmitted (and thus for which it is ambiguous=20
       whether the reply was for the first instance of the packet or a=20
       later instance).=20

In case we have continous retransmissions we do not update the RTT=20
measurement.=20


---------------------
                 In any case, after a retransmission you duplicate RTO,=20
and thus the main use of RTT is somehow lost... :-/

[HP] I'm sending on primary destination, and assuming continuous fast=20
retransmissions on the alternate destination. (FR does not update RTO,=20
right?, and if not FR, T3 expired for the primary, ie. duplicate the RTO=20
on primary). Now the alternate destination is busily used for FR, thus no=20
heartbeats are sent, so no RTT measurement is done. But maybe thats ok...
---------------------

Also, above rule implies, that RTT measuremt (for DATA) is performed only=20
on the *current* destination, where=20
current is the primary, if active. The RtoPending flag is per destination, =

where the SACK =5F=5FMAY=5F=5F come=20
back on a different path. Shouldnt the RtoPending flag relate to the=20
assoc?=20


---------------------
                 I'm not sure if I understand your question... :-/

                 But what you test is the RTT of a specific address. The=20
SACK are supposed to be sent to the source address of the DATA that=20
originates it (note that due to the delayed SACK algorithm this is not=20
always possible). If the SACK comes to another different address, then the =

RTT is not that exact, but this should be a corner case.

                 In any case, you measure the RTT of the rest of addresses =

(not the primary one) using the heartbeats... At the end, you are=20
measuring the RTT of the primary address once every RTO, and you measure=20
the RTT of the other addresses once every heartbeating interval (only=20
*one* of those addresses at a time)...

                 I hope these comments are of any help...

                 Bye!

                 BR Iv=E1n Arias Rodr=EDguez

---------------------


Heinz Prantner=20



--=_alternative 0078CBDD85256B51_=
Content-Type: text/html; charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable


<br><font size=3D2 face=3D"sans-serif">Hi Ivan,</font>
<br>
<br><font size=3D2 face=3D"sans-serif">thank you for your explanations, yes=
, your comments help :-)</font>
<br><font size=3D2 face=3D"sans-serif">please see my few more comments inli=
ne...</font>
<br>
<br><font size=3D2 face=3D"sans-serif">Heinz</font>
<br>
<br>
<br>
<br>
<table width=3D100%>
<tr valign=3Dtop>
<td>
<td><font size=3D1 face=3D"sans-serif"><b>&lt;Ivan.Arias-Rodriguez@nokia.co=
m&gt;</b></font>
<p><font size=3D1 face=3D"sans-serif">01/30/2002 02:10 PM</font>
<br>
<td><font size=3D1 face=3D"Arial">&nbsp; &nbsp; &nbsp; &nbsp; </font>
<br><font size=3D1 face=3D"sans-serif">&nbsp; &nbsp; &nbsp; &nbsp; To: &nbs=
p; &nbsp; &nbsp; &nbsp;&lt;Heinz.Prantner@radisys.com&gt;, &lt;tsvwg@ietf.o=
rg&gt;</font>
<br><font size=3D1 face=3D"sans-serif">&nbsp; &nbsp; &nbsp; &nbsp; cc: &nbs=
p; &nbsp; &nbsp; &nbsp;&lt;Dave.Sclarsky@radisys.com&gt;, &lt;Simon.Zhuang@=
radisys.com&gt;, &lt;OrgSPDCSCTP@RADISYS=5FCORPORATION.radisys.com&gt;</fon=
t>
<br><font size=3D1 face=3D"sans-serif">&nbsp; &nbsp; &nbsp; &nbsp; Subject:=
 &nbsp; &nbsp; &nbsp; &nbsp;RE: [Tsvwg] draft-ietf-tsvwg-sctpimpguide-03.tx=
t: Questions</font></table>
<br>
<br>
<br><font size=3D2 face=3D"Courier New">&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; =
&nbsp; &nbsp; &nbsp; &nbsp;Hi!<br>
<br>
 &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; See some comments =
inline...<br>
<br>
-----Original Message-----<br>
From: ext Heinz.Prantner@radisys.com [mailto:Heinz.Prantner@radisys.com]<br>
Sent: 30. January 2002 20:21<br>
To: tsvwg@ietf.org<br>
Cc: Dave.Sclarsky@radisys.com; Simon.Zhuang@radisys.com; OrgSPDCSCTP@RADISY=
S=5FCORPORATION.radisys.com<br>
Subject: Re: [Tsvwg] draft-ietf-tsvwg-sctpimpguide-03.txt: Questions<br>
<br>
<br>
<br>
Hello Sctp, <br>
<br>
In sctpimpguide chapter *2.8 Issues with Fast Retransmit* in *New text: (Se=
ction 7.2.4)* paragraph 5) it reads: <br>
<br>
&quot;...However, &nbsp;as they are marked for retransmission they will be =
retransmitted <br>
 &nbsp; &nbsp; &nbsp;later on as soon as cwnd allows.&quot; <br>
<br>
This refers to the TSN's marked for fast retransmission, but could not been=
 fast retransmitted due to <br>
restriction in paragraph 3 (only one single packet allowed). <br>
<br>
<br>
---------------------<br>
 &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; Exactly.<br>
---------------------<br>
<br>
<br>
Question: Do we wait for T3 for retransmission, or do we *delay* the fast r=
etransmission until consecutive SACKs <br>
will open the CWND. We feel that the latter is applicable and the impguide =
needs clarification. In our implementation, <br>
<br>
<br>
---------------------<br>
 &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; The idea is the se=
cond one. Take into account that once you have marked one packet for retran=
smission, it is not supposed to be on flight any more, and thus no T3 timer=
 should be running for that TSN. The problem with a Fast retransmission is =
that it shrinks the cwnd. Then, if several TSNs must be fast retransmitted =
and their total size is larger than one MTU, it is highly probable that you=
 won't be able to send all of them...<br>
<br>
 &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; When the cwnd grow=
s later on thanks to the received SACK chunks, then you will be able to ret=
ransmit it...<br>
---------------------<br>
<br>
<br>
we do not reset the TSN.Missing.Report counter, if we can not FR immediatly=
, the next SACK which opens CWND, <br>
will kick off the *delayed fast retransmission* for those TSNs. <br>
<br>
<br>
---------------------<br>
 &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; Uhm... :-/<br>
<br>
 &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; Do you mean that i=
f the next SACK does not acknowledge those TSNs not already fast retransmit=
ted, you make another fast retransmission?<br>
<br>
 &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; But the problem wi=
th this is that you will receive many SACKs. They are on the wire comming b=
ack... If you do that, you will apply the Fast Retransmission algorithm sev=
eral times, and you will lower the cwnd a lot. If you make a fast retransmi=
ssion you will normaly have suddenly more outstanding data than the value o=
f cwnd. Then you will have to wait until cwnd grows (and the number of MTUs=
 on flight lowers) before you can send anything (the chunks marked for retr=
ansmission).</font>
<br>
<br><font size=3D2 face=3D"Courier New">[HP] I was not precise: we basicall=
y do what you describe above in *The idea is the second one*. We do not app=
ly FR algorithm several times. </font>
<br><font size=3D2 face=3D"Courier New">It would shrink the cwnd a lot, as =
you say.<br>
---------------------<br>
<br>
<br>
paragraph 3) says, the cwnd SHOULD be ignored for fast retransmission. What=
 about rwnd? If we receive a SACK <br>
<br>
<br>
---------------------<br>
 &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; Well, if you have =
to retransmit something, the rwnd will never limit you, as you consider tho=
se TSNs to be retransmitted as not being on flight any more, and thus they =
do not occupy any space at the receiver... You decrease rwnd by the size of=
 all the TSNs to be fast retransmitted, and then you add only some of them.=
.. you can't have more than before (though some politicians would say that =
it is possible :-&gt;)</font>
<br>
<br><font size=3D2 face=3D"Courier New">[HP] &quot;...have to retransmit so=
mething, the rwnd will never limit you...&quot;, can we have that in the im=
pguide?</font>
<br><font size=3D2 face=3D"Courier New">The RFC says the opposite:</font>
<br>
<br><font size=3D2><tt>6.1 &nbsp;Transmission of DATA Chunks</tt></font>
<br>
<br><font size=3D2 face=3D"Courier New">...</font>
<br>
<br><font size=3D2><tt>The following general rules MUST be applied by the d=
ata sender for</tt></font>
<br><font size=3D2><tt>&nbsp; &nbsp;transmission and/or retransmission of o=
utbound DATA chunks:</tt></font>
<br><font size=3D2><tt>&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nb=
sp; ^^^^^^^^^^^^^^^^^^^^^</tt></font>
<br><font size=3D2><tt>&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nb=
sp; </tt></font>
<br><font size=3D2><tt>&nbsp; &nbsp;A) At any given time, the data sender M=
UST NOT transmit new data to</tt></font>
<br><font size=3D2><tt>&nbsp; &nbsp; &nbsp; any destination transport addre=
ss if its peer's rwnd indicates</tt></font>
<br><font size=3D2><tt>&nbsp; &nbsp; &nbsp; that the peer has no buffer spa=
ce </tt></font>
<br><font size=3D2 face=3D"Courier New"><br>
---------------------<br>
<br>
<br>
which kicks off fast retransmission, where the SACK advertises rwnd=3D0, we=
 are probably blocked. The impguide <br>
<br>
<br>
---------------------<br>
 &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; Uhm... I get your =
point... So the receiver is effectively shrinking its buffers... bad buy...=
 :-0<br>
<br>
 &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; Maybe it could be =
mentioned...</font>
<br>
<br><font size=3D2 face=3D"Courier New">[HP] according to above statement, =
no rwnd limitation at retrans, it will work ... and might be the only way t=
o help the receiver</font>
<br><font size=3D2 face=3D"Courier New">&nbsp;(the one who advertised rwnd=
=3D0) to free his buffers, since he needs that TSN, otherwise he is blocked=
.<br>
---------------------<br>
<br>
<br>
says 1 single packet according to PMTU is allowed, cwnd SHOULD be ignored, =
but does not mention rwnd. <br>
<br>
<br>
---------------------<br>
 &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; Yeah... You probab=
ly have already that packet on flight...<br>
---------------------<br>
<br>
<br>
The impguide defines max burst rules for FR. Why is cwnd/rwnd not sufficien=
t? What makes a SACK acknowledging a FR different to apply <br>
that rule? <br>
<br>
<br>
---------------------<br>
 &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; Imagine that if th=
ere is one TSN missing and the subsequent TSNs have already arrive. In the =
normal case, when the TSNs are delivered in order to the upper user, then t=
he SCTP receiver will have part of its buffer full due to all those receive=
d TSNs that can't be delivered. So, once the missing TSN arrives, all those=
 TSNs will be delivered to the upper user and rwnd will grow considerably. =
If rwnd was limiting the sender, and not cwnd, suddenly the data sender cou=
ld be allowed to send a lot of TSNs...<br>
<br>
 &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; That's what Max.Bu=
rst is for...<br>
---------------------<br>
<br>
<br>
BTW: we ran into the same problem addressed in *2.8 Issues with Fast Retran=
smit*, having high traffic, delays and losses. <br>
 We realized, that one packet loss will cause the CWND to crumble, due to r=
epeated FR's. <br>
 We *fixed* the problem (before reading the implguide) by saving the number=
 of packets in flight for that TSN <br>
at the first FR, and decrementing the outstanding packets counter (for that=
 TSN) at receiption <br>
of each consecutive SACK. While the counter is greater zero, we did not app=
ly another FR. I have to confess though, that keeping <br>
track of the number of packets in flight is not an easy task. <br>
<br>
Another question regarding RFC2960: <br>
<br>
where it says: <br>
C5) Karn's algorithm: RTT measurements MUST NOT be made using packets <br>
 &nbsp; &nbsp; &nbsp; that were retransmitted (and thus for which it is amb=
iguous <br>
 &nbsp; &nbsp; &nbsp; whether the reply was for the first instance of the p=
acket or a <br>
 &nbsp; &nbsp; &nbsp; later instance). <br>
<br>
In case we have continous retransmissions we do not update the RTT measurem=
ent. <br>
<br>
<br>
---------------------<br>
 &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; In any case, after=
 a retransmission you duplicate RTO, and thus the main use of RTT is someho=
w lost... :-/</font>
<br>
<br><font size=3D2 face=3D"Courier New">[HP] I'm sending on primary destina=
tion, and assuming continuous fast retransmissions on the alternate destina=
tion. (FR does not update RTO, right?, and if not FR, T3 expired for the pr=
imary, ie. duplicate the RTO on primary). Now the alternate destination is =
busily used for FR, thus no heartbeats are sent, so no RTT measurement is d=
one. But maybe thats ok...<br>
---------------------<br>
<br>
Also, above rule implies, that RTT measuremt (for DATA) is performed only o=
n the *current* destination, where <br>
current is the primary, if active. The RtoPending flag is per destination, =
where the SACK =5F=5FMAY=5F=5F come </font>
<br><font size=3D2 face=3D"Courier New">back on a different path. Shouldnt =
the RtoPending flag relate to the assoc? <br>
<br>
<br>
---------------------<br>
 &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; I'm not sure if I =
understand your question... :-/<br>
<br>
 &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; But what you test =
is the RTT of a specific address. The SACK are supposed to be sent to the s=
ource address of the DATA that originates it (note that due to the delayed =
SACK algorithm this is not always possible). If the SACK comes to another d=
ifferent address, then the RTT is not that exact, but this should be a corn=
er case.<br>
<br>
 &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; In any case, you m=
easure the RTT of the rest of addresses (not the primary one) using the hea=
rtbeats... At the end, you are measuring the RTT of the primary address onc=
e every RTO, and you measure the RTT of the other addresses once every hear=
tbeating interval (only *one* of those addresses at a time)...<br>
<br>
 &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; I hope these comme=
nts are of any help...<br>
<br>
 &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; Bye!<br>
<br>
 &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; BR Iv=E1n Arias Ro=
dr=EDguez<br>
<br>
---------------------<br>
<br>
<br>
Heinz Prantner <br>
</font>
<br>
<br>
--=_alternative 0078CBDD85256B51_=--

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Wed Jan 30 18:11:37 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id SAA04528
	for <tsvwg-archive@odin.ietf.org>; Wed, 30 Jan 2002 18:11:37 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id SAA29182
	for tsvwg-archive@odin.ietf.org; Wed, 30 Jan 2002 18:11:38 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id RAA28238;
	Wed, 30 Jan 2002 17:57:52 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id RAA28209
	for <tsvwg@optimus.ietf.org>; Wed, 30 Jan 2002 17:57:50 -0500 (EST)
Received: from mx10.radisys.com (mx10.radisys.com [206.102.10.36])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id RAA04226
	for <tsvwg@ietf.org>; Wed, 30 Jan 2002 17:57:48 -0500 (EST)
From: Heinz.Prantner@radisys.com
Received: by mx10.radisys.com (Postfix, from userid 5)
	id 80817142E2A; Wed, 30 Jan 2002 14:57:20 -0800 (PST)
Received: from UNKNOWN(206.103.52.220), claiming to be "notes.radisys.com"
 via SMTP by mx10, id smtpdAAA0x.QuQ; Wed Jan 30 14:57:12 2002
To: <Ivan.Arias-Rodriguez@nokia.com>
Cc: Dave.Sclarsky@radisys.com, OrgSPDCSCTP@RADISYS_CORPORATION.radisys.com,
        Simon.Zhuang@radisys.com, tsvwg@ietf.org
Subject: RE: [Tsvwg] draft-ietf-tsvwg-sctpimpguide-03.txt: Questions
MIME-Version: 1.0
X-Mailer: Lotus Notes Release 5.0.8  June 18, 2001
Message-ID: <OF34E8A4E7.74390E4E-ON85256B51.007C8B77@radisys.com>
Date: Wed, 30 Jan 2002 17:46:46 -0500
X-MIMETrack: Serialize by Router on HQ_ACS_1/Radisys_Corporation/US(Release 5.0.8 |June
 18, 2001) at 01/30/2002 02:52:03 PM,
	Serialize complete at 01/30/2002 02:52:03 PM
Content-Type: multipart/alternative; boundary="=_alternative 007E1EB185256B51_="
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

This is a multipart message in MIME format.
--=_alternative 007E1EB185256B51_=
Content-Type: text/plain; charset="us-ascii"

<SNIP>
 
In any case I think that I don't get your point. Why is it a problem? If 
you are sending FR then you can't use then to measure RTT, i.e. you 
continue sending heartbeats (look at the definition of "idle address" in 
page 93 of RFC 2960).
 
<SNIP>

 
Uuups, I will then update my code to the correct IDLE definition. Thanks 
for the catch!

Heinz








--=_alternative 007E1EB185256B51_=
Content-Type: text/html; charset="us-ascii"


<br>
<br>
<br>
<br>
<br>
<br>
<br><font size=2 face="Courier New"><br>
<br>
&lt;SNIP&gt;<br>
 <br>
In any case I think that I don't get your point. Why is it a problem? If you are sending FR then you can't use then to measure RTT, i.e. you continue sending heartbeats (look at the definition of &quot;idle address&quot; in page 93 of RFC 2960).<br>
 <br>
&lt;SNIP&gt;</font>
<br>
<br><font size=2 face="Courier New">&nbsp;<br>
</font><font size=2 face="sans-serif">Uuups, I will then update my code to the correct IDLE definition. Thanks for the catch!</font>
<br>
<br><font size=2 face="sans-serif">Heinz</font>
<br><font size=2 face="Courier New"><br>
<br>
<br>
<br>
<br>
</font>
<br>
<br>
--=_alternative 007E1EB185256B51_=--

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Wed Jan 30 18:12:16 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id SAA04545
	for <tsvwg-archive@odin.ietf.org>; Wed, 30 Jan 2002 18:12:16 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id SAA29202
	for tsvwg-archive@odin.ietf.org; Wed, 30 Jan 2002 18:12:18 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id RAA26866;
	Wed, 30 Jan 2002 17:29:00 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id RAA26799
	for <tsvwg@optimus.ietf.org>; Wed, 30 Jan 2002 17:28:57 -0500 (EST)
Received: from mgw-x2.nokia.com (mgw-x2.nokia.com [131.228.20.22])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id RAA03739
	for <tsvwg@ietf.org>; Wed, 30 Jan 2002 17:28:52 -0500 (EST)
From: Ivan.Arias-Rodriguez@nokia.com
Received: from esvir04nok.ntc.nokia.com (esvir04nokt.ntc.nokia.com [172.21.143.36])
	by mgw-x2.nokia.com (Switch-2.1.0/Switch-2.1.0) with ESMTP id g0UMSwZ28196
	for <tsvwg@ietf.org>; Thu, 31 Jan 2002 00:28:58 +0200 (EET)
Received: from esebh004.NOE.Nokia.com (unverified) by esvir04nok.ntc.nokia.com
 (Content Technologies SMTPRS 4.2.5) with ESMTP id <T58c65a615bac158f24077@esvir04nok.ntc.nokia.com>;
 Thu, 31 Jan 2002 00:28:53 +0200
Received: from esebe017.NOE.Nokia.com ([172.21.138.56]) by esebh004.NOE.Nokia.com with Microsoft SMTPSVC(5.0.2195.3779);
	 Thu, 31 Jan 2002 00:28:54 +0200
content-class: urn:content-classes:message
MIME-Version: 1.0
Content-Type: text/plain;
	charset="iso-8859-1"
Subject: RE: [Tsvwg] draft-ietf-tsvwg-sctpimpguide-03.txt: Questions
X-MimeOLE: Produced By Microsoft Exchange V6.0.5762.3
Date: Thu, 31 Jan 2002 00:28:53 +0200
Message-ID: <58AEDE27D48F884285007ABF827FF26308EE5B@esebe017.NOE.Nokia.com>
Thread-Topic: [Tsvwg] draft-ietf-tsvwg-sctpimpguide-03.txt: Questions
Thread-Index: AcGp2Vw51gqxXU5RSkKaz1qQyMZeCAAAYl8A
To: <Heinz.Prantner@radisys.com>, <tsvwg@ietf.org>
Cc: <Dave.Sclarsky@radisys.com>, <OrgSPDCSCTP@RADISYS_CORPORATION.radisys.com>,
        <Simon.Zhuang@radisys.com>
X-OriginalArrivalTime: 30 Jan 2002 22:28:54.0210 (UTC) FILETIME=[7FEBA220:01C1A9DD]
Content-Transfer-Encoding: 8bit
X-MIME-Autoconverted: from quoted-printable to 8bit by optimus.ietf.org id RAA26800
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 8bit

    Hi again!

-----Original Message-----
From: ext Heinz.Prantner@radisys.com [mailto:Heinz.Prantner@radisys.com]
Sent: 30. January 2002 23:49
To: Arias-Rodriguez Ivan (NRC/Helsinki); tsvwg@ietf.org
Cc: Dave.Sclarsky@radisys.com; OrgSPDCSCTP@RADISYS_CORPORATION.radisys.com; Simon.Zhuang@radisys.com
Subject: RE: [Tsvwg] draft-ietf-tsvwg-sctpimpguide-03.txt: Questions



Hi Ivan, 

thank you for your explanations, yes, your comments help :-) 
please see my few more comments inline... 

Heinz 




	<Ivan.Arias-Rodriguez@nokia.com> 


01/30/2002 02:10 PM 


        
        To:        <Heinz.Prantner@radisys.com>, <tsvwg@ietf.org> 
        cc:        <Dave.Sclarsky@radisys.com>, <Simon.Zhuang@radisys.com>, <OrgSPDCSCTP@RADISYS_CORPORATION.radisys.com> 
        Subject:        RE: [Tsvwg] draft-ietf-tsvwg-sctpimpguide-03.txt: Questions



                 Hi!

                See some comments inline...

-----Original Message-----
From: ext Heinz.Prantner@radisys.com [mailto:Heinz.Prantner@radisys.com]
Sent: 30. January 2002 20:21
To: tsvwg@ietf.org
Cc: Dave.Sclarsky@radisys.com; Simon.Zhuang@radisys.com; OrgSPDCSCTP@RADISYS_CORPORATION.radisys.com
Subject: Re: [Tsvwg] draft-ietf-tsvwg-sctpimpguide-03.txt: Questions



Hello Sctp, 

In sctpimpguide chapter *2.8 Issues with Fast Retransmit* in *New text: (Section 7.2.4)* paragraph 5) it reads: 

"...However,  as they are marked for retransmission they will be retransmitted 
     later on as soon as cwnd allows." 

This refers to the TSN's marked for fast retransmission, but could not been fast retransmitted due to 
restriction in paragraph 3 (only one single packet allowed). 


---------------------
                Exactly.
---------------------


Question: Do we wait for T3 for retransmission, or do we *delay* the fast retransmission until consecutive SACKs 
will open the CWND. We feel that the latter is applicable and the impguide needs clarification. In our implementation, 


---------------------
                The idea is the second one. Take into account that once you have marked one packet for retransmission, it is not supposed to be on flight any more, and thus no T3 timer should be running for that TSN. The problem with a Fast retransmission is that it shrinks the cwnd. Then, if several TSNs must be fast retransmitted and their total size is larger than one MTU, it is highly probable that you won't be able to send all of them...

                When the cwnd grows later on thanks to the received SACK chunks, then you will be able to retransmit it...
---------------------


we do not reset the TSN.Missing.Report counter, if we can not FR immediatly, the next SACK which opens CWND, 
will kick off the *delayed fast retransmission* for those TSNs. 


---------------------
                Uhm... :-/

                Do you mean that if the next SACK does not acknowledge those TSNs not already fast retransmitted, you make another fast retransmission?

                But the problem with this is that you will receive many SACKs. They are on the wire comming back... If you do that, you will apply the Fast Retransmission algorithm several times, and you will lower the cwnd a lot. If you make a fast retransmission you will normaly have suddenly more outstanding data than the value of cwnd. Then you will have to wait until cwnd grows (and the number of MTUs on flight lowers) before you can send anything (the chunks marked for retransmission). 

[HP] I was not precise: we basically do what you describe above in *The idea is the second one*. We do not apply FR algorithm several times. 
It would shrink the cwnd a lot, as you say.
---------------------


paragraph 3) says, the cwnd SHOULD be ignored for fast retransmission. What about rwnd? If we receive a SACK 


---------------------
                Well, if you have to retransmit something, the rwnd will never limit you, as you consider those TSNs to be retransmitted as not being on flight any more, and thus they do not occupy any space at the receiver... You decrease rwnd by the size of all the TSNs to be fast retransmitted, and then you add only some of them... you can't have more than before (though some politicians would say that it is possible :->) 

[HP] "...have to retransmit something, the rwnd will never limit you...", can we have that in the impguide? 
The RFC says the opposite: 

6.1  Transmission of DATA Chunks 

... 

The following general rules MUST be applied by the data sender for 
   transmission and/or retransmission of outbound DATA chunks: 
                ^^^^^^^^^^^^^^^^^^^^^ 
                
   A) At any given time, the data sender MUST NOT transmit new data to 
      any destination transport address if its peer's rwnd indicates 
      that the peer has no buffer space 

[Arias-Rodriguez Ivan (NRC/Helsinki)] Yep... When I answered I didn't consider the case when the SACK sender is shrinking its buffers... Probably we should also say something about that in the Fast Retransmit section... :-/
 
 
 
---------------------


which kicks off fast retransmission, where the SACK advertises rwnd=0, we are probably blocked. The impguide 


---------------------
                Uhm... I get your point... So the receiver is effectively shrinking its buffers... bad buy... :-0

                Maybe it could be mentioned... 

[HP] according to above statement, no rwnd limitation at retrans, it will work ... and might be the only way to help the receiver 
 (the one who advertised rwnd=0) to free his buffers, since he needs that TSN, otherwise he is blocked.

 
[Arias-Rodriguez Ivan (NRC/Helsinki)]  Well, you always can have at least one TSN in flight. If due to rwnd limitation you can't send the datagram, you should check if you have any other TSN on flight. As always, if there is no TSN in flight rwnd does not limit.
 
If the receiver has its buffers full, but still need a previous TSN, then it should renege one of the subsequent TSNs in order to be able to deliver something to the upper user and then free some space...
 
Summarizing, I think that the best behaviour is not letting cwnd limit us, but take a look at rwnd. If rwnd is less than what we would like to send (the first TSN marked for retransmission), then we can't send anything unless there isn't any other TSN in flight (taking into account that all those TSNs marked for fast retransmission are not in flight any more)
---------------------


says 1 single packet according to PMTU is allowed, cwnd SHOULD be ignored, but does not mention rwnd. 


---------------------
                Yeah... You probably have already that packet on flight...
---------------------


The impguide defines max burst rules for FR. Why is cwnd/rwnd not sufficient? What makes a SACK acknowledging a FR different to apply 
that rule? 


---------------------
                Imagine that if there is one TSN missing and the subsequent TSNs have already arrive. In the normal case, when the TSNs are delivered in order to the upper user, then the SCTP receiver will have part of its buffer full due to all those received TSNs that can't be delivered. So, once the missing TSN arrives, all those TSNs will be delivered to the upper user and rwnd will grow considerably. If rwnd was limiting the sender, and not cwnd, suddenly the data sender could be allowed to send a lot of TSNs...

                That's what Max.Burst is for...
---------------------


BTW: we ran into the same problem addressed in *2.8 Issues with Fast Retransmit*, having high traffic, delays and losses. 
We realized, that one packet loss will cause the CWND to crumble, due to repeated FR's. 
We *fixed* the problem (before reading the implguide) by saving the number of packets in flight for that TSN 
at the first FR, and decrementing the outstanding packets counter (for that TSN) at receiption 
of each consecutive SACK. While the counter is greater zero, we did not apply another FR. I have to confess though, that keeping 
track of the number of packets in flight is not an easy task. 

Another question regarding RFC2960: 

where it says: 
C5) Karn's algorithm: RTT measurements MUST NOT be made using packets 
      that were retransmitted (and thus for which it is ambiguous 
      whether the reply was for the first instance of the packet or a 
      later instance). 

In case we have continous retransmissions we do not update the RTT measurement. 


---------------------
                In any case, after a retransmission you duplicate RTO, and thus the main use of RTT is somehow lost... :-/ 

[HP] I'm sending on primary destination, and assuming continuous fast retransmissions on the alternate destination. (FR does not update RTO, right?, and if not FR, T3 expired for the primary, ie. duplicate the RTO on primary). Now the alternate destination is busily used for FR, thus no heartbeats are sent, so no RTT measurement is done. But maybe thats ok...

[Arias-Rodriguez Ivan (NRC/Helsinki)] Ah, when you said that you had continuous retransmissions I thought they were triggered by T3. If those retransmissions are due to the Fast Retransmit algorith, then you don't double RTO.
 
In any case I think that I don't get your point. Why is it a problem? If you are sending FR then you can't use then to measure RTT, i.e. you continue sending heartbeats (look at the definition of "idle address" in page 93 of RFC 2960).
 
Moreover, you can only FR once a chunk. So, you do FR the TSN to another address, and then the T3 timer will expire after some time, and you will double its RTO. Meanwhile, either you are not sending anything new to the primary address (and thus the heartbeat algorithm applies) or you are measuring RTT with the new DATA chunks sent...
 
I don't know if the problem is that if you are FR a lot you would never take a measure of the RTT on that secondary address, which is not the case as that address continues being "idle"... :-/
 
Could you explain better the problem?
 
Thanks!
 
BR Iv�n Arias Rodr�guez 
---------------------

Also, above rule implies, that RTT measuremt (for DATA) is performed only on the *current* destination, where 
current is the primary, if active. The RtoPending flag is per destination, where the SACK __MAY__ come 
back on a different path. Shouldnt the RtoPending flag relate to the assoc? 


---------------------
                I'm not sure if I understand your question... :-/

                But what you test is the RTT of a specific address. The SACK are supposed to be sent to the source address of the DATA that originates it (note that due to the delayed SACK algorithm this is not always possible). If the SACK comes to another different address, then the RTT is not that exact, but this should be a corner case.

                In any case, you measure the RTT of the rest of addresses (not the primary one) using the heartbeats... At the end, you are measuring the RTT of the primary address once every RTO, and you measure the RTT of the other addresses once every heartbeating interval (only *one* of those addresses at a time)...

                I hope these comments are of any help...

                Bye!

                BR Iv�n Arias Rodr�guez

---------------------


Heinz Prantner 





_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Thu Jan 31 00:20:39 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id AAA11079
	for <tsvwg-archive@odin.ietf.org>; Thu, 31 Jan 2002 00:20:39 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id AAA13515
	for tsvwg-archive@odin.ietf.org; Thu, 31 Jan 2002 00:20:42 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id AAA12953;
	Thu, 31 Jan 2002 00:03:17 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id AAA12924
	for <tsvwg@optimus.ietf.org>; Thu, 31 Jan 2002 00:03:16 -0500 (EST)
Received: from hutmail.com ([61.79.71.135])
	by ietf.org (8.9.1a/8.9.1a) with SMTP id AAA10839
	for <tsvwg@ietf.org>; Thu, 31 Jan 2002 00:03:01 -0500 (EST)
Message-Id: <200201310503.AAA10839@ietf.org>
Reply-To: turbozet@hutmail.com
From: Lee <turbozet@hutmail.com>
To: <tsvwg@ietf.org>
Mime-Version: 1.0
Content-Type: text/html; charset="ks_c_5601-1987"
Date: Thu, 31 Jan 2002 14:00:59 +0900
X-User: 2.61-hkilfmmu-hojlnt-Ffkiq
Subject: [Tsvwg] (광 고) 획기적인 홍보데이터.
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

<html>
<head>
<meta http-equiv="content-type" content="text/html; charset=euc-kr">
<style>
<!-- A:link { text-decoration:none; } A:visited { text-decoration:none; } A:active { text-decoration:none; } A:hover { text-decoration:none; background:#C6E3F7; } -->
</style>
</head>
<body oncontextmenu="return false" bgcolor="white" text="black" link="blue" vlink="purple" alink ="red" ondragstart ="return false" onselectstart="return false">
<p style="LINE-HEIGHT: 113%"><span style="FONT-SIZE: 10pt"><br>안녕하십니까? 메일데이터를 저렴한 
가격에 드립니다.<br><font face="굴림">본 메일은 정보통신부 권고사항에 의거하여 제목에 
광고라 표시된 메일입니다.<br>본 내용은 스팸메일의 유형에 아무것도 속하지 않습니다.(피라미드, 
행운의편지, 성인관련홍보 등)<br>또한 각 사이트의 공개게시판에서 임의추출한 것이므로 귀하의 
신용정보와 아무런 연관이 없습니다.<br><br>프로그램으로 추출하여&nbsp;작업한 
메일데이터의 특징은 다음과 같습니다.<br></font><font color="#3366ff"><b>다음우표제?</b></font> 
안심하십시오. 철저한 작업으로 한메일이 없습니다. (한메일 포함된 것도 있음)<br><font color="#3366ff"><b>중복메일?</b></font> 
안심하십시오. 중복메일이 없습니다. 수신거부에 편리합니다.<br><font color="#3366ff"><b>통합?</b></font><font color="teal"><b> 
</b></font>편리하게 사용하시도록 똑같은 개수로 나누었습니다.<br><font color="#3366ff"><b>가격?</b></font> 저렴합니다. 
비싸지 않습니다.<br><font color="#3366ff"><b>호환?</b></font> 한줄에 
하나씩, 그리고 TXT로 저장되어 있습니다.<br> &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;어떤 
프로그램과도 호환가능합니다.<br><br>...등등의 장점이 있습니다.<br>드라마틱한 홍보를 
위한 필수데이터! 이런 기회는 다시오지 않습니다.<br>사이트를 참고해 주십시오. 건전한 용도로 쓰지 
않는 분에게 판매하지 않으며,<br>각종 이용용도를 접수후, 엄선하여 몇분에게만 판매합니다.</span></p>
<p style="LINE-HEIGHT: 113%"><span style="FONT-SIZE: 10pt"><b><a href="http://come.to/adver" target="_blank"><font color="#0000cc">come.to/adver</font></a><font color="#0000cc"> 
&nbsp;</font></b><a href="http://adver.ox.ro" target="_blank"><b><font color="#0000cc">adver.ox.ro</font></b></a><a href="http://adver.ce.ro" target="_blank"><b><font color="#0000cc"><br>adver.ce.ro</font></b></a><b><font color="#0000cc"> 
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;</font></b><a href="http://adver.pe.ky" target="_blank"><b><font color="#0000cc">adver.pe.ky</font></b></a></span></p>
<p style="LINE-HEIGHT: 113%"><span style="FONT-SIZE: 10pt"><font color="black">접속이 
느리면 adver2.(ce.ro / pe.ky / ox.ro)로 접속해 주십시오.<br>공지없이 중단되면 
adver3, adver4, 5...이렇게 접속하시면 됩니다.<br>이유없이 사이트를 비방, 방해하는 사람에게 손해배상을 요구합니다.<br>본 사이트는 회사나 기업이 
아닙니다. 오해마시길 바랍니다.<br><b>또한 본 메일은 사용자의 편의를 위해 절대 한번이상 
발송하지 않습니다.<br>따라서 수신거부하실 필요가 없습니다.<br></b>중복메일 받으시는 
경우는 다른 계정으로부터 메일포워딩을 받으시거나,<br>동일한 도메인이 있는 메일 
계정을 사용하는 경우입니다.<br>(예를들어 
kornet.net, kornet21.net 또는 kornet.mn.kr 모두 같은 도메인입니다)<br>주의하도록 
하겠습니다. 너그럽게 이해해 주십시오.<br>문의사항 있으시면 </font><A href="mailto:adver56@lycos.co.kr"><font color="black"><u>메일</u></font></a><font color="black">주십시오. 
계정이 종료되어 리턴되어 돌아온다면 죄송합니다.<br>&nbsp;</font></span></p>
</body>
</html>

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@optimus.ietf.org  Thu Jan 31 08:11:57 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id IAA25389
	for <tsvwg-archive@odin.ietf.org>; Thu, 31 Jan 2002 08:11:56 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id IAA14371
	for tsvwg-archive@odin.ietf.org; Thu, 31 Jan 2002 08:12:00 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id HAA13143;
	Thu, 31 Jan 2002 07:48:20 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id HAA13112
	for <tsvwg@optimus.ietf.org>; Thu, 31 Jan 2002 07:48:18 -0500 (EST)
Received: from mgw-x2.nokia.com (mgw-x2.nokia.com [131.228.20.22])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id HAA24903
	for <tsvwg@ietf.org>; Thu, 31 Jan 2002 07:48:13 -0500 (EST)
From: john.loughney@nokia.com
Received: from esvir03nok.nokia.com (esvir03nokt.ntc.nokia.com [172.21.143.35])
	by mgw-x2.nokia.com (Switch-2.1.0/Switch-2.1.0) with ESMTP id g0VCm6Z19901
	for <tsvwg@ietf.org>; Thu, 31 Jan 2002 14:48:06 +0200 (EET)
Received: from esebh003.NOE.Nokia.com (unverified) by esvir03nok.nokia.com
 (Content Technologies SMTPRS 4.2.5) with ESMTP id <T58c96cf17dac158f23076@esvir03nok.nokia.com>;
 Thu, 31 Jan 2002 14:48:02 +0200
Received: from esebe004.NOE.Nokia.com ([172.21.138.44]) by esebh003.NOE.Nokia.com with Microsoft SMTPSVC(5.0.2195.3779);
	 Thu, 31 Jan 2002 14:48:02 +0200
X-MimeOLE: Produced By Microsoft Exchange V6.0.5762.3
content-class: urn:content-classes:message
MIME-Version: 1.0
Content-Type: text/plain;
	charset="iso-8859-1"
Subject: RE: [Tsvwg] Comment on: draft-ietf-tsvwg-udp-lite-00.txt
Date: Thu, 31 Jan 2002 14:48:01 +0200
Message-ID: <0C1353ABB1DEB74DB067ADFF749C4EEF5D9515@esebe004.NOE.Nokia.com>
Thread-Topic: [Tsvwg] Comment on: draft-ietf-tsvwg-udp-lite-00.txt
Thread-Index: AcGpnrQDThLxvpveSAChWVkQHM33AAAtk4ug
To: <Lars-Erik.Jonsson@epl.ericsson.se>, <tsvwg@ietf.org>
Cc: <lln@cdt.luth.se>, <micke@cs.arizona.edu>, <steve@cs.arizona.edu>
X-OriginalArrivalTime: 31 Jan 2002 12:48:02.0233 (UTC) FILETIME=[84EFBA90:01C1AA55]
Content-Transfer-Encoding: 8bit
X-MIME-Autoconverted: from quoted-printable to 8bit by optimus.ietf.org id HAA13113
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 8bit

Hi Lars-Erik,

> I think you may have misread the UDP Lite draft. Also, the
> potential header compression problem you mention is not as
> difficult as one may believe. See details below...

I don't think it is difficult, see comments in-line.
 
> > RFC3095 ROHC header compression cannot be used for UDP Lite.
> 
> It would just not compress if the Coverage fields is not equal 
> to Length.
>  
> > ...so a new ROHC profile is needed.
> 
> Yes, to be able to compress it.

This is exactly what I meant - 3095 could not compress UDP-lite
& a new profile would be needed.

> For more than a year now, we have been discussing the two alternatives
> of either having a separate identifier or share identifier 
> with UDP. For the reasons described in the draft, the latter alternative 
> has been given up. Anyway, I agree with you that the draft could probably 
> be clearer in this regard, i.e. avoid wording such as "propose to".

I think you need to be more explicit - IANA will balk at the IANA
section if you are not explicit, such as:

   "A new IP protocol identifier will be allocated for UDP Lite."

> By the way, the header compression aspects you refer to have been
> discussed (in Minneapolis last year, I think), and we agreed that for
> HC purposes both approaches would work. The reason is that the current
> RFC3095 profiles would automatically not compress UDP Lite if the 
> coverage field is not equal to length, and new UDP Lite-aware profiles
> could be made able to handle (compress) both UDP variants. However, as
> mentioned above, this is not an issue any longer since we have agreed
> not to share protocol id with UDP.

Sounds reasonable to me.

Thanks,
John

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@ns.ietf.org  Thu Jan 31 14:37:24 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id OAA09439
	for <tsvwg-archive@odin.ietf.org>; Thu, 31 Jan 2002 14:37:24 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id OAA17144
	for tsvwg-archive@odin.ietf.org; Thu, 31 Jan 2002 14:37:28 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id OAA13438;
	Thu, 31 Jan 2002 14:00:17 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id OAA13392
	for <tsvwg@ns.ietf.org>; Thu, 31 Jan 2002 14:00:13 -0500 (EST)
Received: from bfgbhome.inetint.com (IDENT:root@tnt-dal-42-239.dallas.net [209.44.42.239])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id OAA08069;
	Thu, 31 Jan 2002 14:00:01 -0500 (EST)
Received: (from brian@localhost)
	by bfgbhome.inetint.com (8.9.3/8.9.3) id MAA01915;
	Thu, 31 Jan 2002 12:59:56 -0600
Date: Thu, 31 Jan 2002 12:59:56 -0600
From: "Brian F. G. Bidulock" <bidulock@openss7.org>
To: Meyknecht Bernward <Bernward.Meyknecht@icn.siemens.de>
Cc: "'sigtran@ietf.org'" <sigtran@ietf.org>, tsvwg@ietf.org
Message-ID: <20020131125956.J31721@openss7.org>
Reply-To: bidulock@openss7.org
Mail-Followup-To: Meyknecht Bernward <Bernward.Meyknecht@icn.siemens.de>,
	"'sigtran@ietf.org'" <sigtran@ietf.org>, tsvwg@ietf.org
References: <E01FD2A7A400D511A97C0008C791E2DF4AA38D@MCHH262E>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
User-Agent: Mutt/1.2i
In-Reply-To: <E01FD2A7A400D511A97C0008C791E2DF4AA38D@MCHH262E>; from Bernward.Meyknecht@icn.siemens.de on Thu, Jan 31, 2002 at 07:03:35PM +0100
Organization: http://www.openss7.org/
Dsn-Notification-To: <bidulock@openss7.org>
Subject: [Tsvwg] Re: [Sigtran] SCTP: What about the status of flow control in units of messages per-association ?
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

Bernward,

I believe TSVWG is responsible for these drafts.  Your
questions will probably get better answers there.  I have
forwarded this reply to tsvwg@ietf.org.

--brian

Meyknecht Bernward wrote:                               (Thu, 31 Jan 2002 19:03:35)
> 
> Dear all,
> 
> I'm wondering what's the status of the proposed Message Based Flow Control on 
> per-association basis for SCTP (setting flow limits in units of bytes or 
> also in units of messages). Cannot find it in the actual drafts.
> 
> I know the stream related flow control has been discussed and may now be out 
> of discussion, but I think the message based flow control on per-association base 
> still makes sense for our message based SCTP's and should work as 
> proposed in the former drafts.
> 
> What do you think?
> 
> Shouldn't it be included into the actual draft?
> 
> By the way: The abstract section in draft-ietf-tsvwg-addip-sctp-04.txt still 
> mentions the flow limit extensions per-association and per-stream.
> 
> 
> Best regards,
> 
> Bernward
> 
> 
> Siemens AG Muenchen
> ICN WN CC NA J14
> 
> 
> 
> _______________________________________________
> Sigtran mailing list
> Sigtran@ietf.org
> https://www1.ietf.org/mailman/listinfo/sigtran

-- 
Brian F. G. Bidulock
bidulock@openss7.org
http://www.openss7.org/

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@ns.ietf.org  Thu Jan 31 15:02:03 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id PAA10397
	for <tsvwg-archive@odin.ietf.org>; Thu, 31 Jan 2002 15:01:58 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id PAA20428
	for tsvwg-archive@odin.ietf.org; Thu, 31 Jan 2002 15:02:02 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id OAA17792;
	Thu, 31 Jan 2002 14:46:21 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id OAA17746
	for <tsvwg@ns.ietf.org>; Thu, 31 Jan 2002 14:46:18 -0500 (EST)
Received: from motgate.mot.com (motgate.mot.com [129.188.136.100])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id OAA09754;
	Thu, 31 Jan 2002 14:46:12 -0500 (EST)
Received: [from pobox4.mot.com (pobox4.mot.com [10.64.251.243]) by motgate.mot.com (motgate 2.1) with ESMTP id MAA24218; Thu, 31 Jan 2002 12:46:13 -0700 (MST)]
Received: [from relay1.cig.mot.com (relay1.cig.mot.com [136.182.15.23]) by pobox4.mot.com (MOT-pobox4 2.0) with ESMTP id MAA03621; Thu, 31 Jan 2002 12:46:12 -0700 (MST)]
Received: from quantix20.cig.mot.com (quantix20 [160.47.58.31])
	by relay1.cig.mot.com (8.11.4/8.11.4) with ESMTP id g0VJk8929431;
	Thu, 31 Jan 2002 13:46:09 -0600 (CST)
Received: from cig.mot.com (d50-384a.cig.mot.com [160.47.56.74])
	by quantix20.cig.mot.com (8.8.8+Sun/8.8.8) with ESMTP id NAA16867;
	Thu, 31 Jan 2002 13:50:58 -0600 (CST)
Message-ID: <3C599F49.1A93E041@cig.mot.com>
Date: Thu, 31 Jan 2002 13:47:21 -0600
From: Qiaobing Xie <xieqb@cig.mot.com>
X-Mailer: Mozilla 4.77 [en] (Win98; U)
X-Accept-Language: en
MIME-Version: 1.0
To: bidulock@openss7.org
CC: Meyknecht Bernward <Bernward.Meyknecht@icn.siemens.de>, sigtran@ietf.org,
        tsvwg@ietf.org
References: <E01FD2A7A400D511A97C0008C791E2DF4AA38D@MCHH262E> <20020131125956.J31721@openss7.org>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Transfer-Encoding: 7bit
Subject: [Tsvwg] Re: [Sigtran] SCTP: What about the status of flow control in units of
 messages per-association ?
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit

Hi, Bernward,

There were some extensive discussions about what SCTP flow control
really means at the London IETF meeting (if I remember it correctly). We
sort of came to the conclusion that there is still some research work
needed to define a truly effective stream-based flow control function
inside SCTP. What we proposed before is actually a flow limiting
mechanism and its usefulness became a little unclear to us after some
further analysis. 

-Qiaobing

"Brian F. G. Bidulock" wrote:
> 
> Bernward,
> 
> I believe TSVWG is responsible for these drafts.  Your
> questions will probably get better answers there.  I have
> forwarded this reply to tsvwg@ietf.org.
> 
> --brian
> 
> Meyknecht Bernward wrote:                               (Thu, 31 Jan 2002 19:03:35)
> >
> > Dear all,
> >
> > I'm wondering what's the status of the proposed Message Based Flow Control on
> > per-association basis for SCTP (setting flow limits in units of bytes or
> > also in units of messages). Cannot find it in the actual drafts.
> >
> > I know the stream related flow control has been discussed and may now be out
> > of discussion, but I think the message based flow control on per-association base
> > still makes sense for our message based SCTP's and should work as
> > proposed in the former drafts.
> >
> > What do you think?
> >
> > Shouldn't it be included into the actual draft?
> >
> > By the way: The abstract section in draft-ietf-tsvwg-addip-sctp-04.txt still
> > mentions the flow limit extensions per-association and per-stream.
> >
> >
> > Best regards,
> >
> > Bernward
> >
> >
> > Siemens AG Muenchen
> > ICN WN CC NA J14
> >
> >
> >
> > _______________________________________________
> > Sigtran mailing list
> > Sigtran@ietf.org
> > https://www1.ietf.org/mailman/listinfo/sigtran
> 
> --
> Brian F. G. Bidulock
> bidulock@openss7.org
> http://www.openss7.org/
> 
> _______________________________________________
> Sigtran mailing list
> Sigtran@ietf.org
> https://www1.ietf.org/mailman/listinfo/sigtran

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@ns.ietf.org  Thu Jan 31 15:09:32 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id PAA10676
	for <tsvwg-archive@odin.ietf.org>; Thu, 31 Jan 2002 15:09:31 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id PAA20867
	for tsvwg-archive@odin.ietf.org; Thu, 31 Jan 2002 15:09:35 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id OAA18087;
	Thu, 31 Jan 2002 14:47:07 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id OAA18039
	for <tsvwg@ns.ietf.org>; Thu, 31 Jan 2002 14:47:04 -0500 (EST)
Received: from sj-msg-core-2.cisco.com (sj-msg-core-2.cisco.com [171.69.24.11])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id OAA09761
	for <tsvwg@ietf.org>; Thu, 31 Jan 2002 14:46:59 -0500 (EST)
Received: from mira-sjc5-2.cisco.com (mira-sjc5-2.cisco.com [171.71.163.16])
	by sj-msg-core-2.cisco.com (8.11.3/8.9.1) with ESMTP id g0VJjx616369;
	Thu, 31 Jan 2002 11:45:59 -0800 (PST)
Received: from stewart.chicago.il.us (ssh-sj1.cisco.com [171.68.225.134])
	by mira-sjc5-2.cisco.com (Mirapoint)
	with ESMTP id AAY90279;
	Thu, 31 Jan 2002 11:45:58 -0800 (PST)
Message-ID: <3C599EF4.47688888@stewart.chicago.il.us>
Date: Thu, 31 Jan 2002 13:45:57 -0600
From: Randall Stewart <randall@stewart.chicago.il.us>
X-Mailer: Mozilla 4.76 [en] (X11; U; Linux 2.2.12 i386)
X-Accept-Language: en
MIME-Version: 1.0
To: Meyknecht Bernward <Bernward.Meyknecht@icn.siemens.de>
CC: tsvwg@ietf.org, Phil Conrad <conrad@joda.cis.temple.edu>
Subject: Re: [Tsvwg] Re: [Sigtran] SCTP: What about the status of flow control in 
 units of messages per-association ?
References: <E01FD2A7A400D511A97C0008C791E2DF4AA38D@MCHH262E> <20020131125956.J31721@openss7.org>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Transfer-Encoding: 7bit
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org
Content-Transfer-Encoding: 7bit

Bernward:

The flow limit/control feature has been removed from the
add-ip draft and a hook put in for the eventual building
of such a feature (outside of the base SCTP protocol).

Phil Conrad (who is copied on this email) is working on
this.. Phil any progress? 

And it looks like you point out a bug in my editting out
of the flow control.. I missed the abstract.. sigh.. oh
well there is always -05 :>

R

> Meyknecht Bernward wrote:                               (Thu, 31 Jan 2002 19:03:35)
> >
> > Dear all,
> >
> > I'm wondering what's the status of the proposed Message Based Flow Control on
> > per-association basis for SCTP (setting flow limits in units of bytes or
> > also in units of messages). Cannot find it in the actual drafts.
> >
> > I know the stream related flow control has been discussed and may now be out
> > of discussion, but I think the message based flow control on per-association base
> > still makes sense for our message based SCTP's and should work as
> > proposed in the former drafts.
> >
> > What do you think?
> >
> > Shouldn't it be included into the actual draft?
> >
> > By the way: The abstract section in draft-ietf-tsvwg-addip-sctp-04.txt still
> > mentions the flow limit extensions per-association and per-stream.
> >
> >
> > Best regards,
> >
> > Bernward
> >
> >
> > Siemens AG Muenchen
> > ICN WN CC NA J14
> >
> >
> >
> > _______________________________________________
> > Sigtran mailing list
> > Sigtran@ietf.org
> > https://www1.ietf.org/mailman/listinfo/sigtran
> 
> --
> Brian F. G. Bidulock
> bidulock@openss7.org
> http://www.openss7.org/
> 
> _______________________________________________
> tsvwg mailing list
> tsvwg@ietf.org
> https://www1.ietf.org/mailman/listinfo/tsvwg

-- 
Randall R. Stewart
randall@stewart.chicago.il.us 815-342-5222 (cell phone)

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



From daemon@ns.ietf.org  Thu Jan 31 16:44:58 2002
Received: from optimus.ietf.org (ietf.org [132.151.1.19] (may be forged))
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id QAA13512
	for <tsvwg-archive@odin.ietf.org>; Thu, 31 Jan 2002 16:44:57 -0500 (EST)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id QAA26003
	for tsvwg-archive@odin.ietf.org; Thu, 31 Jan 2002 16:45:02 -0500 (EST)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id QAA24563;
	Thu, 31 Jan 2002 16:24:13 -0500 (EST)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id QAA24500
	for <tsvwg@ns.ietf.org>; Thu, 31 Jan 2002 16:24:09 -0500 (EST)
Received: from bfgbhome.inetint.com (IDENT:root@tnt-dal-42-239.dallas.net [209.44.42.239])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id QAA12976;
	Thu, 31 Jan 2002 16:24:00 -0500 (EST)
Received: (from brian@localhost)
	by bfgbhome.inetint.com (8.9.3/8.9.3) id PAA05412;
	Thu, 31 Jan 2002 15:23:48 -0600
Date: Thu, 31 Jan 2002 15:23:48 -0600
From: "Brian F. G. Bidulock" <bidulock@openss7.org>
To: "Phillip T. Conrad" <conrad@joda.cis.temple.edu>
Cc: Meyknecht Bernward <Bernward.Meyknecht@icn.siemens.de>, sigtran@ietf.org,
        tsvwg <tsvwg@ietf.org>
Message-ID: <20020131152348.L31721@openss7.org>
Reply-To: bidulock@openss7.org
Mail-Followup-To: "Phillip T. Conrad" <conrad@joda.cis.temple.edu>,
	Meyknecht Bernward <Bernward.Meyknecht@icn.siemens.de>,
	sigtran@ietf.org, tsvwg <tsvwg@ietf.org>
References: <E01FD2A7A400D511A97C0008C791E2DF4AA38D@MCHH262E> <5.1.0.14.2.20020131144008.02b2ce38@joda.cis.temple.edu>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
User-Agent: Mutt/1.2i
In-Reply-To: <5.1.0.14.2.20020131144008.02b2ce38@joda.cis.temple.edu>; from conrad@joda.cis.temple.edu on Thu, Jan 31, 2002 at 03:22:54PM -0500
Organization: http://www.openss7.org/
Dsn-Notification-To: <bidulock@openss7.org>
Subject: [Tsvwg] Re: [Sigtran] SCTP: What about the status of flow control in units of messages per-association ?
Sender: tsvwg-admin@ietf.org
Errors-To: tsvwg-admin@ietf.org
X-Mailman-Version: 1.0
Precedence: bulk
List-Id: Transport Area Working Group <tsvwg.ietf.org>
X-BeenThere: tsvwg@ietf.org

Phillip,

For signalling application (SIGTRAN UAs) we had a long discussion
on TSV concerning application flow control and hacking the rwnd.
The result of those discussions was the concensus that an application
should not muck with rwnd (i.e., set it to zero) in an attempt to
acheive application flow control.

The two SIGTRAN UAs which are most sensitive to flow controls are
M2PA and M2UA.  M2UA has a Correlation Id and Correlation Id Ack
which can be used on a per link (stream) basis to acheive application
congestion and flow control.  The next version of the M2PA draft will
include sequence numbers or a similar mechanism for application
congestion and flow control independent of SCTP.  The opensource
implementation of M2PA at http://www.openss7.org/ provides such
application congestion and flow control in anticipation of draft-04.

The remaining SIGTRAN SS7 UAs: M3UA, SUA (and the proposed TUA) have
access to the acknowledged data service provided by the RFC2960
SCTP-ULP API and present in some implementations (such as the
opensource implementations of SCTP at http://www.openss7.org/ )
Using the acknowledged data service it is not possible to acheive
full application level congestion and flow control (for which a
proposal for the same Correlation Id and Ack procedures of M2UA
were proposed for M3UA and SUA.)   Nevertheless, the acknowledged
data service does provide for per stream flow control for application
threads competing for cwnd and rwnd without the need for additional
protocol in SCTP.

To summarize, I don't believe that association flow controls would
provide any benefit to SIGTRAN SS7 UAs that could not be, perhaps
more easliy, accomplished in the UA protocols themselves.

The other alternative which has oft times been discussed is to
open a separate SCTP association for each stream of traffic,
and to not use SCTP's streams at all for unbalanced stream loads.
Some think this drastic, some think it suitable.

I would put it to you that even addip is not necessary: an
implementation which uses verification tag, port number and local IP
address only can easily detect when a peer sends from a new IP address
and include its address in its destination list.  Deleting IP addresses
can easily be accomplished by treating aborts on a per-destination
basis rather than a per-association basis.

--brian


Phillip T. Conrad wrote:                                  (Thu, 31 Jan 2002 15:22:54)
> At 07:03 PM 1/31/2002 +0100, Meyknecht Bernward 
> <Bernward.Meyknecht@icn.siemens.de> wrote:
> >I'm wondering what's the status of the proposed Message Based Flow Control on
> >per-association basis for SCTP (setting flow limits in units of bytes or
> >also in units of messages). Cannot find it in the actual drafts.
> >
> >I know the stream related flow control has been discussed and may now be out
> >of discussion, but I think the message based flow control on 
> >per-association base
> >still makes sense for our message based SCTP's and should work as
> >proposed in the former drafts.
> >
> >What do you think?
> 
> I had started some work on implementing all of the limits (byte and 
> message) per stream, and per association; this was work in the reference 
> implementation found at (sctp.chicago.il.us.)   However, I suspended this 
> work because the concept itself ran into some serious "concern" at the 
> London IETF.      In particular, there was concern that extra control 
> traffic would have to be added to make these limits work properly 
> (specifically, delivery acknowledgements), and that this control traffic 
> would pose an unacceptable burden on the protocol.
> 
> After further reflection, I decided to let it drop for now pending further 
> research that I hope will, at some point in the future,  provide some real 
> data to either
> 
> (a) show that one or more of these limits are useful and/or necessary, and 
> the the control traffic does not introduce a significant penalty or else
> 
> (b) show that I was wrong and the critics were right. i.e. to show that 
> this stuff is better done in application layer protocols, not in SCTP 
> itself.     (Note that if the application handles these extra limits, there 
> is still a need for signalling bits.. we've just moved the signalling bits 
> from one layer to another.  But an argument can be made (and was made) that 
> the needs for these limits are application specific enough that they do not 
> belong in SCTP.
> 
> For my part, my main motivation for pursuing per-stream flow limits was to 
> ensure that messages could not be indefinitely postponed (through the 
> scenario listed at the bottom of this message), and in a larger sense to 
> allow the application some control over the competition for window space 
> (and hence bandwidth) among multiple streams.     Through some discussions 
> at the London IETF, a different fix for the indefinite postponement problem 
> was suggested, and slated for inclusion in the implementer's guide (though 
> I'm not sure whether it made it or not... I need to check.)
> 
> In the meantime, if there are some good reasons to bring back the 
> association level message limit, I'd be glad to ressurrect the deleted text 
> and dust off the deleted source code, and try to make these 
> work.   However, since I'm a "transport guy" and not a "telephony 
> signalling guy" I need some help from the sigtran community to "make the 
> case for this" or any other additional flow control limits (beyond the base 
> protocols rwnd and cwnd) that we want to see included as extensions.
> 
> If someone else can make a compelling case for why these limits are needed, 
> and convince the opinion leaders in the two working groups (sigtran and 
> tsv) I'm happy to do the grunt work on the ID language, implementation 
> (i.e. patches for one or both of the two open source implementations) and 
> testing.
> 
> Thanks,
> Phill Conrad <conrad@acm.org>
> Asst. Professor, Temple University, Philadephia PA
> http://netlab.cis.temple.edu
> 
> 
> **********************************************************************
> **********************************************************************
> **********************************************************************
> 
> PS: Here's the original scenario that I suggested as a motivation for 
> per-stream flow limits.
> 
> Consider for example, an application that strictly alternates between 
> writing 400-byte messages to stream 0 and 1300-byte messages to stream 
> 1.   Suppose now that the application process consuming the messages is 
> temporarily suspended due to a context switch, and the sending application 
> fills up the available per-association rwnd buffer space.   If the messages 
> are consumed by two different threads at the receiving application, it is 
> possible that the window will open up space 800 or 1200 bytes at a 
> time.  In this situation, there is nothing in RFC2960 that would prevent an 
> implementation from always being "greedy", and choosing from among the 
> packets at the head of each stream one that will fit in the currently 
> available min(rwnd,cwnd) space.    In fact, because of concern about head 
> of line blocking, some folks insisted that this is exactly what an 
> implementation SHOULD do!  I argue, on the contrary, that a greedy approach 
> can result in head-of-line blocking for the *larger* chunks at the expense 
> of smaller ones, similar to what happens with a "shortest job first" 
> strategy for scheduling jobs.
> 
> Per stream flow control fixes this by reserving flow control space for each 
> individual stream.   It allow an application to ensure the progress of each 
> stream, at the expense of reduced throughput (if the window/rtt < 
> bandwidth*delay product), and at the expense of a minimum packet size on 
> each stream (equal to the per-stream byte limit.)   Note that as an 
> extension, it is an entirely optional feature, and would not restrict 
> applications that do not choose to use it in anyway.
> 
> However, a more lightweight fix was suggested for this 
> problem.  Specifically, it was suggested that language be added to the 
> implementer's guide to ensure that the order in which messages are 
> submitted to SCTP by the Upper Layer Protocol must be preserved as the 
> initial sending order (note that this does NOT re-introduce head-of-line 
> blocking of the type that multiple streams in SCTP is designed to prevent; 
> it is still possible to deliver chunks out of sequence at the *receiver*, 
> for example, when a chunk with a lower TSN from an *unrelated* stream is 
> lost. )    This was agreed to by several folks during a discussion at the 
> London IETF, though I don't know if it ever made it into the implementer's 
> guide.   (I should check on this.)
> 
> ptc
> 
> 
> _______________________________________________
> Sigtran mailing list
> Sigtran@ietf.org
> https://www1.ietf.org/mailman/listinfo/sigtran

-- 
Brian F. G. Bidulock
bidulock@openss7.org
http://www.openss7.org/

_______________________________________________
tsvwg mailing list
tsvwg@ietf.org
https://www1.ietf.org/mailman/listinfo/tsvwg



