From daemon@optimus.ietf.org  Sat Jun  1 14:27:42 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 OAA26090
	for <tsvwg-archive@odin.ietf.org>; Sat, 1 Jun 2002 14:27:42 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id OAA14720
	for tsvwg-archive@odin.ietf.org; Sat, 1 Jun 2002 14:28:09 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id NAA13641;
	Sat, 1 Jun 2002 13:55:04 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id NAA13612
	for <tsvwg@optimus.ietf.org>; Sat, 1 Jun 2002 13:55:02 -0400 (EDT)
Received: from bfgbhome.inetint.com (IDENT:root@tnt-dal-42-248.dallas.net [209.44.42.248])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id NAA25841
	for <tsvwg@ietf.org>; Sat, 1 Jun 2002 13:54:33 -0400 (EDT)
Received: (from brian@localhost)
	by bfgbhome.inetint.com (8.9.3/8.9.3) id MAA26500;
	Sat, 1 Jun 2002 12:54:59 -0500
Date: Sat, 1 Jun 2002 12:54:59 -0500
From: "Brian F. G. Bidulock" <bidulock@openss7.org>
To: tsvwg@ietf.org, Randall Stewart <randall@stewart.chicago.il.us>
Message-ID: <20020601125459.A26453@openss7.org>
Reply-To: bidulock@openss7.org
Mime-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
User-Agent: Mutt/1.2i
Organization: http://www.openss7.org/
Dsn-Notification-To: <bidulock@openss7.org>
Content-Transfer-Encoding: 8bit
Subject: [Tsvwg] Cause code conflict.
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,

There is a conflict in cause code values between the IG and ADD-IP:

draft-ietf-tsvwg-addip-sctp-05.txt: Section 3.3

    Cause Code     
    Value          Cause Code 
    ---------      ----------------
    0x000B          Request to Delete Last Remaining IP Address.

draft-ietf-tsvwg-sctpimpguide-05.txt: Section 2.6.2

      Cause Code
      Value           Cause Code
      ---------      ----------------
      ...
      11              Restart of an association with new addresses

Please move one of these cause codes to remove the conflict.

--brian

-- 
Brian F. G. Bidulock    ¦ The reasonable man adapts himself to the ¦
bidulock@openss7.org    ¦ world; the unreasonable one persists in  ¦
http://www.openss7.org/ ¦ trying  to adapt the  world  to himself. ¦
                        ¦ Therefore  all  progress  depends on the ¦
                        ¦ unreasonable man. -- George Bernard Shaw ¦

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



From daemon@optimus.ietf.org  Sat Jun  1 16:55:30 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 QAA27496
	for <tsvwg-archive@odin.ietf.org>; Sat, 1 Jun 2002 16:55:30 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id QAA19719
	for tsvwg-archive@odin.ietf.org; Sat, 1 Jun 2002 16:55:57 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id QAA18494;
	Sat, 1 Jun 2002 16:21:32 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id QAA18465
	for <tsvwg@optimus.ietf.org>; Sat, 1 Jun 2002 16:21:31 -0400 (EDT)
Received: from bfgbhome.inetint.com (IDENT:root@tnt-dal-42-248.dallas.net [209.44.42.248])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id QAA27086
	for <tsvwg@ietf.org>; Sat, 1 Jun 2002 16:21:02 -0400 (EDT)
Received: (from brian@localhost)
	by bfgbhome.inetint.com (8.9.3/8.9.3) id PAA28439;
	Sat, 1 Jun 2002 15:21:28 -0500
Date: Sat, 1 Jun 2002 15:21:28 -0500
From: "Brian F. G. Bidulock" <bidulock@openss7.org>
To: tsvwg@ietf.org, Randall Stewart <randall@stewart.chicago.il.us>
Message-ID: <20020601152128.B28346@openss7.org>
Reply-To: bidulock@openss7.org
Mime-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
User-Agent: Mutt/1.2i
Organization: http://www.openss7.org/
Dsn-Notification-To: <bidulock@openss7.org>
Content-Transfer-Encoding: 8bit
Subject: [Tsvwg] SCTP IG (SACKs)
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,

draft-ietf-tsvwg-sctpimguide-05.txt Sction 2.7 places a MUST on
sending gap reports.  Before when this was a SHOULD, it was obvious
that gap ack blocks in a SACK which exceeded an MTU could be
truncated.  Now it is not so clear.  What should an implementation
do if the number of gap ack blocks in a SACK exceeds the path MTU?
Truncate?  Force IP fragmentation?

--brian

-- 
Brian F. G. Bidulock    ¦ The reasonable man adapts himself to the ¦
bidulock@openss7.org    ¦ world; the unreasonable one persists in  ¦
http://www.openss7.org/ ¦ trying  to adapt the  world  to himself. ¦
                        ¦ Therefore  all  progress  depends on the ¦
                        ¦ unreasonable man. -- George Bernard Shaw ¦

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



From daemon@optimus.ietf.org  Sun Jun  2 17:41:38 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 RAA19551
	for <tsvwg-archive@odin.ietf.org>; Sun, 2 Jun 2002 17:41:38 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id RAA21192
	for tsvwg-archive@odin.ietf.org; Sun, 2 Jun 2002 17:42:06 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id RAA20138;
	Sun, 2 Jun 2002 17:15:09 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id RAA20109
	for <tsvwg@optimus.ietf.org>; Sun, 2 Jun 2002 17:15:07 -0400 (EDT)
Received: from ilsa.franken.de (postfix@ilsa.franken.de [193.175.24.42])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id RAA19314
	for <tsvwg@ietf.org>; Sun, 2 Jun 2002 17:14:39 -0400 (EDT)
Received: from localhost (pD9E4FC6D.dip.t-dialin.net [217.228.252.109])
	by ilsa.franken.de (Postfix) with ESMTP
	id 7BFA8245DE; Sun,  2 Jun 2002 23:14:46 +0200 (CEST)
Date: Sun, 2 Jun 2002 23:14:40 +0200
Subject: Re: [Tsvwg] Cause code conflict.
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Mime-Version: 1.0 (Apple Message framework v481)
Cc: tsvwg@ietf.org, Randall Stewart <randall@stewart.chicago.il.us>
To: bidulock@openss7.org
From: Michael Tuexen <Michael.Tuexen@icn.siemens.de>
In-Reply-To: <20020601125459.A26453@openss7.org>
Message-Id: <C0AA71D4-766D-11D6-8A30-0030654C1AB6@icn.siemens.de>
X-Mailer: Apple Mail (2.481)
Content-Transfer-Encoding: 8bit
X-MIME-Autoconverted: from quoted-printable to 8bit by optimus.ietf.org id RAA20110
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

Brian,

the AddIP Cause Code will be shifted. The error codes in the IG should
stay contiguous. The error causes in the AddIP ID will start at 0x100.

It will be fixed in the next release of the IDs.

Best regards
Michael

On Saturday, June 1, 2002, at 07:54 PM, Brian F. G. Bidulock wrote:

> Randall,
>
> There is a conflict in cause code values between the IG and ADD-IP:
>
> draft-ietf-tsvwg-addip-sctp-05.txt: Section 3.3
>
>     Cause Code
>     Value          Cause Code
>     ---------      ----------------
>     0x000B          Request to Delete Last Remaining IP Address.
>
> draft-ietf-tsvwg-sctpimpguide-05.txt: Section 2.6.2
>
>       Cause Code
>       Value           Cause Code
>       ---------      ----------------
>       ...
>       11              Restart of an association with new addresses
>
> Please move one of these cause codes to remove the conflict.
>
> --brian
>
> --
> Brian F. G. Bidulock    ¦ The reasonable man adapts himself to the ¦
> bidulock@openss7.org    ¦ world; the unreasonable one persists in  ¦
> http://www.openss7.org/ ¦ trying  to adapt the  world  to himself. ¦
>                         ¦ Therefore  all  progress  depends on the ¦
>                         ¦ unreasonable man. -- George Bernard Shaw ¦
>
> _______________________________________________
> tsvwg mailing list
> tsvwg@ietf.org
> https://www1.ietf.org/mailman/listinfo/tsvwg
>
Michael.Tuexen@micmac.franken.de


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



From daemon@optimus.ietf.org  Sun Jun  2 17:45:33 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 RAA19597
	for <tsvwg-archive@odin.ietf.org>; Sun, 2 Jun 2002 17:45:33 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id RAA21298
	for tsvwg-archive@odin.ietf.org; Sun, 2 Jun 2002 17:46:00 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id RAA20349;
	Sun, 2 Jun 2002 17:24:40 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id RAA20324
	for <tsvwg@optimus.ietf.org>; Sun, 2 Jun 2002 17:24:38 -0400 (EDT)
Received: from bfgbhome.inetint.com (IDENT:root@tnt-dal-42-248.dallas.net [209.44.42.248])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id RAA19430
	for <tsvwg@ietf.org>; Sun, 2 Jun 2002 17:24:08 -0400 (EDT)
Received: (from brian@localhost)
	by bfgbhome.inetint.com (8.9.3/8.9.3) id QAA15926;
	Sun, 2 Jun 2002 16:24:30 -0500
Date: Sun, 2 Jun 2002 16:24:30 -0500
From: "Brian F. G. Bidulock" <bidulock@openss7.org>
To: Michael Tuexen <Michael.Tuexen@icn.siemens.de>
Cc: tsvwg@ietf.org, Randall Stewart <randall@stewart.chicago.il.us>
Subject: Re: [Tsvwg] Cause code conflict.
Message-ID: <20020602162430.F28346@openss7.org>
Reply-To: bidulock@openss7.org
References: <20020601125459.A26453@openss7.org> <C0AA71D4-766D-11D6-8A30-0030654C1AB6@icn.siemens.de>
Mime-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
User-Agent: Mutt/1.2i
In-Reply-To: <C0AA71D4-766D-11D6-8A30-0030654C1AB6@icn.siemens.de>; from Michael.Tuexen@icn.siemens.de on Sun, Jun 02, 2002 at 11:14:40PM +0200
Organization: http://www.openss7.org/
Dsn-Notification-To: <bidulock@openss7.org>
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

Michael,

Thanx for the info.

--brian

On Sun, 02 Jun 2002, Michael Tuexen wrote:

> Brian,
> 
> the AddIP Cause Code will be shifted. The error codes in the IG should
> stay contiguous. The error causes in the AddIP ID will start at 0x100.
> 
> It will be fixed in the next release of the IDs.
> 
> Best regards
> Michael
> 
> On Saturday, June 1, 2002, at 07:54 PM, Brian F. G. Bidulock wrote:
> 
> > Randall,
> >
> > There is a conflict in cause code values between the IG and ADD-IP:
> >
> > draft-ietf-tsvwg-addip-sctp-05.txt: Section 3.3
> >
> >     Cause Code
> >     Value          Cause Code
> >     ---------      ----------------
> >     0x000B          Request to Delete Last Remaining IP Address.
> >
> > draft-ietf-tsvwg-sctpimpguide-05.txt: Section 2.6.2
> >
> >       Cause Code
> >       Value           Cause Code
> >       ---------      ----------------
> >       ...
> >       11              Restart of an association with new addresses
> >
> > Please move one of these cause codes to remove the conflict.
> >
> > --brian
> >
> > --
> > Brian F. G. Bidulock    ¦ The reasonable man adapts himself to the ¦
> > bidulock@openss7.org    ¦ world; the unreasonable one persists in  ¦
> > http://www.openss7.org/ ¦ trying  to adapt the  world  to himself. ¦
> >                         ¦ Therefore  all  progress  depends on the ¦
> >                         ¦ unreasonable man. -- George Bernard Shaw ¦
> >
> > _______________________________________________
> > tsvwg mailing list
> > tsvwg@ietf.org
> > https://www1.ietf.org/mailman/listinfo/tsvwg
> >
> Michael.Tuexen@micmac.franken.de
> 
> 
> _______________________________________________
> tsvwg mailing list
> tsvwg@ietf.org
> https://www1.ietf.org/mailman/listinfo/tsvwg

-- 
Brian F. G. Bidulock    ¦ The reasonable man adapts himself to the ¦
bidulock@openss7.org    ¦ world; the unreasonable one persists in  ¦
http://www.openss7.org/ ¦ trying  to adapt the  world  to himself. ¦
                        ¦ Therefore  all  progress  depends on the ¦
                        ¦ unreasonable man. -- George Bernard Shaw ¦

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



From daemon@optimus.ietf.org  Mon Jun  3 15:52:05 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 PAA29085
	for <tsvwg-archive@odin.ietf.org>; Mon, 3 Jun 2002 15:52:05 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id PAA26074
	for tsvwg-archive@odin.ietf.org; Mon, 3 Jun 2002 15:52:34 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id PAA24684;
	Mon, 3 Jun 2002 15:20:26 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id PAA24656
	for <tsvwg@optimus.ietf.org>; Mon, 3 Jun 2002 15:20:23 -0400 (EDT)
Received: from motgate.mot.com (motgate.mot.com [129.188.136.100])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id PAA28235
	for <tsvwg@ietf.org>; Mon, 3 Jun 2002 15:19:52 -0400 (EDT)
Received: [from pobox3.mot.com (pobox3.mot.com [10.64.251.242]) by motgate.mot.com (motgate 2.1) with ESMTP id MAA14057; Mon, 3 Jun 2002 12:20:19 -0700 (MST)]
Received: [from relay1.cig.mot.com (relay1.cig.mot.com [136.182.15.23]) by pobox3.mot.com (MOT-pobox3 2.0) with ESMTP id MAA26404; Mon, 3 Jun 2002 12:20:04 -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 g53JGZb13185;
	Mon, 3 Jun 2002 14:16:35 -0500 (CDT)
Received: from email.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 OAA07496;
	Mon, 3 Jun 2002 14:16:56 -0500 (CDT)
Message-ID: <3CFBC11C.1EFA747@email.mot.com>
Date: Mon, 03 Jun 2002 14:18:52 -0500
From: Qiaobing Xie <qxie1@email.mot.com>
X-Mailer: Mozilla 4.77 [en] (Win98; U)
X-Accept-Language: en
MIME-Version: 1.0
To: bidulock@openss7.org
CC: Qiaobing Xie <Qiaobing.Xie@motorola.com>,
        "Armando L. Caro Jr." <acaro@mail.eecis.udel.edu>, tsvwg@ietf.org
Subject: Re: [Tsvwg] More on PRSCTP-00
References: <Pine.GSO.4.33.0205311736060.8809-100000@ren.eecis.udel.edu> <3CF80F65.58E43813@email.mot.com> <20020531195333.D11468@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

Brian,

I see what you are getting at now. This is another issue that was
intensely discussed off-line before. It is really a tradeoff between
design simplicity and more efficiency. See more comments below.

"Brian F. G. Bidulock" wrote:
> 
> Qiaobing,
> 
> Well, that's not really what I was gettin at: consider just the
> receiver's view of this sequence:
> 
> <-- TSH   SID  SSN
> 
>     101    1    2
>     102    1    3
>     103    2   57
>     104    2   58
> 
>     FORWARD TSN 106
> 
>     108    1    5
>     109    1    6
>     110    2   60
>     111    2   61
> 
> Now, because the receiver cannot tell whether SID/SSN (1/4) was
> dropped or whether SID/SSN (2/60) was dropped (until it receives the
                              ^^^^^
       you must meant to say (2/59) here

> missing TSN 107), missing TSN 107 has blocked both streams.  If more
> streams were involved in the same way, they would all be blocked by
> the loss of one TSN.  This *is* inter-stream HOL blocking.
      ^^^^^^^^^^^^^^^
This is a misstatement. As you also came to the realization below, you
must lose >2 TSNs across both streams for the above scenario to happen
(in your example, you assumed the loss of 3 TSNs across both streams).
If you only lose one TSN, it is impossible to affect more than 2 streams
and thus there is *no chance* for any inter-stream HOL blocking.

> The problem is not as great as I first thought (all streams
> blocked), really it is just streams that have dropping that are
> blocked, but the problem is still of concern: particularly as it is
> congestion which causes drops in PR-SCTP in the first place, one can
> expect packet losses at about the same time.

It is not totally unreasonable to let a stream suffer some msg delivery
delay when the stream is missing a TSN in the first place.

> 
> The delay introducted is blocking (in the example), delivery on
> stream 1 and stream 2 until TSN 107 can be finally and successfully
> delivered.

It is so because you actually lost TSN 105, 106, and 107 to start the
scenario and each of the 2 stream lost at least one TSN, otherwise it
would be affected.

> 
> If FORWARD TSN included the fact that, say, stream 1 should "expect"
> SSN 5 and stream 2 should "expect" SSN 60 (because 4 and 59 have
> been dropped), then there would be no blocking.

Yes, it will help if we include SSN info in the Fwd-TSN, but the
question is how to encode it. Your suggestion here does not scale well
because, say, if we have 10k outbound streams, how can you encode the
"expected" SSN for each one of them in a Fwd-TSN?

An alternative might be to enlist in the Fwd-TSN each "dropped" chunk
with its TSN/SID/SSN triplet. But this means that we will still "burn" 8
bytes for each dropped data chunk, while one of the primary ideas of not
to re-send expired data is to save network bandwidth...

Maybe this info can be provided as optional, i.e., the Fwd-TSN will
always indicate the new cumTSN position, and also OPTIONALLY provide a
list of TSN/SID/SSN triplets to help the receiver message delivery. This
can even be a partial list if too many TSNs are dropped/skipped since it
will still help the receiver somewhat...

-Qiaobing

> 
> In general I don't see a problem with putting the SID/SSN in the
> FORWARD TSN.  Certainly the sizeof of the FOWARD TSN chunk could be
> limited in similar ways to SACK chunks (advisory only), then the
> approach appears to be a sound one.
> 
> --brian
>

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



From daemon@optimus.ietf.org  Mon Jun  3 16:12:02 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 QAA29624
	for <tsvwg-archive@odin.ietf.org>; Mon, 3 Jun 2002 16:12:01 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id QAA27047
	for tsvwg-archive@odin.ietf.org; Mon, 3 Jun 2002 16:12:28 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id PAA25851;
	Mon, 3 Jun 2002 15:44:12 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id PAA25823
	for <tsvwg@optimus.ietf.org>; Mon, 3 Jun 2002 15:44:10 -0400 (EDT)
Received: from bfgbhome.inetint.com (IDENT:root@tnt-dal-42-248.dallas.net [209.44.42.248])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id PAA28774
	for <tsvwg@ietf.org>; Mon, 3 Jun 2002 15:43:36 -0400 (EDT)
Received: (from brian@localhost)
	by bfgbhome.inetint.com (8.9.3/8.9.3) id OAA30907;
	Mon, 3 Jun 2002 14:43:42 -0500
Date: Mon, 3 Jun 2002 14:43:42 -0500
From: "Brian F. G. Bidulock" <bidulock@openss7.org>
To: Qiaobing Xie <qxie1@email.mot.com>
Cc: Qiaobing Xie <Qiaobing.Xie@motorola.com>,
        "Armando L. Caro Jr." <acaro@mail.eecis.udel.edu>, tsvwg@ietf.org
Subject: Re: [Tsvwg] More on PRSCTP-00
Message-ID: <20020603144342.A30667@openss7.org>
Reply-To: bidulock@openss7.org
References: <Pine.GSO.4.33.0205311736060.8809-100000@ren.eecis.udel.edu> <3CF80F65.58E43813@email.mot.com> <20020531195333.D11468@openss7.org> <3CFBC11C.1EFA747@email.mot.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
User-Agent: Mutt/1.2i
In-Reply-To: <3CFBC11C.1EFA747@email.mot.com>; from qxie1@email.mot.com on Mon, Jun 03, 2002 at 02:18:52PM -0500
Organization: http://www.openss7.org/
Dsn-Notification-To: <bidulock@openss7.org>
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

Qiaobing,

On Mon, 03 Jun 2002, Qiaobing Xie wrote:

> Brian,
> 
> I see what you are getting at now. This is another issue that was
> intensely discussed off-line before. It is really a tradeoff between
> design simplicity and more efficiency. See more comments below.
> 
> "Brian F. G. Bidulock" wrote:
> > 
> > Qiaobing,
> > 
> > Well, that's not really what I was gettin at: consider just the
> > receiver's view of this sequence:
> > 
> > <-- TSH   SID  SSN
> > 
> >     101    1    2
> >     102    1    3
> >     103    2   57
> >     104    2   58
> > 
> >     FORWARD TSN 106
> > 
> >     108    1    5
> >     109    1    6
> >     110    2   60
> >     111    2   61
> > 
> > Now, because the receiver cannot tell whether SID/SSN (1/4) was
> > dropped or whether SID/SSN (2/60) was dropped (until it receives the
>                               ^^^^^
>        you must meant to say (2/59) here

Yes, of course.

> 
> > missing TSN 107), missing TSN 107 has blocked both streams.  If more
> > streams were involved in the same way, they would all be blocked by
> > the loss of one TSN.  This *is* inter-stream HOL blocking.
>       ^^^^^^^^^^^^^^^
> This is a misstatement. As you also came to the realization below, you
> must lose >2 TSNs across both streams for the above scenario to happen
> (in your example, you assumed the loss of 3 TSNs across both streams).
> If you only lose one TSN, it is impossible to affect more than 2 streams
> and thus there is *no chance* for any inter-stream HOL blocking.

That is not true, especially for SIGTRAN where DATA chunks are quite small.
Loss of one SCTP packet could easily result in the loss of 32 TSNs at once
making interstream blocking between 33 streams quite possible and even
probably in this situation.

> 
> > The problem is not as great as I first thought (all streams
> > blocked), really it is just streams that have dropping that are
> > blocked, but the problem is still of concern: particularly as it is
> > congestion which causes drops in PR-SCTP in the first place, one can
> > expect packet losses at about the same time.
> 
> It is not totally unreasonable to let a stream suffer some msg delivery
> delay when the stream is missing a TSN in the first place.

That's not the issue, the issue is suffering a delay when a TSN has been
dropped by PR-SCTP and not as a result of packet loss.  If dropping the
packet does not boost performance from the application's perspective, why
drop packets?

> 
> > 
> > The delay introducted is blocking (in the example), delivery on
> > stream 1 and stream 2 until TSN 107 can be finally and successfully
> > delivered.
> 
> It is so because you actually lost TSN 105, 106, and 107 to start the
> scenario and each of the 2 stream lost at least one TSN, otherwise it
> would be affected.

The example for 33 streams is a valid one as well.  It is actually quite
probably that packet loss will occur around the same time as dropping,
because congestion leads to both loss and dropping under "timed reliability".

> 
> > 
> > If FORWARD TSN included the fact that, say, stream 1 should "expect"
> > SSN 5 and stream 2 should "expect" SSN 60 (because 4 and 59 have
> > been dropped), then there would be no blocking.
> 
> Yes, it will help if we include SSN info in the Fwd-TSN, but the
> question is how to encode it. Your suggestion here does not scale well
> because, say, if we have 10k outbound streams, how can you encode the
> "expected" SSN for each one of them in a Fwd-TSN?

It is only necessary to code the "Expected SSN" for streams which have
dropped TSNs.  Consider it as a Forward-SSN.  It should never be necessary
to code 10k streams but could be treated as truncation on GAP reports
nevertheless.

> 
> An alternative might be to enlist in the Fwd-TSN each "dropped" chunk
> with its TSN/SID/SSN triplet. But this means that we will still "burn" 8
> bytes for each dropped data chunk, while one of the primary ideas of not
> to re-send expired data is to save network bandwidth...

Sending Forward-SSN would only take about 8 bytes and would be included
no more than once per stream dropping for no more than 8 bytes per dropped
packet.  The DATA chunk header is 16 bytes, so you will be saving quite a
few bytes.

> 
> Maybe this info can be provided as optional, i.e., the Fwd-TSN will
> always indicate the new cumTSN position, and also OPTIONALLY provide a
> list of TSN/SID/SSN triplets to help the receiver message delivery. This
> can even be a partial list if too many TSNs are dropped/skipped since it
> will still help the receiver somewhat...

Sending a Forward-SSN for only those streams that drop ordered packets
within the cumm-ack to Forward-TSN range would require far less data
in the FORWARD TSN, yet provide only what is required to deliver around
drops, takes 4 bytes less data than your triplet approach, and could still
be truncated when required.  I would say MUST, because this is as important
to SCTP's ability to avoid this interstream blocking as are GAP reports.

--brian

> 
> -Qiaobing
> 
> > 
> > In general I don't see a problem with putting the SID/SSN in the
> > FORWARD TSN.  Certainly the sizeof of the FOWARD TSN chunk could be
> > limited in similar ways to SACK chunks (advisory only), then the
> > approach appears to be a sound one.
> > 
> > --brian
> >
> 
> _______________________________________________
> tsvwg mailing list
> tsvwg@ietf.org
> https://www1.ietf.org/mailman/listinfo/tsvwg

-- 
Brian F. G. Bidulock    ¦ The reasonable man adapts himself to the ¦
bidulock@openss7.org    ¦ world; the unreasonable one persists in  ¦
http://www.openss7.org/ ¦ trying  to adapt the  world  to himself. ¦
                        ¦ Therefore  all  progress  depends on the ¦
                        ¦ unreasonable man. -- George Bernard Shaw ¦

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



From daemon@optimus.ietf.org  Mon Jun  3 17:07:29 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 RAA01020
	for <tsvwg-archive@odin.ietf.org>; Mon, 3 Jun 2002 17:07:29 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id RAA00519
	for tsvwg-archive@odin.ietf.org; Mon, 3 Jun 2002 17:07:53 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id QAA27344;
	Mon, 3 Jun 2002 16:19:51 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id QAA27313
	for <tsvwg@optimus.ietf.org>; Mon, 3 Jun 2002 16:19:49 -0400 (EDT)
Received: from mail.eecis.udel.edu (louie.udel.edu [128.4.40.12])
	by ietf.org (8.9.1a/8.9.1a) with SMTP id QAA29806
	for <tsvwg@ietf.org>; Mon, 3 Jun 2002 16:19:15 -0400 (EDT)
Received: from ren.eecis.udel.edu by mail.eecis.udel.edu id aa14381;
          3 Jun 2002 16:19 EDT
Date: Mon, 3 Jun 2002 16:19:01 -0400 (EDT)
From: "Armando L. Caro Jr." <acaro@mail.eecis.udel.edu>
To: "Brian F. G. Bidulock" <bidulock@openss7.org>
cc: Qiaobing Xie <qxie1@email.mot.com>,
        Qiaobing Xie <Qiaobing.Xie@motorola.com>, tsvwg@ietf.org
MMDF-Warning:  Parse error in original version of preceding line at mail.eecis.udel.edu
Subject: Re: [Tsvwg] More on PRSCTP-00
In-Reply-To: <20020603144342.A30667@openss7.org>
Message-ID: <Pine.GSO.4.33.0206031603240.23103-100000@ren.eecis.udel.edu>
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

On Mon, 3 Jun 2002, Brian F. G. Bidulock wrote:

> Sending a Forward-SSN for only those streams that drop ordered packets
> within the cumm-ack to Forward-TSN range would require far less data
> in the FORWARD TSN, yet provide only what is required to deliver around
> drops, takes 4 bytes less data than your triplet approach, and could still
> be truncated when required.

So you are proposing a new forward tsn chunk like the following?

         0                   1                   2                   3
         0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
        +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
        |1 1 0 0 0 0 0 0|  Chunk Flags  |0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0|
        +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
        |                      New Cumulative TSN                       |
        +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
        |      Num of Forward SSNs = N  |         ??                    |
        +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
        |      Stream Identifier S      |   Stream Sequence Number n    |
        +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
        /                                                               /
        \                              ...                              \
        /                                                               /
        +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
        |      Stream Identifier S      |   Stream Sequence Number n    |
        +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

> I would say MUST, because this is as important to SCTP's ability to
> avoid this interstream blocking as are GAP reports.

I think MUST is good, otherwise PR-SCTP may not provided much better
performance than plain SCTP.

~armando

0--                                                --0
| Armando L. Caro Jr.     |       acaro@cis.udel.edu |
| University of Delaware  |  www.cis.udel.edu/~acaro |
0--                                                --0



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



From daemon@optimus.ietf.org  Mon Jun  3 17:10:20 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 RAA01117
	for <tsvwg-archive@odin.ietf.org>; Mon, 3 Jun 2002 17:10:20 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id RAA00649
	for tsvwg-archive@odin.ietf.org; Mon, 3 Jun 2002 17:10:48 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id QAA28712;
	Mon, 3 Jun 2002 16:40:52 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id QAA28685
	for <tsvwg@optimus.ietf.org>; Mon, 3 Jun 2002 16:40:50 -0400 (EDT)
Received: from mail.eecis.udel.edu (louie.udel.edu [128.4.40.12])
	by ietf.org (8.9.1a/8.9.1a) with SMTP id QAA00336
	for <tsvwg@ietf.org>; Mon, 3 Jun 2002 16:40:21 -0400 (EDT)
Received: from ren.eecis.udel.edu by mail.eecis.udel.edu id aa19113;
          3 Jun 2002 16:40 EDT
Date: Mon, 3 Jun 2002 16:40:27 -0400 (EDT)
From: "Armando L. Caro Jr." <acaro@mail.eecis.udel.edu>
To: "Brian F. G. Bidulock" <bidulock@openss7.org>
cc: tsvwg@ietf.org, Randall Stewart <randall@stewart.chicago.il.us>
MMDF-Warning:  Parse error in original version of preceding line at mail.eecis.udel.edu
Subject: Re: [Tsvwg] SCTP IG (SACKs)
In-Reply-To: <20020601152128.B28346@openss7.org>
Message-ID: <Pine.GSO.4.33.0206031620140.23103-100000@ren.eecis.udel.edu>
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

On Sat, 1 Jun 2002, Brian F. G. Bidulock wrote:

> draft-ietf-tsvwg-sctpimguide-05.txt Sction 2.7 places a MUST on
                                      ^^^^^^^^^^
you mean section 2.8

> sending gap reports.  Before when this was a SHOULD, it was obvious
> that gap ack blocks in a SACK which exceeded an MTU could be
> truncated.  Now it is not so clear.  What should an implementation
> do if the number of gap ack blocks in a SACK exceeds the path MTU?
> Truncate?  Force IP fragmentation?

I think what is desired is that SACKs SHOULD send gap reports, where the
only reason for not sending a gap report is due to truncation. So what
about the following wording...

"Any received DATA chunks with TSN greater than the value in the
Cumulative TSN Ack field are reported in the Gap Ack Block fields. The
SCTP endpoint MUST report as many Gap Ack Blocks that can fit in a single
SACK chunk limited by the current path MTU."

~armando

0--                                                --0
| Armando L. Caro Jr.     |       acaro@cis.udel.edu |
| University of Delaware  |  www.cis.udel.edu/~acaro |
0--                                                --0


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



From daemon@optimus.ietf.org  Mon Jun  3 17:13:53 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 RAA01227
	for <tsvwg-archive@odin.ietf.org>; Mon, 3 Jun 2002 17:13:53 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id RAA00826
	for tsvwg-archive@odin.ietf.org; Mon, 3 Jun 2002 17:14:21 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id QAA28539;
	Mon, 3 Jun 2002 16:38:51 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id QAA28510
	for <tsvwg@optimus.ietf.org>; Mon, 3 Jun 2002 16:38:49 -0400 (EDT)
Received: from bfgbhome.inetint.com (IDENT:root@tnt-dal-42-248.dallas.net [209.44.42.248])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id QAA00264
	for <tsvwg@ietf.org>; Mon, 3 Jun 2002 16:38:15 -0400 (EDT)
Received: (from brian@localhost)
	by bfgbhome.inetint.com (8.9.3/8.9.3) id PAA31418;
	Mon, 3 Jun 2002 15:38:29 -0500
Date: Mon, 3 Jun 2002 15:38:29 -0500
From: "Brian F. G. Bidulock" <bidulock@openss7.org>
To: "Armando L. Caro Jr." <acaro@mail.eecis.udel.edu>
Cc: Qiaobing Xie <qxie1@email.mot.com>,
        Qiaobing Xie <Qiaobing.Xie@motorola.com>, tsvwg@ietf.org
Subject: Re: [Tsvwg] More on PRSCTP-00
Message-ID: <20020603153829.C30667@openss7.org>
Reply-To: bidulock@openss7.org
References: <20020603144342.A30667@openss7.org> <Pine.GSO.4.33.0206031603240.23103-100000@ren.eecis.udel.edu>
Mime-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
User-Agent: Mutt/1.2i
In-Reply-To: <Pine.GSO.4.33.0206031603240.23103-100000@ren.eecis.udel.edu>; from acaro@mail.eecis.udel.edu on Mon, Jun 03, 2002 at 04:19:01PM -0400
Organization: http://www.openss7.org/
Dsn-Notification-To: <bidulock@openss7.org>
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

Armando,

Yes!  Maybe the "??" field could be used to indicate the total
number of dropped TSNs (of course maxing out at 0xffff) reported
by this FORWARD TSN chunk.  The receiver might be able to use that
to determine whether it has accounted for all drops with the
provided Forward SSNs.

The procedure would be upon receipt of a FORWARD TSN chunk to perform
the cummulative ack adjustments already provided in the draft, however,
the next expected stream sequence number for each inbound stream would
be adjusted forward to the SSN contained in the Forward SSNs.

--brian

On Mon, 03 Jun 2002, Armando L. Caro Jr. wrote:

> On Mon, 3 Jun 2002, Brian F. G. Bidulock wrote:
> 
> > Sending a Forward-SSN for only those streams that drop ordered packets
> > within the cumm-ack to Forward-TSN range would require far less data
> > in the FORWARD TSN, yet provide only what is required to deliver around
> > drops, takes 4 bytes less data than your triplet approach, and could still
> > be truncated when required.
> 
> So you are proposing a new forward tsn chunk like the following?
> 
>          0                   1                   2                   3
>          0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
>         +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
>         |1 1 0 0 0 0 0 0|  Chunk Flags  |0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0|
>         +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
>         |                      New Cumulative TSN                       |
>         +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
>         |      Num of Forward SSNs = N  |         ??                    |
>         +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
>         |      Stream Identifier S      |   Stream Sequence Number n    |
>         +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
>         /                                                               /
>         \                              ...                              \
>         /                                                               /
>         +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
>         |      Stream Identifier S      |   Stream Sequence Number n    |
>         +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
> 
> > I would say MUST, because this is as important to SCTP's ability to
> > avoid this interstream blocking as are GAP reports.
> 
> I think MUST is good, otherwise PR-SCTP may not provided much better
> performance than plain SCTP.
> 
> ~armando
> 
> 0--                                                --0
> | Armando L. Caro Jr.     |       acaro@cis.udel.edu |
> | University of Delaware  |  www.cis.udel.edu/~acaro |
> 0--                                                --0
> 
> 
> 
> _______________________________________________
> tsvwg mailing list
> tsvwg@ietf.org
> https://www1.ietf.org/mailman/listinfo/tsvwg

-- 
Brian F. G. Bidulock    ¦ The reasonable man adapts himself to the ¦
bidulock@openss7.org    ¦ world; the unreasonable one persists in  ¦
http://www.openss7.org/ ¦ trying  to adapt the  world  to himself. ¦
                        ¦ Therefore  all  progress  depends on the ¦
                        ¦ unreasonable man. -- George Bernard Shaw ¦

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



From daemon@optimus.ietf.org  Mon Jun  3 17:14:51 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 RAA01290
	for <tsvwg-archive@odin.ietf.org>; Mon, 3 Jun 2002 17:14:50 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id RAA00890
	for tsvwg-archive@odin.ietf.org; Mon, 3 Jun 2002 17:15:19 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id QAA28868;
	Mon, 3 Jun 2002 16:45:01 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id QAA28838
	for <tsvwg@optimus.ietf.org>; Mon, 3 Jun 2002 16:44:58 -0400 (EDT)
Received: from mail.eecis.udel.edu (louie.udel.edu [128.4.40.12])
	by ietf.org (8.9.1a/8.9.1a) with SMTP id QAA00408
	for <tsvwg@ietf.org>; Mon, 3 Jun 2002 16:44:29 -0400 (EDT)
Received: from ren.eecis.udel.edu by mail.eecis.udel.edu id aa19987;
          3 Jun 2002 16:44 EDT
Date: Mon, 3 Jun 2002 16:44:23 -0400 (EDT)
From: "Armando L. Caro Jr." <acaro@mail.eecis.udel.edu>
To: "Brian F. G. Bidulock" <bidulock@openss7.org>
cc: Qiaobing Xie <qxie1@email.mot.com>,
        Qiaobing Xie <Qiaobing.Xie@motorola.com>, tsvwg@ietf.org
MMDF-Warning:  Parse error in original version of preceding line at mail.eecis.udel.edu
Subject: Re: [Tsvwg] More on PRSCTP-00
In-Reply-To: <20020603153829.C30667@openss7.org>
Message-ID: <Pine.GSO.4.33.0206031643230.23103-100000@ren.eecis.udel.edu>
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

On Mon, 3 Jun 2002, Brian F. G. Bidulock wrote:

> Yes!  Maybe the "??" field could be used to indicate the total
> number of dropped TSNs (of course maxing out at 0xffff) reported
> by this FORWARD TSN chunk.  The receiver might be able to use that
> to determine whether it has accounted for all drops with the
> provided Forward SSNs.

yeah, _maybe_

> The procedure would be upon receipt of a FORWARD TSN chunk to perform
> the cummulative ack adjustments already provided in the draft, however,
> the next expected stream sequence number for each inbound stream would
> be adjusted forward to the SSN contained in the Forward SSNs.

Yeah, this is the same idea we had... but yours is better in that we don't
have to report foward ssns for all streams... only those that need it.

~armando

0--                                                --0
| Armando L. Caro Jr.     |       acaro@cis.udel.edu |
| University of Delaware  |  www.cis.udel.edu/~acaro |
0--                                                --0


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



From daemon@optimus.ietf.org  Mon Jun  3 17:27:59 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 RAA01632
	for <tsvwg-archive@odin.ietf.org>; Mon, 3 Jun 2002 17:27:55 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id RAA01692
	for tsvwg-archive@odin.ietf.org; Mon, 3 Jun 2002 17:28:23 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id QAA29273;
	Mon, 3 Jun 2002 16:57:56 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id QAA29242
	for <tsvwg@optimus.ietf.org>; Mon, 3 Jun 2002 16:57:54 -0400 (EDT)
Received: from bfgbhome.inetint.com (IDENT:root@tnt-dal-42-248.dallas.net [209.44.42.248])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id QAA00745
	for <tsvwg@ietf.org>; Mon, 3 Jun 2002 16:57:22 -0400 (EDT)
Received: (from brian@localhost)
	by bfgbhome.inetint.com (8.9.3/8.9.3) id PAA31598;
	Mon, 3 Jun 2002 15:57:42 -0500
Date: Mon, 3 Jun 2002 15:57:42 -0500
From: "Brian F. G. Bidulock" <bidulock@openss7.org>
To: "Armando L. Caro Jr." <acaro@mail.eecis.udel.edu>
Cc: tsvwg@ietf.org, Randall Stewart <randall@stewart.chicago.il.us>
Subject: Re: [Tsvwg] SCTP IG (SACKs)
Message-ID: <20020603155742.D30667@openss7.org>
Reply-To: bidulock@openss7.org
References: <20020601152128.B28346@openss7.org> <Pine.GSO.4.33.0206031620140.23103-100000@ren.eecis.udel.edu>
Mime-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
User-Agent: Mutt/1.2i
In-Reply-To: <Pine.GSO.4.33.0206031620140.23103-100000@ren.eecis.udel.edu>; from acaro@mail.eecis.udel.edu on Mon, Jun 03, 2002 at 04:40:27PM -0400
Organization: http://www.openss7.org/
Dsn-Notification-To: <bidulock@openss7.org>
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

Armando,

On Mon, 03 Jun 2002, Armando L. Caro Jr. wrote:

> On Sat, 1 Jun 2002, Brian F. G. Bidulock wrote:
> 
> > draft-ietf-tsvwg-sctpimguide-05.txt Sction 2.7 places a MUST on
>                                       ^^^^^^^^^^
> you mean section 2.8
> 
> > sending gap reports.  Before when this was a SHOULD, it was obvious
> > that gap ack blocks in a SACK which exceeded an MTU could be
> > truncated.  Now it is not so clear.  What should an implementation
> > do if the number of gap ack blocks in a SACK exceeds the path MTU?
> > Truncate?  Force IP fragmentation?
> 
> I think what is desired is that SACKs SHOULD send gap reports, where the
> only reason for not sending a gap report is due to truncation. So what
> about the following wording...
> 
> "Any received DATA chunks with TSN greater than the value in the
> Cumulative TSN Ack field are reported in the Gap Ack Block fields. The
> SCTP endpoint MUST report as many Gap Ack Blocks that can fit in a single
> SACK chunk limited by the current path MTU."

That would be fine with me.  I just want to truncate on path MTU
(which is quite a few gap reports and or dups).  Perhaps it should
be mentioned that duplicate TSN reports should be truncated in
preference of reporting GAPs.

--brian

> 
> ~armando
> 
> 0--                                                --0
> | Armando L. Caro Jr.     |       acaro@cis.udel.edu |
> | University of Delaware  |  www.cis.udel.edu/~acaro |
> 0--                                                --0
> 
> 
> _______________________________________________
> tsvwg mailing list
> tsvwg@ietf.org
> https://www1.ietf.org/mailman/listinfo/tsvwg

-- 
Brian F. G. Bidulock    ¦ The reasonable man adapts himself to the ¦
bidulock@openss7.org    ¦ world; the unreasonable one persists in  ¦
http://www.openss7.org/ ¦ trying  to adapt the  world  to himself. ¦
                        ¦ Therefore  all  progress  depends on the ¦
                        ¦ unreasonable man. -- George Bernard Shaw ¦

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



From daemon@ns.ietf.org  Mon Jun  3 18:19:54 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 SAA02876
	for <tsvwg-archive@odin.ietf.org>; Mon, 3 Jun 2002 18:19:54 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id SAA04641
	for tsvwg-archive@odin.ietf.org; Mon, 3 Jun 2002 18:20:23 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id RAA03112;
	Mon, 3 Jun 2002 17:48:34 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id RAA03082
	for <tsvwg@optimus.ietf.org>; Mon, 3 Jun 2002 17:48:32 -0400 (EDT)
Received: from ilsa.franken.de (postfix@ilsa.franken.de [193.175.24.42])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id RAA02289
	for <tsvwg@ietf.org>; Mon, 3 Jun 2002 17:48:02 -0400 (EDT)
Received: from localhost (pD9E4F402.dip.t-dialin.net [217.228.244.2])
	by ilsa.franken.de (Postfix) with ESMTP
	id 8E75F245CF; Mon,  3 Jun 2002 23:48:28 +0200 (CEST)
Date: Mon, 3 Jun 2002 23:48:19 +0200
Subject: Re: [Tsvwg] The unusable addresses in <draft-stewart-tsvwg-sctp-ipv4-00.txt>
Content-Type: text/plain; charset=US-ASCII; format=flowed
Mime-Version: 1.0 (Apple Message framework v481)
Cc: tsvwg@ietf.org
To: "Daisy Chang" <daisyc@us.ibm.com>
From: Michael Tuexen <Michael.Tuexen@icn.siemens.de>
In-Reply-To: <OFAD08B8E1.06D5E58C-ON85256BC9.00534EEF@raleigh.ibm.com>
Message-Id: <9E70FD6D-773B-11D6-BAEA-0030654C1AB6@icn.siemens.de>
Content-Transfer-Encoding: 7bit
X-Mailer: Apple Mail (2.481)
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 Daisy,

see my comments below.

Best regards
Michael

On Thursday, May 30, 2002, at 05:28 PM, Daisy Chang wrote:

> Hi,
>       In <draft-stewart-tsvwg-sctp-ipv4-00.txt>, the unusable addresses 
> are
> defined as
>
>  Level 0: Addresses unusable with SCTP: 0.0.0.0/8, 224.0.0.0/4,
>       198.18.0.0/24, 192.88.99.0/24.
>
>       Questions:
>  - Should broadcast address be included as well?
Well, broadcast addresses are NOT allowed, but it is difficult to list 
them
(at least when CIDR comes into play). So I will change the sentence

    In this document the IPv4 addresses are divided into several
    different levels:

to

    In this document the unicast IPv4 addresses are divided into several
    different levels:

and omit the 224.0.0.0/4 from level 0.
>  - <draft-iana-special-ipv4-03.txt> defines addresses used for benchmark
> tests is 198.18.0.0/15. Is there any reason to change it to 
> 198.18.0.0/24
> for SCTP?
No, of course not. It is simply a typo.
>
> Thanks,
> Daisy Chang
Thank you for the comments.
> 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
>
>
Michael.Tuexen@micmac.franken.de


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



From daemon@ns.ietf.org  Mon Jun  3 21:38:57 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 VAA05614
	for <tsvwg-archive@odin.ietf.org>; Mon, 3 Jun 2002 21:38:56 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id VAA14340
	for tsvwg-archive@odin.ietf.org; Mon, 3 Jun 2002 21:39:25 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id VAA13059;
	Mon, 3 Jun 2002 21:11:43 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id VAA13033
	for <tsvwg@ns.ietf.org>; Mon, 3 Jun 2002 21:11:42 -0400 (EDT)
Received: from motgate2.mot.com (motgate2.mot.com [136.182.1.10])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id VAA05287
	for <tsvwg@ietf.org>; Mon, 3 Jun 2002 21:11:12 -0400 (EDT)
Received: [from pobox.mot.com (pobox.mot.com [129.188.137.100]) by motgate2.mot.com (motgate2 2.1) with ESMTP id SAA06605; Mon, 3 Jun 2002 18:11:42 -0700 (MST)]
Received: [from relay1.cig.mot.com (relay1.cig.mot.com [136.182.15.23]) by pobox.mot.com (MOT-pobox 2.0) with ESMTP id SAA10136; Mon, 3 Jun 2002 18:11:39 -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 g541Bbb07619;
	Mon, 3 Jun 2002 20:11:37 -0500 (CDT)
Received: from email.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 UAA07768;
	Mon, 3 Jun 2002 20:11:57 -0500 (CDT)
Message-ID: <3CFC144D.65697A41@email.mot.com>
Date: Mon, 03 Jun 2002 20:13:49 -0500
From: Qiaobing Xie <qxie1@email.mot.com>
X-Mailer: Mozilla 4.77 [en] (Win98; U)
X-Accept-Language: en
MIME-Version: 1.0
To: bidulock@openss7.org
CC: Qiaobing Xie <Qiaobing.Xie@motorola.com>,
        "Armando L. Caro Jr." <acaro@mail.eecis.udel.edu>, tsvwg@ietf.org
Subject: Re: [Tsvwg] More on PRSCTP-00
References: <Pine.GSO.4.33.0205311736060.8809-100000@ren.eecis.udel.edu> <3CF80F65.58E43813@email.mot.com> <20020531195333.D11468@openss7.org> <3CFBC11C.1EFA747@email.mot.com> <20020603144342.A30667@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

.....
> > > missing TSN 107), missing TSN 107 has blocked both streams.  If more
> > > streams were involved in the same way, they would all be blocked by
> > > the loss of one TSN.  This *is* inter-stream HOL blocking.
> >       ^^^^^^^^^^^^^^^
> > This is a misstatement. As you also came to the realization below, you
> > must lose >2 TSNs across both streams for the above scenario to happen
> > (in your example, you assumed the loss of 3 TSNs across both streams).
> > If you only lose one TSN, it is impossible to affect more than 2 streams
> > and thus there is *no chance* for any inter-stream HOL blocking.
> 
> That is not true, 
   ^^^^^^^^^^^^^^

Brian,

I am sorry but IT IS TRUE! I am merely pointing out the fact that if you
lose just one TSN then there won't be an inter-stream HOL blocking as
you wrongly stated. Do you agree with this point?

> especially for SIGTRAN where DATA chunks are quite small.
> Loss of one SCTP packet could easily result in the loss of 32 TSNs at once
> making interstream blocking between 33 streams quite possible and even
> probably in this situation.

This is an entirely different point than what I was talking about. Yes,
if the app tends to bundle a number of small messages from different
streams in a single package, then multiple streams will be
affected/delayed if you lose the packet. But this is always the case,
regardless you use PR-SCTP or not. We all know this for a long time and
PR-SCTP is not designed to address this issue.

In reality, let's say you bundle 32 TSNs from different strms in a
packet and it gets lost in the network, and at the time of retrans,
let's assume that, of the 32 TSNs, 10 have expired. With SCTP your
sender will re-send all the 32 TSNs, probably bundled into another
packet. With PR-SCTP, you will send a Fwd-TSN to skip the 10 expired
TSNs if you could, and re-send the other 22 TSNs that are still valid,
and you would very likely bundle your Fwd-TSN and the 22 TSNs into a
packet. In such a case, PR-SCTP would never cause any extra delay than
SCTP.

....
> That's not the issue, the issue is suffering a delay when a TSN has been
> dropped by PR-SCTP and not as a result of packet loss.  If dropping the
> packet does not boost performance from the application's perspective, why
> drop packets?

Also from "Armando L. Caro Jr.":
> I think MUST is good, otherwise PR-SCTP may not provided much better
> performance than plain SCTP.
> 

This is a big misunderstanding of the original intention of PR-SCTP (and
that of U-SCTP and the primitive "lifetime" feature in SCTP). We didn't
introduce this feature as a way to "boost performance from the
application's perspective". Instead, this was motivated by the question
"why we keep retransmitting expired user messages and waste the network
bandwidth?"

Just think about it, when a lost TSN(s) is detected, the only difference
between a PR-SCTP sender and a SCTP sender is that the former sends a
Fwd-TSN and the latter re-send the lost TSN(s). In either case the lost
TSN problem will take a one-way network delay to resolve (assuming that
the Fwd-TSN or the re-send will not be lost again). There is no time
advantage between the both. But there is a clear advantage for PR-SCTP
in terms of saving in the network bandwidth (however, if we keep
inflating the size of Fwd-TSN, this advantage will become diminished).

I'd agree that there may be some circumstantial advantages for PR-SCTP,
but even if they exist, they are mostly secondary and case-dependent.

In summary, I still don't see clearly how PR-SCTP introduces extra delay
to the delivery of inter-stream messages. Comparing to what SCTP does,
PR-SCTP does not do any better, nor does it do any worse in this
department. But this is ok to me, since it is mainly designed to improve
network bandwidth efficacy by not retransmitting expired user data. 

-Qiaobing

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



From daemon@ns.ietf.org  Mon Jun  3 21:48:30 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 VAA06559
	for <tsvwg-archive@odin.ietf.org>; Mon, 3 Jun 2002 21:48:30 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id VAA15268
	for tsvwg-archive@odin.ietf.org; Mon, 3 Jun 2002 21:48:59 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id VAA13513;
	Mon, 3 Jun 2002 21:26:49 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id VAA13427
	for <tsvwg@ns.ietf.org>; Mon, 3 Jun 2002 21:26:45 -0400 (EDT)
Received: from motgate2.mot.com (motgate2.mot.com [136.182.1.10])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id VAA05467
	for <tsvwg@ietf.org>; Mon, 3 Jun 2002 21:26:15 -0400 (EDT)
Received: [from pobox.mot.com (pobox.mot.com [129.188.137.100]) by motgate2.mot.com (motgate2 2.1) with ESMTP id SAA13860; Mon, 3 Jun 2002 18:26:43 -0700 (MST)]
Received: [from relay1.cig.mot.com (relay1.cig.mot.com [136.182.15.23]) by pobox.mot.com (MOT-pobox 2.0) with ESMTP id SAA13637; Mon, 3 Jun 2002 18:26:41 -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 g541Qcb08276;
	Mon, 3 Jun 2002 20:26:39 -0500 (CDT)
Received: from email.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 UAA07783;
	Mon, 3 Jun 2002 20:26:59 -0500 (CDT)
Message-ID: <3CFC17D2.25849F3C@email.mot.com>
Date: Mon, 03 Jun 2002 20:28:50 -0500
From: Qiaobing Xie <qxie1@email.mot.com>
X-Mailer: Mozilla 4.77 [en] (Win98; U)
X-Accept-Language: en
MIME-Version: 1.0
To: bidulock@openss7.org
CC: "Armando L. Caro Jr." <acaro@mail.eecis.udel.edu>, tsvwg@ietf.org,
        Randall Stewart <randall@stewart.chicago.il.us>
Subject: Re: [Tsvwg] SCTP IG (SACKs)
References: <20020601152128.B28346@openss7.org> <Pine.GSO.4.33.0206031620140.23103-100000@ren.eecis.udel.edu> <20020603155742.D30667@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

...
> Perhaps it should
> be mentioned that duplicate TSN reports should be truncated in
> preference of reporting GAPs.

I think this is a good point.

-Qiaobing

> 
> --brian
>

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



From daemon@ns.ietf.org  Mon Jun  3 23:24:47 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 XAA08230
	for <tsvwg-archive@odin.ietf.org>; Mon, 3 Jun 2002 23:24:46 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id XAA19434
	for tsvwg-archive@odin.ietf.org; Mon, 3 Jun 2002 23:25:14 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id XAA18672;
	Mon, 3 Jun 2002 23:01:28 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id XAA18642
	for <tsvwg@ns.ietf.org>; Mon, 3 Jun 2002 23:01:26 -0400 (EDT)
Received: from bfgbhome.inetint.com (IDENT:root@tnt-dal-42-248.dallas.net [209.44.42.248])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id XAA07920
	for <tsvwg@ietf.org>; Mon, 3 Jun 2002 23:00:55 -0400 (EDT)
Received: (from brian@localhost)
	by bfgbhome.inetint.com (8.9.3/8.9.3) id VAA02409;
	Mon, 3 Jun 2002 21:59:52 -0500
Date: Mon, 3 Jun 2002 21:59:52 -0500
From: "Brian F. G. Bidulock" <bidulock@openss7.org>
To: Qiaobing Xie <qxie1@email.mot.com>
Cc: Qiaobing Xie <Qiaobing.Xie@motorola.com>,
        "Armando L. Caro Jr." <acaro@mail.eecis.udel.edu>, tsvwg@ietf.org
Subject: Re: [Tsvwg] More on PRSCTP-00
Message-ID: <20020603215952.E30667@openss7.org>
Reply-To: bidulock@openss7.org
References: <Pine.GSO.4.33.0205311736060.8809-100000@ren.eecis.udel.edu> <3CF80F65.58E43813@email.mot.com> <20020531195333.D11468@openss7.org> <3CFBC11C.1EFA747@email.mot.com> <20020603144342.A30667@openss7.org> <3CFC144D.65697A41@email.mot.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
User-Agent: Mutt/1.2i
In-Reply-To: <3CFC144D.65697A41@email.mot.com>; from qxie1@email.mot.com on Mon, Jun 03, 2002 at 08:13:49PM -0500
Organization: http://www.openss7.org/
Dsn-Notification-To: <bidulock@openss7.org>
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

Qiaobing,

On Mon, 03 Jun 2002, Qiaobing Xie wrote:

> .....
> > > > missing TSN 107), missing TSN 107 has blocked both streams.  If more
> > > > streams were involved in the same way, they would all be blocked by
> > > > the loss of one TSN.  This *is* inter-stream HOL blocking.
> > >       ^^^^^^^^^^^^^^^
> > > This is a misstatement. As you also came to the realization below, you
> > > must lose >2 TSNs across both streams for the above scenario to happen
> > > (in your example, you assumed the loss of 3 TSNs across both streams).
> > > If you only lose one TSN, it is impossible to affect more than 2 streams
> > > and thus there is *no chance* for any inter-stream HOL blocking.
> > 
> > That is not true, 
>    ^^^^^^^^^^^^^^
> 
> Brian,
> 
> I am sorry but IT IS TRUE! I am merely pointing out the fact that if you
> lose just one TSN then there won't be an inter-stream HOL blocking as
> you wrongly stated. Do you agree with this point?

No I do not agree: consider the receiver's view of the following:

  <--- TSN 101  SID 1  SSN 1
  <--- TSN 102  SID 2  SSN 1
  <--- TSN 103  SID 3  SSN 1
  <--- TSN 104  SID 4  SSN 1
  <--- TSN 105  SID 5  SSN 1
  <--- TSN 106  SID 6  SSN 1
  <--- TSN 107  SID 7  SSN 1
  <--- TSN 108  SID 8  SSN 1
  <--- TSN 109  SID 9  SSN 1

  <--- FOWARD TSN = 118 (9 TSNs dropped)

  <--- TSN 120  SID 1  SSN 3
  <--- TSN 121  SID 2  SSN 3
  <--- TSN 122  SID 3  SSN 3
  <--- TSN 123  SID 4  SSN 3
  <--- TSN 124  SID 5  SSN 3
  <--- TSN 125  SID 6  SSN 3
  <--- TSN 126  SID 7  SSN 3
  <--- TSN 127  SID 8  SSN 3
  <--- TSN 128  SID 9  SSN 3

Here *loss* of a single TSN has blocked 9 streams.  The receiver
cannot know if TSN 119 contains SSN 2 for any of the 9 streams.
None of the SSN 3's for any of the nine streams can be delivered
until TSN 119 has been successfully retransmitted.  This is a severe
blocking situation.

If you consider the use of a "Forward SSN" then you have something
like:

  <--- TSN 101  SID 1  SSN 1
  <--- TSN 102  SID 2  SSN 1
  <--- TSN 103  SID 3  SSN 1
  <--- TSN 104  SID 4  SSN 1
  <--- TSN 105  SID 5  SSN 1
  <--- TSN 106  SID 6  SSN 1
  <--- TSN 107  SID 7  SSN 1
  <--- TSN 108  SID 8  SSN 1
  <--- TSN 109  SID 9  SSN 1

  <--- FOWARD TSN = 118 (9 TSNs dropped)
       SID 1 Forward SSN = 3
       SID 2 Forward SSN = 3
       SID 3 Forward SSN = 3
       SID 4 Forward SSN = 3
       SID 5 Forward SSN = 3
       SID 6 Forward SSN = 3
       SID 7 Forward SSN = 3
       SID 8 Forward SSN = 3

  <--- TSN 120  SID 1  SSN 3
  <--- TSN 121  SID 2  SSN 3
  <--- TSN 122  SID 3  SSN 3
  <--- TSN 123  SID 4  SSN 3
  <--- TSN 124  SID 5  SSN 3
  <--- TSN 125  SID 6  SSN 3
  <--- TSN 126  SID 7  SSN 3
  <--- TSN 127  SID 8  SSN 3
  <--- TSN 128  SID 9  SSN 3

Then it is obvious to the receiver that SSN 3 can be delivered
around the lost packet for SIDs 1 thru 8, but that SSN 3 for SID 9
must await retransmission of TSN 119.  (Because in the example it
must contain SID 9 SSN 2).

4 bytes per streams which has dropped packets in the FOWARD TSN is
then no more consuptive of packet payload than GAP reports in SACKs.

 
> > especially for SIGTRAN where DATA chunks are quite small.
> > Loss of one SCTP packet could easily result in the loss of 32 TSNs at once
> > making interstream blocking between 33 streams quite possible and even
> > probably in this situation.
> 
> This is an entirely different point than what I was talking about. Yes,
> if the app tends to bundle a number of small messages from different
> streams in a single package, then multiple streams will be
> affected/delayed if you lose the packet. But this is always the case,
> regardless you use PR-SCTP or not. We all know this for a long time and
> PR-SCTP is not designed to address this issue.

There is nothing in SCTP RFC 2960 operation which keeps a non
PR-SCTP implementation from delivering SSNs around lost TSNs
regardless of user data size or bundling.  In th above case,
_except_for_the_effect_of_dropped_packets_in_PR_SCTP_ SCTP would
normally delivery the next sequential SSN in the stream around
missing TSNs.  So what you way above does not ring true to me.

> In reality, let's say you bundle 32 TSNs from different strms in a
> packet and it gets lost in the network, and at the time of retrans,
> let's assume that, of the 32 TSNs, 10 have expired. With SCTP your
> sender will re-send all the 32 TSNs, probably bundled into another
> packet. With PR-SCTP, you will send a Fwd-TSN to skip the 10 expired
> TSNs if you could, and re-send the other 22 TSNs that are still valid,
> and you would very likely bundle your Fwd-TSN and the 22 TSNs into a
> packet. In such a case, PR-SCTP would never cause any extra delay than
> SCTP.

Yes it would, due to the effect above.  The dropped PR-SCTP TSNs
break the normal increment by one rule of the SSN that regular SCTP
uses to determine when TSNs from one stream can be delivered around
TSNs lost for other streams.  It is only because regular SCTP
implicitly knows that the next expected SSN in an ordered stream is
one greater than the last received SSN which allows SCTP to do this.
Regardless of lost TSNs, I know that I have the last received SSN+1
in  an ordered stream that it can be delivered regardless of missing
(lost or reordered) TSNs.  When a receiver receives a FORWARD TSN,
it breaks this rule that SSNs must increase by one, temporarily
destroying regular SCTP's ability to avoid HOL blocking by
delivering around lost TSNs.  I'm sorry but PR-SCTP degrades normal
SCTP performance in this regard.  Placing a Forward SSN in the
FORWARD TSN chunk for those streams which have had SSNs dropped,
restores this relationship.

> 
> ....
> > That's not the issue, the issue is suffering a delay when a TSN has been
> > dropped by PR-SCTP and not as a result of packet loss.  If dropping the
> > packet does not boost performance from the application's perspective, why
> > drop packets?
> 
> Also from "Armando L. Caro Jr.":
> > I think MUST is good, otherwise PR-SCTP may not provided much better
> > performance than plain SCTP.
> > 
> 
> This is a big misunderstanding of the original intention of PR-SCTP (and
> that of U-SCTP and the primitive "lifetime" feature in SCTP). We didn't
> introduce this feature as a way to "boost performance from the
> application's perspective". Instead, this was motivated by the question
> "why we keep retransmitting expired user messages and waste the network
> bandwidth?"

But please not at the expense of reintroducing HOL blocking and
making the application perform worse under PR-SCTP than under
regular SCTP.  There is no need to do this at the application's
unnecessary expense.

> 
> Just think about it, when a lost TSN(s) is detected, the only difference
> between a PR-SCTP sender and a SCTP sender is that the former sends a
> Fwd-TSN and the latter re-send the lost TSN(s). In either case the lost
> TSN problem will take a one-way network delay to resolve (assuming that
> the Fwd-TSN or the re-send will not be lost again). There is no time
> advantage between the both. But there is a clear advantage for PR-SCTP
> in terms of saving in the network bandwidth (however, if we keep
> inflating the size of Fwd-TSN, this advantage will become diminished).

The saving in the network might be in terms of TSNs, but the
degradation to the application is because of the poor treatment of
SSNs: the abrogation of the rule that SSNs must increase by one.

> I'd agree that there may be some circumstantial advantages for PR-SCTP,
> but even if they exist, they are mostly secondary and case-dependent.

This is true for GAP reports as well.  I think that experience with
GAP reports has shown that the implementation can quite efficiently
make use of GAP reports to save network resources while still
permitting the impelmentation to avoid HOL blocking.  The same
should easily be true for Forward SSNs, they permit the application
to continue to avoid HOL blocking will allowing the network
efficiency of dropping stale packets rather than retransmitting
them.

> In summary, I still don't see clearly how PR-SCTP introduces extra delay
> to the delivery of inter-stream messages. Comparing to what SCTP does,
> PR-SCTP does not do any better, nor does it do any worse in this
> department. But this is ok to me, since it is mainly designed to improve
> network bandwidth efficacy by not retransmitting expired user data. 

Take a really close look at the example above.  Consider where
65,536 streams have just each dropped the next ordered packet because
of a timer expiry, and one TSN is subsequently lost.  All 65,536
streams are blocked from delivering the next TSN to the ULP until the
single (yes one) lost TSN is retransmitted.

Again, consider that this behavior is not a corner case: the same
congestion conditions which cause "timed reliability" dropping also
cause packet loss in non-ECN networks.  The congestion which caused
a the packet drops ("timed reliabilty" expiry) is apt to drop some
next packet due to the same congestion.

--brian

-- 
Brian F. G. Bidulock    ¦ The reasonable man adapts himself to the ¦
bidulock@openss7.org    ¦ world; the unreasonable one persists in  ¦
http://www.openss7.org/ ¦ trying  to adapt the  world  to himself. ¦
                        ¦ Therefore  all  progress  depends on the ¦
                        ¦ unreasonable man. -- George Bernard Shaw ¦

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



From daemon@ns.ietf.org  Tue Jun  4 20:55:47 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 UAA20856
	for <tsvwg-archive@odin.ietf.org>; Tue, 4 Jun 2002 20:55:42 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id UAA03988
	for tsvwg-archive@odin.ietf.org; Tue, 4 Jun 2002 20:56:07 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id UAA01791;
	Tue, 4 Jun 2002 20:12:09 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id UAA01760
	for <tsvwg@ns.ietf.org>; Tue, 4 Jun 2002 20:12:06 -0400 (EDT)
Received: from motgate3.mot.com (motgate3.mot.com [144.189.100.103])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id UAA20170
	for <tsvwg@ietf.org>; Tue, 4 Jun 2002 20:11:35 -0400 (EDT)
Received: [from pobox.mot.com (pobox.mot.com [129.188.137.100]) by motgate3.mot.com (motgate3 2.1) with ESMTP id RAA19071 for <tsvwg@ietf.org>; Tue, 4 Jun 2002 17:11:47 -0700 (MST)]
Received: [from relay1.cig.mot.com (relay1.cig.mot.com [136.182.15.23]) by pobox.mot.com (MOT-pobox 2.0) with ESMTP id RAA06906 for <tsvwg@ietf.org>; Tue, 4 Jun 2002 17:12:03 -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 g550Bqb25000;
	Tue, 4 Jun 2002 19:11:56 -0500 (CDT)
Received: from email.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 TAA13851;
	Tue, 4 Jun 2002 19:12:13 -0500 (CDT)
Message-ID: <3CFD57D8.5637E431@email.mot.com>
Date: Tue, 04 Jun 2002 19:14:16 -0500
From: Qiaobing Xie <qxie1@email.mot.com>
X-Mailer: Mozilla 4.77 [en] (Win98; U)
X-Accept-Language: en
MIME-Version: 1.0
To: bidulock@openss7.org
CC: Qiaobing Xie <Qiaobing.Xie@motorola.com>,
        "Armando L. Caro Jr." <acaro@mail.eecis.udel.edu>, tsvwg@ietf.org
Subject: Re: [Tsvwg] More on PRSCTP-00
References: <Pine.GSO.4.33.0205311736060.8809-100000@ren.eecis.udel.edu> <3CF80F65.58E43813@email.mot.com> <20020531195333.D11468@openss7.org> <3CFBC11C.1EFA747@email.mot.com> <20020603144342.A30667@openss7.org> <3CFC144D.65697A41@email.mot.com> <20020603215952.E30667@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

....
> Take a really close look at the example above.  Consider where
> 65,536 streams have just each dropped the next ordered packet because
> of a timer expiry, and one TSN is subsequently lost.  All 65,536
> streams are blocked from delivering the next TSN to the ULP until the
> single (yes one) lost TSN is retransmitted.

Brian,

Huh, I see now why we are having problem understand each other's
points... we are using different terminology:

When I said "dropped TSNs", I spoke from the receiver's perspective and
referred to TSNs the receiver sees as missing in the TSN space. While
you use "dropped TSNs" to refer to those that are assigned to data
chunks but never sent out (and you use "lost TSNs" to refer to those
that are put on the wire and then lost in the network).

With this clarified, I'd say your above example is more than a corner
case - it is actually very strange. It basically shows that your data
sender assigns 65535 TSNs (one for each stream) to messages in the
sender's outqueue and then changes its mind and never put those 65535
TSNs on the wire.

I'd say that to a certain degree this is equivalent to that the sender
intentionally create the 65535 holes in the TSN space (but somehow still
expects the receiver to perform). This example in fact illustrates the
importance of the note in RFC2960 that the sender should always try to
postpone the assignment of TSNs to the last moment. A good
implementation should expire messages that have never been transmitted
without burning holes in the TSN space. Fixing holes in the receiver's
TSN space (with either plain retransmission of base SCTP or Fwd-TSN of
PR-SCTP) will always cost you. One would think the sender should always
try its best not to create holes in the TSN space. 

-Qiaobing

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



From daemon@ns.ietf.org  Tue Jun  4 21:22:58 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 VAA21277
	for <tsvwg-archive@odin.ietf.org>; Tue, 4 Jun 2002 21:22:58 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id VAA05409
	for tsvwg-archive@odin.ietf.org; Tue, 4 Jun 2002 21:23:23 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id UAA03748;
	Tue, 4 Jun 2002 20:50:43 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id UAA03723
	for <tsvwg@ns.ietf.org>; Tue, 4 Jun 2002 20:50:42 -0400 (EDT)
Received: from bfgbhome.inetint.com (IDENT:root@tnt-dal-42-248.dallas.net [209.44.42.248])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id UAA20753
	for <tsvwg@ietf.org>; Tue, 4 Jun 2002 20:50:10 -0400 (EDT)
Received: (from brian@localhost)
	by bfgbhome.inetint.com (8.9.3/8.9.3) id TAA16353;
	Tue, 4 Jun 2002 19:50:20 -0500
Date: Tue, 4 Jun 2002 19:50:20 -0500
From: "Brian F. G. Bidulock" <bidulock@openss7.org>
To: Qiaobing Xie <qxie1@email.mot.com>
Cc: Qiaobing Xie <Qiaobing.Xie@motorola.com>,
        "Armando L. Caro Jr." <acaro@mail.eecis.udel.edu>, tsvwg@ietf.org
Subject: Re: [Tsvwg] More on PRSCTP-00
Message-ID: <20020604195020.A16020@openss7.org>
Reply-To: bidulock@openss7.org
References: <Pine.GSO.4.33.0205311736060.8809-100000@ren.eecis.udel.edu> <3CF80F65.58E43813@email.mot.com> <20020531195333.D11468@openss7.org> <3CFBC11C.1EFA747@email.mot.com> <20020603144342.A30667@openss7.org> <3CFC144D.65697A41@email.mot.com> <20020603215952.E30667@openss7.org> <3CFD57D8.5637E431@email.mot.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
User-Agent: Mutt/1.2i
In-Reply-To: <3CFD57D8.5637E431@email.mot.com>; from qxie1@email.mot.com on Tue, Jun 04, 2002 at 07:14:16PM -0500
Organization: http://www.openss7.org/
Dsn-Notification-To: <bidulock@openss7.org>
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

Qiaobing,

On Tue, 04 Jun 2002, Qiaobing Xie wrote:

> ....
> > Take a really close look at the example above.  Consider where
> > 65,536 streams have just each dropped the next ordered packet because
> > of a timer expiry, and one TSN is subsequently lost.  All 65,536
> > streams are blocked from delivering the next TSN to the ULP until the
> > single (yes one) lost TSN is retransmitted.
> 
> Brian,
> 
> Huh, I see now why we are having problem understand each other's
> points... we are using different terminology:
> 
> When I said "dropped TSNs", I spoke from the receiver's perspective and
> referred to TSNs the receiver sees as missing in the TSN space. While
> you use "dropped TSNs" to refer to those that are assigned to data
> chunks but never sent out (and you use "lost TSNs" to refer to those
> that are put on the wire and then lost in the network).

Yes.  PR-SCTP talks of TSNs being "dropped".  I use the terminology
from the PR-SCTP draft when referring to "dropped" TSNs.

> With this clarified, I'd say your above example is more than a corner
> case - it is actually very strange. It basically shows that your data
> sender assigns 65535 TSNs (one for each stream) to messages in the
> sender's outqueue and then changes its mind and never put those 65535
> TSNs on the wire.

The example is from the receiver's view.  It says nothing about whether
the sender placed the TSNs on the wire or not.  It may have, and they
may have been "lost" yet expired (timed reliability) before retransmission
resulting in the FORWARD TSN.  This is not strange, it is even highly
likely if the "timed reliability" interval is about 4 RTT.

> I'd say that to a certain degree this is equivalent to that the sender
> intentionally create the 65535 holes in the TSN space (but somehow still
> expects the receiver to perform). This example in fact illustrates the
> importance of the note in RFC2960 that the sender should always try to
> postpone the assignment of TSNs to the last moment. A good
> implementation should expire messages that have never been transmitted
> without burning holes in the TSN space. Fixing holes in the receiver's
> TSN space (with either plain retransmission of base SCTP or Fwd-TSN of
> PR-SCTP) will always cost you. One would think the sender should always
> try its best not to create holes in the TSN space. 

Sorry, but these TSNs may have been sent by the sender (meaning that
TSNs were assigned and cannot be deferred any longer), but expired before
successful retransmission.  In the example of 9 streams under times of
congestion, it is quite possible that 9 TSNs were lost in a single SCTP
packet, or even in multiple retransmission, the FORWARD TSN got thru
(because it was small), and subsequent TSNs transmitted on the stream are
also "lost".

There is no assumption of a poorly designed sender in the example.

This is a problem: a real problem, easily addressed by the Forward SSN
approach.

--brian

-- 
Brian F. G. Bidulock    ¦ The reasonable man adapts himself to the ¦
bidulock@openss7.org    ¦ world; the unreasonable one persists in  ¦
http://www.openss7.org/ ¦ trying  to adapt the  world  to himself. ¦
                        ¦ Therefore  all  progress  depends on the ¦
                        ¦ unreasonable man. -- George Bernard Shaw ¦

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



From daemon@ns.ietf.org  Wed Jun  5 16:41:37 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 QAA18508
	for <tsvwg-archive@odin.ietf.org>; Wed, 5 Jun 2002 16:41:37 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id QAA00282
	for tsvwg-archive@odin.ietf.org; Wed, 5 Jun 2002 16:42:08 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id QAA28548;
	Wed, 5 Jun 2002 16:09:43 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id QAA28523
	for <tsvwg@ns.ietf.org>; Wed, 5 Jun 2002 16:09:41 -0400 (EDT)
Received: from mail.eecis.udel.edu (louie.udel.edu [128.4.40.12])
	by ietf.org (8.9.1a/8.9.1a) with SMTP id QAA16453
	for <tsvwg@ietf.org>; Wed, 5 Jun 2002 16:09:10 -0400 (EDT)
Received: from ren.eecis.udel.edu by mail.eecis.udel.edu id aa00121;
          5 Jun 2002 16:09 EDT
Date: Wed, 5 Jun 2002 16:09:03 -0400 (EDT)
From: "Armando L. Caro Jr." <acaro@mail.eecis.udel.edu>
Reply-To: "Armando L. Caro Jr." <acaro@acm.org>
To: Randall Stewart <randall@stewart.chicago.il.us>
cc: Phil Conrad <conrad@joda.cis.temple.edu>,
        "Michael A. Ramalho" <mramalho@cisco.com>,
        Michael Tuexen <michael.tuexen@icn.siemens.de>, tsvwg@ietf.org
MMDF-Warning:  Parse error in original version of preceding line at mail.eecis.udel.edu
Subject: Re: [Tsvwg] More on PRSCTP-00
In-Reply-To: <3CFE56F9.347AC316@stewart.chicago.il.us>
Message-ID: <Pine.GSO.4.33.0206051605010.1041-100000@ren.eecis.udel.edu>
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

On Wed, 5 Jun 2002, Randall Stewart wrote:

> After watching this debate.. I think I have come to the conclusion that
> Brian is correct :)
>
> Michael R. is here... I will discuss it with him.. I want to get
> the other co-author's take on this one... If the others agree with
> Brian (and me :>) then I think we will put this in.... also I would
> like to hear what the rest of the WG feels on this???

Well, you already know I agree with this change.

~armando

0--                                                --0
| Armando L. Caro Jr.     |       acaro@cis.udel.edu |
| University of Delaware  |  www.cis.udel.edu/~acaro |
0--                                                --0


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



From daemon@ns.ietf.org  Wed Jun  5 17:00:58 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 RAA19553
	for <tsvwg-archive@odin.ietf.org>; Wed, 5 Jun 2002 17:00:57 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id RAA02095
	for tsvwg-archive@odin.ietf.org; Wed, 5 Jun 2002 17:01:28 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id QAA29162;
	Wed, 5 Jun 2002 16:24:25 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id QAA29131
	for <tsvwg@ns.ietf.org>; Wed, 5 Jun 2002 16:24:21 -0400 (EDT)
Received: from motgate2.mot.com (motgate2.mot.com [136.182.1.10])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id QAA17566
	for <tsvwg@ietf.org>; Wed, 5 Jun 2002 16:23:49 -0400 (EDT)
Received: [from mothost.mot.com (mothost.mot.com [129.188.137.101]) by motgate2.mot.com (motgate2 2.1) with ESMTP id NAA10435; Wed, 5 Jun 2002 13:24:21 -0700 (MST)]
Received: [from relay1.cig.mot.com (relay1.cig.mot.com [136.182.15.23]) by mothost.mot.com (MOT-pobox 2.0) with ESMTP id NAA13960; Wed, 5 Jun 2002 13:24:16 -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 g55KOFb04086;
	Wed, 5 Jun 2002 15:24:15 -0500 (CDT)
Received: from email.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 PAA27023;
	Wed, 5 Jun 2002 15:24:36 -0500 (CDT)
Message-ID: <3CFE7403.59FB6F4C@email.mot.com>
Date: Wed, 05 Jun 2002 15:26:43 -0500
From: Qiaobing Xie <qxie1@email.mot.com>
X-Mailer: Mozilla 4.77 [en] (Win98; U)
X-Accept-Language: en
MIME-Version: 1.0
To: bidulock@openss7.org
CC: Qiaobing Xie <Qiaobing.Xie@motorola.com>,
        "Armando L. Caro Jr." <acaro@mail.eecis.udel.edu>, tsvwg@ietf.org
Subject: Re: [Tsvwg] More on PRSCTP-00
References: <Pine.GSO.4.33.0205311736060.8809-100000@ren.eecis.udel.edu> <3CF80F65.58E43813@email.mot.com> <20020531195333.D11468@openss7.org> <3CFBC11C.1EFA747@email.mot.com> <20020603144342.A30667@openss7.org> <3CFC144D.65697A41@email.mot.com> <20020603215952.E30667@openss7.org> <3CFD57D8.5637E431@email.mot.com> <20020604195020.A16020@openss7.org>
Content-Type: text/plain; charset=iso-8859-1
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

Brian,

"Brian F. G. Bidulock" wrote:
....
> Sorry, but these TSNs may have been sent by the sender (meaning that
> TSNs were assigned and cannot be deferred any longer), but expired before
> successful retransmission.  In the example of 9 streams under times of
> congestion, it is quite possible that 9 TSNs were lost in a single SCTP
                                        ^^^^^^^^^^^^^^^^
> packet, or even in multiple retransmission the FORWARD TSN got thru
> (because it was small), and subsequent TSNs transmitted on the stream are
> also "lost".

Then, your (repeated) statement (citing one of them here):

> > Here *loss* of a single TSN has blocked 9 streams.
    ^^^^^^^^^^^^^^^^^^^^^^^^^

is untrue (or at least grossly mis-stated). 

> This is a problem: a real problem, easily addressed by the Forward SSN
> approach.

I've said at the very beginning that I am not "violently" against the
idea of adding SSN info to Fwd-TSN. All I've been trying to point out to
you is that your scenario in which multiple streams can be blocked can
only result from the loss of multiple TSNs. This is not a problem
induced by PR-SCTP, the same thing can happen with the base SCTP. 

To make myself clear:

- Is the problem as serious as you claimed to be? 
No. You overstated it. 

- Whether we will be able to alleviate this problem by adding some
complexity to PR-SCTP design?
Yes, as I've already said that adding SSN to Fwd-TSN can have benefits. 

- Whether the added complexity worth the benefit? 
Maybe or maybe not, I am a doubter but I will not violentely object it
if others want to do it.

-Qiaobing

> 
> --brian
> 
> --
> Brian F. G. Bidulock    ¦ The reasonable man adapts himself to the ¦
> bidulock@openss7.org    ¦ world; the unreasonable one persists in  ¦
> http://www.openss7.org/ ¦ trying  to adapt the  world  to himself. ¦
>                         ¦ Therefore  all  progress  depends on the ¦
>                         ¦ unreasonable man. -- George Bernard Shaw ¦

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



From daemon@ns.ietf.org  Wed Jun  5 17:07:30 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 RAA20045
	for <tsvwg-archive@odin.ietf.org>; Wed, 5 Jun 2002 17:07:25 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id RAA02444
	for tsvwg-archive@odin.ietf.org; Wed, 5 Jun 2002 17:07:56 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id QAA29366;
	Wed, 5 Jun 2002 16:28:31 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id QAA29339
	for <tsvwg@ns.ietf.org>; Wed, 5 Jun 2002 16:28:29 -0400 (EDT)
Received: from motgate3.mot.com (motgate3.mot.com [144.189.100.103])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id QAA17836
	for <tsvwg@ietf.org>; Wed, 5 Jun 2002 16:27:57 -0400 (EDT)
Received: [from pobox.mot.com (pobox.mot.com [129.188.137.100]) by motgate3.mot.com (motgate3 2.1) with ESMTP id NAA29618; Wed, 5 Jun 2002 13:28:06 -0700 (MST)]
Received: [from relay1.cig.mot.com (relay1.cig.mot.com [136.182.15.23]) by pobox.mot.com (MOT-pobox 2.0) with ESMTP id NAA24221; Wed, 5 Jun 2002 13:28:23 -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 g55KSJb04471;
	Wed, 5 Jun 2002 15:28:19 -0500 (CDT)
Received: from email.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 PAA27070;
	Wed, 5 Jun 2002 15:28:40 -0500 (CDT)
Message-ID: <3CFE74F8.147675FE@email.mot.com>
Date: Wed, 05 Jun 2002 15:30:48 -0500
From: Qiaobing Xie <qxie1@email.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: bidulock@openss7.org, Qiaobing Xie <Qiaobing.Xie@motorola.com>,
        "Armando L. Caro Jr." <acaro@mail.eecis.udel.edu>, tsvwg@ietf.org,
        Michael Tuexen <michael.tuexen@icn.siemens.de>,
        "Michael A. Ramalho" <mramalho@cisco.com>,
        Phil Conrad <conrad@joda.cis.temple.edu>
Subject: Re: [Tsvwg] More on PRSCTP-00
References: <Pine.GSO.4.33.0205311736060.8809-100000@ren.eecis.udel.edu> <3CF80F65.58E43813@email.mot.com> <20020531195333.D11468@openss7.org> <3CFBC11C.1EFA747@email.mot.com> <20020603144342.A30667@openss7.org> <3CFC144D.65697A41@email.mot.com> <20020603215952.E30667@openss7.org> <3CFD57D8.5637E431@email.mot.com> <20020604195020.A16020@openss7.org> <3CFE4001.BCE81C14@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

Hi, Randy,

Pls see my latest reply to Brian...

-Qiaobing

Randall Stewart wrote:
> 
> Brian/Qiaobing:
> 
> I have read through this interesting debate. Qiaobing and
> I had a major discussion on this way back when it was
> considered U-SCTP. Renee Revis first pointed this out to me.. but
> from a different persepective i.e. the receiver being able to
> efficently process stuck TSN's ...
> 
> After a lengthy debate I agreed to not have the SSN/SID listed
> in the FWD TSN... I did NOT think of the HOL blocking issue.. or
> I would have used it as a issue (I was the one arguing for the
> SSN/SID in the FWD-TSN :->).
> 
> After watching this debate.. I think I have come to the conclusion that
> Brian is correct :)
> 
> Michael R. is here... I will discuss it with him.. I want to get
> the other co-author's take on this one... If the others agree with
> Brian (and me :>) then I think we will put this in.... also I would
> like to hear what the rest of the WG feels on this???
> 
> I think it is reasonable and not that much overhead to include JUST
> the dropped streams NEXT expected SSN... You would be adding 4 bytes
> for EVERY stream dropping one or more TSN's... note that this does
> NOT mean 4 bytes per TSN but 4 bytes per where one or more TSN's were
> dropped..
> 
> I think this is reasonable.. it improves processing AND avoids HOL
> blocking..
> 
> Michael T/Phil? what do you think... Michael R.. I will talk to you at
> the
> next class break..
> 
> 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  Wed Jun  5 18:08:36 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 SAA22933
	for <tsvwg-archive@odin.ietf.org>; Wed, 5 Jun 2002 18:08:31 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id SAA05695
	for tsvwg-archive@odin.ietf.org; Wed, 5 Jun 2002 18:09:01 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id RAA03974;
	Wed, 5 Jun 2002 17:35:22 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id RAA03944
	for <tsvwg@ns.ietf.org>; Wed, 5 Jun 2002 17:35:20 -0400 (EDT)
Received: from bfgbhome.inetint.com (IDENT:root@tnt-dal-42-242.dallas.net [209.44.42.242])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id RAA21177
	for <tsvwg@ietf.org>; Wed, 5 Jun 2002 17:34:46 -0400 (EDT)
Received: (from brian@localhost)
	by bfgbhome.inetint.com (8.9.3/8.9.3) id QAA03977;
	Wed, 5 Jun 2002 16:35:02 -0500
Date: Wed, 5 Jun 2002 16:35:02 -0500
From: "Brian F. G. Bidulock" <bidulock@openss7.org>
To: Qiaobing Xie <qxie1@email.mot.com>
Cc: Qiaobing Xie <Qiaobing.Xie@motorola.com>,
        "Armando L. Caro Jr." <acaro@mail.eecis.udel.edu>, tsvwg@ietf.org
Subject: Re: [Tsvwg] More on PRSCTP-00
Message-ID: <20020605163502.A2012@openss7.org>
Reply-To: bidulock@openss7.org
References: <Pine.GSO.4.33.0205311736060.8809-100000@ren.eecis.udel.edu> <3CF80F65.58E43813@email.mot.com> <20020531195333.D11468@openss7.org> <3CFBC11C.1EFA747@email.mot.com> <20020603144342.A30667@openss7.org> <3CFC144D.65697A41@email.mot.com> <20020603215952.E30667@openss7.org> <3CFD57D8.5637E431@email.mot.com> <20020604195020.A16020@openss7.org> <3CFE7403.59FB6F4C@email.mot.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
User-Agent: Mutt/1.2i
In-Reply-To: <3CFE7403.59FB6F4C@email.mot.com>; from qxie1@email.mot.com on Wed, Jun 05, 2002 at 03:26:43PM -0500
Organization: http://www.openss7.org/
Dsn-Notification-To: <bidulock@openss7.org>
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

Qiaobing,

On Wed, 05 Jun 2002, Qiaobing Xie wrote:

> Brian,
> 
> "Brian F. G. Bidulock" wrote:
> ....
> > Sorry, but these TSNs may have been sent by the sender (meaning that
> > TSNs were assigned and cannot be deferred any longer), but expired before
> > successful retransmission.  In the example of 9 streams under times of
> > congestion, it is quite possible that 9 TSNs were lost in a single SCTP
>                                         ^^^^^^^^^^^^^^^^
> > packet, or even in multiple retransmission the FORWARD TSN got thru
> > (because it was small), and subsequent TSNs transmitted on the stream are
> > also "lost".
> 
> Then, your (repeated) statement (citing one of them here):
> 
> > > Here *loss* of a single TSN has blocked 9 streams.
>     ^^^^^^^^^^^^^^^^^^^^^^^^^
> 
> is untrue (or at least grossly mis-stated). 

Pardon, let us say loss of a single SCTP packet.
It is still HOL blocking.

--brian

> 
> > This is a problem: a real problem, easily addressed by the Forward SSN
> > approach.
> 
> I've said at the very beginning that I am not "violently" against the
> idea of adding SSN info to Fwd-TSN. All I've been trying to point out to
> you is that your scenario in which multiple streams can be blocked can
> only result from the loss of multiple TSNs. This is not a problem
> induced by PR-SCTP, the same thing can happen with the base SCTP. 
> 
> To make myself clear:
> 
> - Is the problem as serious as you claimed to be? 
> No. You overstated it. 
> 
> - Whether we will be able to alleviate this problem by adding some
> complexity to PR-SCTP design?
> Yes, as I've already said that adding SSN to Fwd-TSN can have benefits. 
> 
> - Whether the added complexity worth the benefit? 
> Maybe or maybe not, I am a doubter but I will not violentely object it
> if others want to do it.
> 
> -Qiaobing
> 
> > 
> > --brian
> > 
> > --
> > Brian F. G. Bidulock    ¦ The reasonable man adapts himself to the ¦
> > bidulock@openss7.org    ¦ world; the unreasonable one persists in  ¦
> > http://www.openss7.org/ ¦ trying  to adapt the  world  to himself. ¦
> >                         ¦ Therefore  all  progress  depends on the ¦
> >                         ¦ unreasonable man. -- George Bernard Shaw ¦

-- 
Brian F. G. Bidulock    ¦ The reasonable man adapts himself to the ¦
bidulock@openss7.org    ¦ world; the unreasonable one persists in  ¦
http://www.openss7.org/ ¦ trying  to adapt the  world  to himself. ¦
                        ¦ Therefore  all  progress  depends on the ¦
                        ¦ unreasonable man. -- George Bernard Shaw ¦

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



From daemon@ns.ietf.org  Wed Jun  5 18:26:11 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 SAA23967
	for <tsvwg-archive@odin.ietf.org>; Wed, 5 Jun 2002 18:26:11 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id SAA06168
	for tsvwg-archive@odin.ietf.org; Wed, 5 Jun 2002 18:26:41 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id RAA04765;
	Wed, 5 Jun 2002 17:58:37 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id RAA04738
	for <tsvwg@ns.ietf.org>; Wed, 5 Jun 2002 17:58:32 -0400 (EDT)
Received: from motgate3.mot.com (motgate3.mot.com [144.189.100.103])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id RAA22287
	for <tsvwg@ietf.org>; Wed, 5 Jun 2002 17:58:02 -0400 (EDT)
Received: [from pobox4.mot.com (pobox4.mot.com [10.64.251.243]) by motgate3.mot.com (motgate3 2.1) with ESMTP id OAA20582; Wed, 5 Jun 2002 14:58:06 -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 OAA03041; Wed, 5 Jun 2002 14:58:22 -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 g55LwFb10810;
	Wed, 5 Jun 2002 16:58:15 -0500 (CDT)
Received: from email.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 QAA28056;
	Wed, 5 Jun 2002 16:58:36 -0500 (CDT)
Message-ID: <3CFE8A04.24188859@email.mot.com>
Date: Wed, 05 Jun 2002 17:00:36 -0500
From: Qiaobing Xie <qxie1@email.mot.com>
X-Mailer: Mozilla 4.77 [en] (Win98; U)
X-Accept-Language: en
MIME-Version: 1.0
To: Phillip Conrad <conrad@acm.org>
CC: Randall Stewart <randall@stewart.chicago.il.us>, bidulock@openss7.org,
        Qiaobing Xie <Qiaobing.Xie@motorola.com>,
        "Armando L. Caro Jr." <acaro@mail.eecis.udel.edu>, tsvwg@ietf.org,
        Michael Tuexen <michael.tuexen@icn.siemens.de>,
        "Michael A. Ramalho" <mramalho@cisco.com>
Subject: Re: [Tsvwg] More on PRSCTP-00
References: <Pine.GSO.4.33.0205311736060.8809-100000@ren.eecis.udel.edu>
	 <3CF80F65.58E43813@email.mot.com> <20020531195333.D11468@openss7.org>
	 <3CFBC11C.1EFA747@email.mot.com> <20020603144342.A30667@openss7.org>
	 <3CFC144D.65697A41@email.mot.com> <20020603215952.E30667@openss7.org>
	 <3CFD57D8.5637E431@email.mot.com> <20020604195020.A16020@openss7.org> <5.1.0.14.2.20020605155749.0236b050@joda.cis.temple.edu>
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

Phil,

Please take my little challenges below as a way to help you making your
case tighter :-)

Phillip Conrad wrote:
...
> The next thing that happens is that TSNs 110 through 118 are "dropped" in
> the technical sense of that word; see the defn in the current draft.  This
> means that the sender gives up on sending them, and issues a forward
> TSN.   However, no Forward TSN is issued for TSN119.  This may be because
> the messages on stream 9 are treated as Reliable.   This is NOT an unusual
> situation; 

What unusual here is that you seem to imply that you send the Fwd-TSN
without retransmitting the *reliable but lost* TSN119. Remember, TSN119
is lost *together* with the other 8 TSNs, and the decision point of
sending Fwd-TSN is *deliberately* crafted to be the same as that for
normal retransmission. 

Therefore, what should be usual is that when the sender reached the
conclusion that it needs to send the Fwd-TSN for the 8 "dropped" TSNs,
it MUST also have reached the conclusion at the same moment to
retransmit TSN119 (furthermore, I'd think in more implementations the
Fwd-TSN will also be piggybacked with the retrans TSN119).

The stream blocking analysis (by Brian) only holds if the Fwd-TSN
arrives without the retransmitted TSN119. But if both arrive at the
same, would the analysis still hold?

> ...We have all the PR streams
> potentially being HOL blocked by loss of a single packet.  I think we need
> to fix this.

Interesting point. But we seem to accept the same thing (streams
potentially being HOL blocked by loss of a single packet) in base SCTP.
How come?

regards,
-Qiaobing

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



From daemon@ns.ietf.org  Wed Jun  5 18:35:41 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 SAA24615
	for <tsvwg-archive@odin.ietf.org>; Wed, 5 Jun 2002 18:35:41 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id SAA06818
	for tsvwg-archive@odin.ietf.org; Wed, 5 Jun 2002 18:36:11 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id SAA05653;
	Wed, 5 Jun 2002 18:07:53 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id SAA05622
	for <tsvwg@ns.ietf.org>; Wed, 5 Jun 2002 18:07:50 -0400 (EDT)
Received: from motgate.mot.com (motgate.mot.com [129.188.136.100])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id SAA22870
	for <tsvwg@ietf.org>; Wed, 5 Jun 2002 18:07:20 -0400 (EDT)
Received: [from pobox4.mot.com (pobox4.mot.com [10.64.251.243]) by motgate.mot.com (motgate 2.1) with ESMTP id PAA11981; Wed, 5 Jun 2002 15:07:48 -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 PAA05676; Wed, 5 Jun 2002 15:07:48 -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 g55M7ib11568;
	Wed, 5 Jun 2002 17:07:45 -0500 (CDT)
Received: from email.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 RAA28163;
	Wed, 5 Jun 2002 17:08:06 -0500 (CDT)
Message-ID: <3CFE8C3F.C935774C@email.mot.com>
Date: Wed, 05 Jun 2002 17:10:07 -0500
From: Qiaobing Xie <qxie1@email.mot.com>
X-Mailer: Mozilla 4.77 [en] (Win98; U)
X-Accept-Language: en
MIME-Version: 1.0
To: bidulock@openss7.org
CC: Qiaobing Xie <Qiaobing.Xie@motorola.com>,
        "Armando L. Caro Jr." <acaro@mail.eecis.udel.edu>, tsvwg@ietf.org
Subject: Re: [Tsvwg] More on PRSCTP-00
References: <Pine.GSO.4.33.0205311736060.8809-100000@ren.eecis.udel.edu> <3CF80F65.58E43813@email.mot.com> <20020531195333.D11468@openss7.org> <3CFBC11C.1EFA747@email.mot.com> <20020603144342.A30667@openss7.org> <3CFC144D.65697A41@email.mot.com> <20020603215952.E30667@openss7.org> <3CFD57D8.5637E431@email.mot.com> <20020604195020.A16020@openss7.org> <3CFE7403.59FB6F4C@email.mot.com> <20020605163502.A2012@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

"Brian F. G. Bidulock" wrote:
...
> Pardon, let us say loss of a single SCTP packet.
> It is still HOL blocking.

Again, what's really funny is that we suddenly start to see this HOL
blocking caused by the loss of a single SCTP packet (presumably
containing multiple TSNs from different streams) as an issue. We seem to
be able to live with it in the base SCTP for a long time.

-Qiaobing

> 
> --brian

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



From daemon@ns.ietf.org  Wed Jun  5 18:43:15 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 SAA25085
	for <tsvwg-archive@odin.ietf.org>; Wed, 5 Jun 2002 18:43:14 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id SAA07025
	for tsvwg-archive@odin.ietf.org; Wed, 5 Jun 2002 18:43:44 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id SAA06005;
	Wed, 5 Jun 2002 18:18:52 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id SAA05975
	for <tsvwg@ns.ietf.org>; Wed, 5 Jun 2002 18:18:49 -0400 (EDT)
Received: from bfgbhome.inetint.com (IDENT:root@tnt-dal-42-242.dallas.net [209.44.42.242])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id SAA23522
	for <tsvwg@ietf.org>; Wed, 5 Jun 2002 18:18:18 -0400 (EDT)
Received: (from brian@localhost)
	by bfgbhome.inetint.com (8.9.3/8.9.3) id RAA04549;
	Wed, 5 Jun 2002 17:18:26 -0500
Date: Wed, 5 Jun 2002 17:18:26 -0500
From: "Brian F. G. Bidulock" <bidulock@openss7.org>
To: Qiaobing Xie <qxie1@email.mot.com>
Cc: Phillip Conrad <conrad@acm.org>,
        Randall Stewart <randall@stewart.chicago.il.us>,
        Qiaobing Xie <Qiaobing.Xie@motorola.com>,
        "Armando L. Caro Jr." <acaro@mail.eecis.udel.edu>, tsvwg@ietf.org,
        Michael Tuexen <michael.tuexen@icn.siemens.de>,
        "Michael A. Ramalho" <mramalho@cisco.com>
Subject: Re: [Tsvwg] More on PRSCTP-00
Message-ID: <20020605171826.D2012@openss7.org>
Reply-To: bidulock@openss7.org
References: <3CF80F65.58E43813@email.mot.com> <20020531195333.D11468@openss7.org> <3CFBC11C.1EFA747@email.mot.com> <20020603144342.A30667@openss7.org> <3CFC144D.65697A41@email.mot.com> <20020603215952.E30667@openss7.org> <3CFD57D8.5637E431@email.mot.com> <20020604195020.A16020@openss7.org> <5.1.0.14.2.20020605155749.0236b050@joda.cis.temple.edu> <3CFE8A04.24188859@email.mot.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
User-Agent: Mutt/1.2i
In-Reply-To: <3CFE8A04.24188859@email.mot.com>; from qxie1@email.mot.com on Wed, Jun 05, 2002 at 05:00:36PM -0500
Organization: http://www.openss7.org/
Dsn-Notification-To: <bidulock@openss7.org>
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

Qiaobing,

On Wed, 05 Jun 2002, Qiaobing Xie wrote:

> Phil,
> 
> Please take my little challenges below as a way to help you making your
> case tighter :-)
> 
> Phillip Conrad wrote:
> ...
> > The next thing that happens is that TSNs 110 through 118 are "dropped" in
> > the technical sense of that word; see the defn in the current draft.  This
> > means that the sender gives up on sending them, and issues a forward
> > TSN.   However, no Forward TSN is issued for TSN119.  This may be because
> > the messages on stream 9 are treated as Reliable.   This is NOT an unusual
> > situation; 
> 
> What unusual here is that you seem to imply that you send the Fwd-TSN
> without retransmitting the *reliable but lost* TSN119. Remember, TSN119
> is lost *together* with the other 8 TSNs, and the decision point of
> sending Fwd-TSN is *deliberately* crafted to be the same as that for
> normal retransmission. 
> 
> Therefore, what should be usual is that when the sender reached the
> conclusion that it needs to send the Fwd-TSN for the 8 "dropped" TSNs,
> it MUST also have reached the conclusion at the same moment to
> retransmit TSN119 (furthermore, I'd think in more implementations the
> Fwd-TSN will also be piggybacked with the retrans TSN119).
> 
> The stream blocking analysis (by Brian) only holds if the Fwd-TSN
> arrives without the retransmitted TSN119. But if both arrive at the
> same, would the analysis still hold?

The draft allows the FORWARD TSN to be sent immediately upon dropping and
there is no requirement to bundle retransmitted TSNs with the FORWARD TSN.

> 
> > ...We have all the PR streams
> > potentially being HOL blocked by loss of a single packet.  I think we need
> > to fix this.
> 
> Interesting point. But we seem to accept the same thing (streams
> potentially being HOL blocked by loss of a single packet) in base SCTP.
> How come?

We don't.

--brian

-- 
Brian F. G. Bidulock    ¦ The reasonable man adapts himself to the ¦
bidulock@openss7.org    ¦ world; the unreasonable one persists in  ¦
http://www.openss7.org/ ¦ trying  to adapt the  world  to himself. ¦
                        ¦ Therefore  all  progress  depends on the ¦
                        ¦ unreasonable man. -- George Bernard Shaw ¦

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



From daemon@ns.ietf.org  Wed Jun  5 18:51:35 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 SAA25570
	for <tsvwg-archive@odin.ietf.org>; Wed, 5 Jun 2002 18:51:34 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id SAA07215
	for tsvwg-archive@odin.ietf.org; Wed, 5 Jun 2002 18:52:04 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id SAA05906;
	Wed, 5 Jun 2002 18:16:12 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id SAA05877
	for <tsvwg@ns.ietf.org>; Wed, 5 Jun 2002 18:16:09 -0400 (EDT)
Received: from bfgbhome.inetint.com (IDENT:root@tnt-dal-42-242.dallas.net [209.44.42.242])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id SAA23332
	for <tsvwg@ietf.org>; Wed, 5 Jun 2002 18:15:38 -0400 (EDT)
Received: (from brian@localhost)
	by bfgbhome.inetint.com (8.9.3/8.9.3) id RAA04533;
	Wed, 5 Jun 2002 17:15:56 -0500
Date: Wed, 5 Jun 2002 17:15:56 -0500
From: "Brian F. G. Bidulock" <bidulock@openss7.org>
To: Qiaobing Xie <qxie1@email.mot.com>
Cc: Qiaobing Xie <Qiaobing.Xie@motorola.com>,
        "Armando L. Caro Jr." <acaro@mail.eecis.udel.edu>, tsvwg@ietf.org
Subject: Re: [Tsvwg] More on PRSCTP-00
Message-ID: <20020605171556.C2012@openss7.org>
Reply-To: bidulock@openss7.org
References: <20020531195333.D11468@openss7.org> <3CFBC11C.1EFA747@email.mot.com> <20020603144342.A30667@openss7.org> <3CFC144D.65697A41@email.mot.com> <20020603215952.E30667@openss7.org> <3CFD57D8.5637E431@email.mot.com> <20020604195020.A16020@openss7.org> <3CFE7403.59FB6F4C@email.mot.com> <20020605163502.A2012@openss7.org> <3CFE8C3F.C935774C@email.mot.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
User-Agent: Mutt/1.2i
In-Reply-To: <3CFE8C3F.C935774C@email.mot.com>; from qxie1@email.mot.com on Wed, Jun 05, 2002 at 05:10:07PM -0500
Organization: http://www.openss7.org/
Dsn-Notification-To: <bidulock@openss7.org>
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

Qiaobing,

On Wed, 05 Jun 2002, Qiaobing Xie wrote:

> "Brian F. G. Bidulock" wrote:
> ...
> > Pardon, let us say loss of a single SCTP packet.
> > It is still HOL blocking.
> 
> Again, what's really funny is that we suddenly start to see this HOL
> blocking caused by the loss of a single SCTP packet (presumably
> containing multiple TSNs from different streams) as an issue. We seem to
> be able to live with it in the base SCTP for a long time.

But these TSNs were also dropped.  It is the further loss of a single
TSN which causes the blocking.  It is not the case that the loss of a
single TSN from one stream blocks other streams, reagardless of the history
leading up to the loss of the TSN.

I'm afraid you're on your own on this one.

--brian

-- 
Brian F. G. Bidulock    ¦ The reasonable man adapts himself to the ¦
bidulock@openss7.org    ¦ world; the unreasonable one persists in  ¦
http://www.openss7.org/ ¦ trying  to adapt the  world  to himself. ¦
                        ¦ Therefore  all  progress  depends on the ¦
                        ¦ unreasonable man. -- George Bernard Shaw ¦

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



From daemon@ns.ietf.org  Wed Jun  5 19:05:21 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 TAA26154
	for <tsvwg-archive@odin.ietf.org>; Wed, 5 Jun 2002 19:05:21 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id TAA07985
	for tsvwg-archive@odin.ietf.org; Wed, 5 Jun 2002 19:05:51 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id SAA06758;
	Wed, 5 Jun 2002 18:33:25 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id SAA06729
	for <tsvwg@ns.ietf.org>; Wed, 5 Jun 2002 18:33:23 -0400 (EDT)
Received: from motgate.mot.com (motgate.mot.com [129.188.136.100])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id SAA24433
	for <tsvwg@ietf.org>; Wed, 5 Jun 2002 18:32:53 -0400 (EDT)
Received: [from mothost.mot.com (mothost.mot.com [129.188.137.101]) by motgate.mot.com (motgate 2.1) with ESMTP id PAA28463; Wed, 5 Jun 2002 15:33:18 -0700 (MST)]
Received: [from relay1.cig.mot.com (relay1.cig.mot.com [136.182.15.23]) by mothost.mot.com (MOT-pobox 2.0) with ESMTP id PAA28689; Wed, 5 Jun 2002 15:33:16 -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 g55MXFb13008;
	Wed, 5 Jun 2002 17:33:15 -0500 (CDT)
Received: from email.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 RAA28440;
	Wed, 5 Jun 2002 17:33:36 -0500 (CDT)
Message-ID: <3CFE9238.930A5B15@email.mot.com>
Date: Wed, 05 Jun 2002 17:35:36 -0500
From: Qiaobing Xie <qxie1@email.mot.com>
X-Mailer: Mozilla 4.77 [en] (Win98; U)
X-Accept-Language: en
MIME-Version: 1.0
To: bidulock@openss7.org
CC: Qiaobing Xie <Qiaobing.Xie@motorola.com>,
        "Armando L. Caro Jr." <acaro@mail.eecis.udel.edu>, tsvwg@ietf.org
Subject: Re: [Tsvwg] More on PRSCTP-00
References: <20020531195333.D11468@openss7.org> <3CFBC11C.1EFA747@email.mot.com> <20020603144342.A30667@openss7.org> <3CFC144D.65697A41@email.mot.com> <20020603215952.E30667@openss7.org> <3CFD57D8.5637E431@email.mot.com> <20020604195020.A16020@openss7.org> <3CFE7403.59FB6F4C@email.mot.com> <20020605163502.A2012@openss7.org> <3CFE8C3F.C935774C@email.mot.com> <20020605171556.C2012@openss7.org>
Content-Type: text/plain; charset=iso-8859-1
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

...
> I'm afraid you're on your own on this one.

Sadly but that seems to be the case :-) 

After all, democratic consensus building is meant to bring what the
*group as a whole* views as the best. But, make no mistake, we are *not*
the same as what's called "designed by committed". We are better than
"them" :-)

-Qiaobing

> 
> --brian
> 
> --
> Brian F. G. Bidulock    ¦ The reasonable man adapts himself to the ¦
> bidulock@openss7.org    ¦ world; the unreasonable one persists in  ¦
> http://www.openss7.org/ ¦ trying  to adapt the  world  to himself. ¦
>                         ¦ Therefore  all  progress  depends on the ¦
>                         ¦ unreasonable man. -- George Bernard Shaw ¦

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



From daemon@ns.ietf.org  Wed Jun  5 19:41:12 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 TAA27933
	for <tsvwg-archive@odin.ietf.org>; Wed, 5 Jun 2002 19:41:11 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id TAA09421
	for tsvwg-archive@odin.ietf.org; Wed, 5 Jun 2002 19:41:42 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id TAA08119;
	Wed, 5 Jun 2002 19:07:47 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id TAA08038
	for <tsvwg@ns.ietf.org>; Wed, 5 Jun 2002 19:07:44 -0400 (EDT)
Received: from bfgbhome.inetint.com (IDENT:root@tnt-dal-42-242.dallas.net [209.44.42.242])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id TAA26256
	for <tsvwg@ietf.org>; Wed, 5 Jun 2002 19:07:10 -0400 (EDT)
Received: (from brian@localhost)
	by bfgbhome.inetint.com (8.9.3/8.9.3) id SAA05239;
	Wed, 5 Jun 2002 18:07:15 -0500
Date: Wed, 5 Jun 2002 18:07:15 -0500
From: "Brian F. G. Bidulock" <bidulock@openss7.org>
To: Randall Stewart <randall@stewart.chicago.il.us>
Cc: Phillip Conrad <conrad@acm.org>, Qiaobing Xie <qxie1@email.mot.com>,
        Qiaobing Xie <Qiaobing.Xie@motorola.com>,
        "Armando L. Caro Jr." <acaro@mail.eecis.udel.edu>, tsvwg@ietf.org,
        Michael Tuexen <michael.tuexen@icn.siemens.de>,
        "Michael A. Ramalho" <mramalho@cisco.com>
Subject: Re: [Tsvwg] More on PRSCTP-00
Message-ID: <20020605180715.E2012@openss7.org>
Reply-To: bidulock@openss7.org
References: <3CF80F65.58E43813@email.mot.com> <20020531195333.D11468@openss7.org> <3CFBC11C.1EFA747@email.mot.com> <20020603144342.A30667@openss7.org> <3CFC144D.65697A41@email.mot.com> <20020603215952.E30667@openss7.org> <3CFD57D8.5637E431@email.mot.com> <20020604195020.A16020@openss7.org> <5.1.0.14.2.20020605155749.0236b050@joda.cis.temple.edu> <3CFE8964.81F6314F@stewart.chicago.il.us>
Mime-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
User-Agent: Mutt/1.2i
In-Reply-To: <3CFE8964.81F6314F@stewart.chicago.il.us>; from randall@stewart.chicago.il.us on Wed, Jun 05, 2002 at 04:57:56PM -0500
Organization: http://www.openss7.org/
Dsn-Notification-To: <bidulock@openss7.org>
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

Randall,

On Wed, 05 Jun 2002, Randall Stewart wrote:

> A few comments..
> 
> Phillip Conrad wrote:
> > 
> > Dear Colleagues:
> >       I agree with Randy that Brian has raised an important issue for us to
> > consider, and I think has explained it quite well with a very clear,
> > illustrative example.
> > 
> > The question is one of a tradeoff between two designs:
> > 
> > (A) current design: simpler, fewer bits on the wire, but may (re)-introduce
> > an HOL blocking problem
> 
> I must disagree with your "it is simpler" ... it is NOT. The question is
> where is the burden. Having implemented pr-sctp in the kernel of freebsd
> NOT having the stream number/SSN in the FWD-TSN means I must keep
> an extra set of linked lists and such to track any stream that has
> pending stuck data. When a fwd-tsn arrives I must HUNT all of these
> streams and look at each inbound queue.
> 
> The counter to this is the sender (in Brian and my earlier proposal
> that was to Q only) must track this info to send with the FWD-TSN.
> Actually
> less of a burden IMHO...

I agree.  I discovered this situation while implementing PR-SCTP in the
Linux Kernel.  The normal receiver can rely on the fact that the next
expected SSN is the last fully received SSN plus 1.  PR-SCTP without
a forward SSN causes the receive to have to update this expected value
with the next SSN in the reorder queue whenever there are no TSN gaps.
This even might require that when TSN gaps are filled that the receiver
walk all inbound stream structures to update the next expected SSN.
In an independent implementation, I have arrived at the same conclusion:
that the reciever has more work to do for FORWARD TSN without a forward
SSN.

In fact the receiver has this work to do if it has ever received a
FORWARD TSN.  This is because it is not possible for the receiver to
know a priori which streams could be affected by this scenairo.  Therefore,
the load on the receiver is increased even when this case does not occur.
It is the mere possibility of its occurence which introduces the load.
(Each time gaps close, the next expected SSN for all streams need to be
re-examined whether there are SSN gaps or not.)

On the sender side, the sender must always track the last sent SSN and
assign the next SSN (++SSN).  Indeed the sender when generating a FORWARD
TSN chunk would need to traverse outbound stream structures with dropped
TSNs to formulate Forward TSNs, but this is likely to be a far shorter
list than at the receiver and non-PR streams would never need to be
involved.

--brian

> > 
> > (B) alternative design: more complex, may require larger Forward TSN
> > chunks, but avoids re-introducing an HOL blocking problem, and thus
> > improves performance in some situations.
> 
> 
> see above it is NOT more complex... 
> 
> > 
> > The question of whether to choose A or B depends on whether you believe
> > that the situation Brian describes is a "corner case" or not.   So I've
> > walked through Brian's example, and have determined that I don't think it
> > is a corner case at all.   So, I'm inclined to agree with Randy and Brian
> > that we should consider a change.
> > 
> > But I'd like to invite others to walk through my line of reasoning to test
> > whether I've reasoned this correctly.
> 
> 
> I agree it is NOT a corner case..
> 
> > 
> > My rationale that Brian's example is NOT a corner case:
> > ========================================
> > 
> > In Brian's example, messages are being sent round-robin on a number of
> > streams.  This seems like a fairly reasonable use of SCTP streams, do that
> > aspect does not strike me as unusual.
> 
> I agree :>
> 
> > 
> > Now for the next part of this explanation, I'm going to have to infer
> > certain things, so, Brian, please correct me if I infer wrongly.   *One*
> > possibility for the sequence of events leading to the
> > situation Brian describes is that there is a burst loss of data chunks
> > TSN110 through TSN119 (ten chunks, but possibly less than ten packets..
> > perhaps even only ONE packet if the chunks are small enough to be
> > bundled!)... and then data chunks TSN120 through TSN128 make it through
> > OK.   This does not strike me as particularly unusual; a loss of a single
> > packet followed by a successful packet is sufficient to cause this situation.
> > 
> 
> Or the other possiblility would be a set of bundled chunks in one packet
> being lost.. either way is the same result I would think..
> 
> > The next thing that happens is that TSNs 110 through 118 are "dropped" in
> > the technical sense of that word; see the defn in the current draft.  This
> > means that the sender gives up on sending them, and issues a forward
> > TSN.   However, no Forward TSN is issued for TSN119.  This may be because
> > the messages on stream 9 are treated as Reliable.   This is NOT an unusual
> > situation; it is perfectly reasonably to assume that an application might
> > have "k" streams with data that can be dropped, and "n - k" streams of data
> > that must be treated as reliable.  In the case of Brian's example, n is 10,
> > and k is 9, but I think that his example will still produce similar results
> > even for other values of "k" and "n".
> > 
> 
> Note.. streams themselves are not reliable or partially reliable. With
> the new draft it is the message itself which is sent with the
> reliablilty
> set to "fully reliable". One possilbe implementation would be to make
> things correlate to a particular stream.. but this is just an
> implementation
> option..
> 
> > Bottom line:
> > ========
> > 
> > I don't think its a corner case at all.   We have all the PR streams
> > potentially being HOL blocked by loss of a single packet.  I think we need
> > to fix this.
> 
> I agree..
> 
> > 
> > The solution of forward SSNs is one viable solution I think.  There may be
> > others, but that's as good as any for right now I think.   There may be
> > another solution as well, but let me do some writing and thinking about it
> > off line before I muddy the waters with it here... just to make sure it is
> > a viable alternative.
> 
> agreed
> 
> 
> > 
> > Randy... I'm willing to make the text changes and send a proposed version
> > around to the co-authors for comment if there is consensus to move in that
> > direction.   Let me know.
> > 
> 
> I will get it out this weekend if I may.. there are some interesting
> things that I have thought about that must be covered... I had something
> like this in an old u-sctp draft (during Q and I's debate earlier :>)
> 
> 
> 
> > 
> > (Side note: I was worried that the "technical definition" of the word
> > "dropped" would cause confusion, and it already has.   Perhaps should use
> > another word in the draft to describe the operation where the sender
> > decides to no longer transmit and/or retransmit a specific data chunk.  How
> > about the term "abandoned" in place of "dropped"?)
> 
> 
> Sounds good..
> 
> note, Michael R also agrees .. or seems to .. we will talk more..
> 
> R
> 
> 
> > 
> > /ptc
> > 
> > At 11:44 AM 6/5/2002 -0500, Randall Stewart wrote:
> > >Brian/Qiaobing:
> > >
> > >I have read through this interesting debate. Qiaobing and
> > >I had a major discussion on this way back when it was
> > >considered U-SCTP. Renee Revis first pointed this out to me.. but
> > >from a different persepective i.e. the receiver being able to
> > >efficently process stuck TSN's ...
> > >
> > >After a lengthy debate I agreed to not have the SSN/SID listed
> > >in the FWD TSN... I did NOT think of the HOL blocking issue.. or
> > >I would have used it as a issue (I was the one arguing for the
> > >SSN/SID in the FWD-TSN :->).
> > >
> > >After watching this debate.. I think I have come to the conclusion that
> > >Brian is correct :)
> > >
> > >Michael R. is here... I will discuss it with him.. I want to get
> > >the other co-author's take on this one... If the others agree with
> > >Brian (and me :>) then I think we will put this in.... also I would
> > >like to hear what the rest of the WG feels on this???
> > >
> > >
> > >I think it is reasonable and not that much overhead to include JUST
> > >the dropped streams NEXT expected SSN... You would be adding 4 bytes
> > >for EVERY stream dropping one or more TSN's... note that this does
> > >NOT mean 4 bytes per TSN but 4 bytes per where one or more TSN's were
> > >dropped..
> > >
> > >
> > >I think this is reasonable.. it improves processing AND avoids HOL
> > >blocking..
> > >
> > >
> > >Michael T/Phil? what do you think... Michael R.. I will talk to you at
> > >the
> > >next class break..
> > >
> > >
> > >
> > >R
> > >
> > >--
> > >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)

-- 
Brian F. G. Bidulock    ¦ The reasonable man adapts himself to the ¦
bidulock@openss7.org    ¦ world; the unreasonable one persists in  ¦
http://www.openss7.org/ ¦ trying  to adapt the  world  to himself. ¦
                        ¦ Therefore  all  progress  depends on the ¦
                        ¦ unreasonable man. -- George Bernard Shaw ¦

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



From daemon@ns.ietf.org  Wed Jun  5 19:52:20 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 TAA28554
	for <tsvwg-archive@odin.ietf.org>; Wed, 5 Jun 2002 19:52:16 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id TAA09725
	for tsvwg-archive@odin.ietf.org; Wed, 5 Jun 2002 19:52:46 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id TAA08568;
	Wed, 5 Jun 2002 19:29:26 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id TAA08539
	for <tsvwg@ns.ietf.org>; Wed, 5 Jun 2002 19:29:24 -0400 (EDT)
Received: from mail.eecis.udel.edu (louie.udel.edu [128.4.40.12])
	by ietf.org (8.9.1a/8.9.1a) with SMTP id TAA27282
	for <tsvwg@ietf.org>; Wed, 5 Jun 2002 19:28:53 -0400 (EDT)
Received: from ren.eecis.udel.edu by mail.eecis.udel.edu id aa19417;
          5 Jun 2002 19:28 EDT
Date: Wed, 5 Jun 2002 19:28:43 -0400 (EDT)
From: "Armando L. Caro Jr." <acaro@mail.eecis.udel.edu>
Reply-To: "Armando L. Caro Jr." <acaro@acm.org>
To: Phillip Conrad <conrad@acm.org>
cc: Randall Stewart <randall@stewart.chicago.il.us>, bidulock@openss7.org,
        Qiaobing Xie <qxie1@email.mot.com>,
        Qiaobing Xie <Qiaobing.Xie@motorola.com>, tsvwg@ietf.org,
        Michael Tuexen <michael.tuexen@icn.siemens.de>,
        "Michael A. Ramalho" <mramalho@cisco.com>
MMDF-Warning:  Parse error in original version of preceding line at mail.eecis.udel.edu
Subject: Re: [Tsvwg] More on PRSCTP-00
In-Reply-To: <5.1.0.14.2.20020605155749.0236b050@joda.cis.temple.edu>
Message-ID: <Pine.GSO.4.33.0206051923070.6820-100000@ren.eecis.udel.edu>
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

On Wed, 5 Jun 2002, Phillip Conrad wrote:

> The question is one of a tradeoff between two designs:
>
> (A) current design: simpler, fewer bits on the wire, but may (re)-introduce
> an HOL blocking problem
>
> (B) alternative design: more complex, may require larger Forward TSN
> chunks, but avoids re-introducing an HOL blocking problem, and thus
> improves performance in some situations.

(A) is actually more complex from the receiver's point of view. (B) gives
the receiver explicit SID/SSN information about "dropped" TSNs.

> I don't think its a corner case at all.   We have all the PR streams

No, it isn't.

> (Side note: I was worried that the "technical definition" of the word
> "dropped" would cause confusion, and it already has.   Perhaps should use
> another word in the draft to describe the operation where the sender
> decides to no longer transmit and/or retransmit a specific data chunk.  How
> about the term "abandoned" in place of "dropped"?)

I agree that "dropped" is somewhat ambiguious... abandoned is a better
term.

~armando

0--                                                --0
| Armando L. Caro Jr.     |       acaro@cis.udel.edu |
| University of Delaware  |  www.cis.udel.edu/~acaro |
0--                                                --0



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



From daemon@ns.ietf.org  Wed Jun  5 20:16:26 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 UAA29706
	for <tsvwg-archive@odin.ietf.org>; Wed, 5 Jun 2002 20:16:26 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id UAA11115
	for tsvwg-archive@odin.ietf.org; Wed, 5 Jun 2002 20:16:56 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id TAA09467;
	Wed, 5 Jun 2002 19:43:45 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id TAA09436
	for <tsvwg@ns.ietf.org>; Wed, 5 Jun 2002 19:43:43 -0400 (EDT)
Received: from bfgbhome.inetint.com (IDENT:root@tnt-dal-42-242.dallas.net [209.44.42.242])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id TAA28062
	for <tsvwg@ietf.org>; Wed, 5 Jun 2002 19:43:10 -0400 (EDT)
Received: (from brian@localhost)
	by bfgbhome.inetint.com (8.9.3/8.9.3) id SAA05563;
	Wed, 5 Jun 2002 18:43:39 -0500
Date: Wed, 5 Jun 2002 18:43:39 -0500
From: "Brian F. G. Bidulock" <bidulock@openss7.org>
To: tsvwg@ietf.org, Randall Stewart <randall@stewart.chicago.il.us>
Subject: Re: [Tsvwg] PRSTCP-00
Message-ID: <20020605184339.G2012@openss7.org>
Reply-To: bidulock@openss7.org
References: <20020530040537.A21085@openss7.org>
Mime-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
User-Agent: Mutt/1.2i
In-Reply-To: <20020530040537.A21085@openss7.org>; from bidulock@openss7.org on Thu, May 30, 2002 at 04:05:37AM -0500
Organization: http://www.openss7.org/
Dsn-Notification-To: <bidulock@openss7.org>
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

Randall,

Could you take another look at this one?  Moving the FORWARD TSN
along on the basis of a gap ack doesn't appear to be a good idea.
When a subsequent SACK could reneg (or truncate) the gap ack
resulting in the loss of a non-abandonned (even reliable) TSN.

--brian

On Thu, 30 May 2002, Brian F. G. Bidulock wrote:

> Randall,
> 
> Two comments and a question:
> 
> 1) My compliments on an excellent draft.
> 
> 2) Page 11:
> 
>    "The data sender MUST then perform the following additional steps:"
> 
>    And
> 
>    "C2) Try to further advance..."
> 
>    C2) shows the Adv.Ack.Pt advancing over gap-acked TSNs.  What
>    happens if the peer is in the process of a renege on those
>    gap-acked TSNs?  When the SACK reneging on the gap-acked TSNs
>    arrives, the reneged TSNs will be behind the Adv.Ack.Pt.  Also,
>    when the peer receives the FORWARD TSN, it will be forced to
>    update its cumulative ack point beyond the reneged TSNs.  These
>    reneged TSN might be fully reliable TSNs which have now been
>    dropped.
> 
>    Considering reneging, I don't think that the PRSCTP sender should
>    advance the Adv.Ack.Pt past gap-acked TSNs.  It should wait until
>    those TSNs are cumulatively acked (no longer renegable -- is that
>    a word?) before advancing the Adv.Ack.Pt further.
> 

[snip]

-- 
Brian F. G. Bidulock    ¦ The reasonable man adapts himself to the ¦
bidulock@openss7.org    ¦ world; the unreasonable one persists in  ¦
http://www.openss7.org/ ¦ trying  to adapt the  world  to himself. ¦
                        ¦ Therefore  all  progress  depends on the ¦
                        ¦ unreasonable man. -- George Bernard Shaw ¦

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



From daemon@ns.ietf.org  Wed Jun  5 20:24:10 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 UAA00021
	for <tsvwg-archive@odin.ietf.org>; Wed, 5 Jun 2002 20:24:05 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id UAA11388
	for tsvwg-archive@odin.ietf.org; Wed, 5 Jun 2002 20:24:10 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id TAA09813;
	Wed, 5 Jun 2002 19:55:20 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id TAA09787
	for <tsvwg@ns.ietf.org>; Wed, 5 Jun 2002 19:55:18 -0400 (EDT)
Received: from motgate.mot.com (motgate.mot.com [129.188.136.100])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id TAA28686
	for <tsvwg@ietf.org>; Wed, 5 Jun 2002 19:54:47 -0400 (EDT)
Received: [from mothost.mot.com (mothost.mot.com [129.188.137.101]) by motgate.mot.com (motgate 2.1) with ESMTP id QAA13439; Wed, 5 Jun 2002 16:55:17 -0700 (MST)]
Received: [from relay1.cig.mot.com (relay1.cig.mot.com [136.182.15.23]) by mothost.mot.com (MOT-pobox 2.0) with ESMTP id QAA17803; Wed, 5 Jun 2002 16:55:14 -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 g55NtDb17800;
	Wed, 5 Jun 2002 18:55:13 -0500 (CDT)
Received: from email.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 SAA29304;
	Wed, 5 Jun 2002 18:55:34 -0500 (CDT)
Message-ID: <3CFEA56B.FAF5EDEC@email.mot.com>
Date: Wed, 05 Jun 2002 18:57:31 -0500
From: Qiaobing Xie <qxie1@email.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: Qiaobing Xie <Qiaobing.Xie@motorola.com>, Phillip Conrad <conrad@acm.org>,
        bidulock@openss7.org,
        "Armando L. Caro Jr." <acaro@mail.eecis.udel.edu>, tsvwg@ietf.org,
        Michael Tuexen <michael.tuexen@icn.siemens.de>,
        "Michael A. Ramalho" <mramalho@cisco.com>
Subject: Re: [Tsvwg] More on PRSCTP-00
References: <Pine.GSO.4.33.0205311736060.8809-100000@ren.eecis.udel.edu>
			 <3CF80F65.58E43813@email.mot.com> <20020531195333.D11468@openss7.org>
			 <3CFBC11C.1EFA747@email.mot.com> <20020603144342.A30667@openss7.org>
			 <3CFC144D.65697A41@email.mot.com> <20020603215952.E30667@openss7.org>
			 <3CFD57D8.5637E431@email.mot.com> <20020604195020.A16020@openss7.org> <5.1.0.14.2.20020605155749.0236b050@joda.cis.temple.edu> <3CFE8A04.24188859@email.mot.com> <3CFEA024.394BC38F@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

R,

...
> No, I don't think you can make this assumption.. see above... only
> if TSN119 was bundled with the rest...

Well, I didn't make this assumption, but Phil (and Brian) did. If you go
back to read Phil's email, *he* built his case assuming the 9 TSNs are
bundled in a single packet. This is why I said the case Phil used to
illustrate the problem can use some tighten-up :-)

...
> > Interesting point. But we seem to accept the same thing (streams
> > potentially being HOL blocked by loss of a single packet) in base SCTP.
> > How come?
> >
> But a single loss can never block more than ONE stream.. I think
> this is the difference...

The argument from Brian and Phil is based on a single loss of a packet
that contains multiple TSNs from different streams. I am arguing if this
were to happen to the base SCTP, all streams that own one or more TSNs
in the lost packet will be affected in the same way.

Anyway, I think you guys have already made up your mind on this... I
have no objection as said before.

-Q

> 
> 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  Thu Jun  6 19:29:25 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 TAA13687
	for <tsvwg-archive@odin.ietf.org>; Thu, 6 Jun 2002 19:29:25 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id TAA05930
	for tsvwg-archive@odin.ietf.org; Thu, 6 Jun 2002 19:29:56 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id SAA00311;
	Thu, 6 Jun 2002 18:45:10 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id SAA00274
	for <tsvwg@optimus.ietf.org>; Thu, 6 Jun 2002 18:45:08 -0400 (EDT)
Received: from bfgbhome.inetint.com (IDENT:root@tnt-dal-42-242.dallas.net [209.44.42.242])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id SAA01491
	for <tsvwg@ietf.org>; Thu, 6 Jun 2002 18:44:33 -0400 (EDT)
Received: (from brian@localhost)
	by bfgbhome.inetint.com (8.9.3/8.9.3) id QAA20748;
	Thu, 6 Jun 2002 16:43:17 -0500
Date: Thu, 6 Jun 2002 16:43:17 -0500
From: "Brian F. G. Bidulock" <bidulock@openss7.org>
To: Randall Stewart <randall@stewart.chicago.il.us>
Cc: tsvwg@ietf.org
Message-ID: <20020606164317.A20729@openss7.org>
Reply-To: bidulock@openss7.org
References: <3CFF9108.FA4C1000@stewart.chicago.il.us>
Mime-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
User-Agent: Mutt/1.2i
In-Reply-To: <3CFF9108.FA4C1000@stewart.chicago.il.us>; from randall@stewart.chicago.il.us on Thu, Jun 06, 2002 at 11:42:48AM -0500
Organization: http://www.openss7.org/
Dsn-Notification-To: <bidulock@openss7.org>
Content-Transfer-Encoding: 8bit
Subject: [Tsvwg] Re: [Fwd: PRSTCP-00]
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,

Please see comments below...

On Thu, 06 Jun 2002, Randall Stewart wrote:

> "Brian F. G. Bidulock" wrote:
> > 
> > Randall,
> > 
> > Two comments and a question:
> > 
> > 1) My compliments on an excellent draft.
> 
> Thanks... this was not only my work .. but the excellent 
> ideas and help of my co-authors as well... 
> 
> > 
> > 2) Page 11:
> > 
> >    "The data sender MUST then perform the following additional steps:"
> > 
> >    And
> > 
> >    "C2) Try to further advance..."
> > 
> >    C2) shows the Adv.Ack.Pt advancing over gap-acked TSNs.  What
> >    happens if the peer is in the process of a renege on those
> >    gap-acked TSNs?  When the SACK reneging on the gap-acked TSNs
> >    arrives, the reneged TSNs will be behind the Adv.Ack.Pt.  Also,
> >    when the peer receives the FORWARD TSN, it will be forced to
> >    update its cumulative ack point beyond the reneged TSNs.  These
> >    reneged TSN might be fully reliable TSNs which have now been
> >    dropped.
> 
> Hmm... I had not thought in detail on this case...
> 
> Now so I send
> 
> Last SACK
> cum-ack=90
> Gap=92-94
> Gap=96-99
> 
> and I now (as the sender) want
> to skip to 101..
> FWD-TSN(101)------->
> 
>                    but in the mean time my peer is doing
>                       <----SACK(90) ... NO GAPS..
> 
> 
> If I do the above (as the current draft states) when the
> SACK arrives the SACK is the SAME point as the CUM-ACK... so
> It is not old or behind... I don't strike anything... and I would
> in theory go to put a mark on the 92-94 and the 96-99 .. but they
> are already marked as being skipped...

What if the SSNs in these TSNs are reliable?  They have been
dropped and yet they have not been successfully delivered to
the peer.  This compromises the reliability of all other "reliable"
messages.

--brian

> 
> Hmmm... I am not sure I see a problem... what I think would
> happen is that we would see 90, try to advance the cum-ack
> and once again generate a 
> 
> FWD-TSN(101)-------------> 
> 
> in response to the SACK(90)...
> 
> Am I missing something?
> 
> > 
> >    Considering reneging, I don't think that the PRSCTP sender should
> >    advance the Adv.Ack.Pt past gap-acked TSNs.  It should wait until
> >    those TSNs are cumulatively acked (no longer renegable -- is that
> >    a word?) before advancing the Adv.Ack.Pt further.
> 
> I am not sure why.. maybe I miss something? can you explain?
> 
> > 
> > Q) I understand that it is only one possible PRSCTP user interface
> >    methodology, but what is the use of a transport-based "timed
> >    reliability"?  I agree that application timed reliability (from
> >    transport user to transport user) might be useful, but this
> >    concept of "timed reliability" does not take into account the
> >    amount of time that data waits queued in the peer's receive queue
> >    before being delivered to the application.  As such, it only
> >    really times the amount of time in the send and retransmission
> >    queues which is related to transmit congestion only: when an
> 
> True... if the peer is NOT READING ... this time is not
> accounted for...
> 
> 
> >    association becomes congested, data with the shortest lifetimes
> >    will be dropped.  It also appears to mean that data with the
> >    longer lifetimes have higher priority in times of congestion.
> >    With a mix of partial and full reliability data, all the partial
> >    data will be dropped in times of congestion and all the full
> >    reliability data will queue indefinitely waiting for the congestion
> >    to abate.  What has this to do with "reliability"?  It appears
> >    more reminicent of the Unix "nice" command.  Unfortunately,
> >    setting the priority of a message by setting its lifetime
> >    parameter would require a deep understanding of the queueing
> >    model and the traffic distribution of all other messages in all
> >    other streams.  It seems unlikely that a useful value could be
> >    determined except for the most narrow of applications.
> > 
> 
> I don't know if I agree here... I am not strongly married to
> this particular service but I do think it is useful. I can see
> sending a message that is going to be useless in 4 seconds from
> the time I send it.. so unless I get it ACK'd by the
> peer there is NO chance the peer can use it..
> 
> If the peer is not reading and I have this timelness constraint
> I see this as more of a bug in the reader than a issue with
> it not reading...
> 
> I think there are quite a few uses for this service but I
> will let my co-authors comment on this.. 
> 
> I like a buffer constrained model myself...
> 
> >    Is there a good real-world example of where transport-based
> >    "timed reliability" would be most useful?  The previous concept
> >    of limiting the number of retransmissions appeared far more
> >    useful for such things as RTP/SCTP (e.g. number of retransmission
> >    set to zero).
> 
> I will let Phil respond to this... he may have some other ideas.
> 
> I will be writting a service here shortly on buffer contrained
> use of PR-SCTP... more targeted for ipfix...
> 
> 
> > 
> > Well, suppose thats really a bunch of comments and a bunch of
> > questiosns...
> > 
> 
> 
> R
> 
> -- 
> Randall R. Stewart
> randall@stewart.chicago.il.us 815-342-5222 (cell phone)


-- 
Brian F. G. Bidulock    ¦ The reasonable man adapts himself to the ¦
bidulock@openss7.org    ¦ world; the unreasonable one persists in  ¦
http://www.openss7.org/ ¦ trying  to adapt the  world  to himself. ¦
                        ¦ Therefore  all  progress  depends on the ¦
                        ¦ unreasonable man. -- George Bernard Shaw ¦

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



From daemon@optimus.ietf.org  Thu Jun  6 19:37:06 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 TAA15402
	for <tsvwg-archive@odin.ietf.org>; Thu, 6 Jun 2002 19:37:01 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id TAA06811
	for tsvwg-archive@odin.ietf.org; Thu, 6 Jun 2002 19:37:32 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id SAA01857;
	Thu, 6 Jun 2002 18:56:48 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id SAA01829
	for <tsvwg@optimus.ietf.org>; Thu, 6 Jun 2002 18:56:46 -0400 (EDT)
Received: from motgate.mot.com (motgate.mot.com [129.188.136.100])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id SAA05356
	for <tsvwg@ietf.org>; Thu, 6 Jun 2002 18:56:14 -0400 (EDT)
Received: [from mothost.mot.com (mothost.mot.com [129.188.137.101]) by motgate.mot.com (motgate 2.1) with ESMTP id OAA14880; Thu, 6 Jun 2002 14:28:31 -0700 (MST)]
Received: [from relay1.cig.mot.com (relay1.cig.mot.com [136.182.15.23]) by mothost.mot.com (MOT-pobox 2.0) with ESMTP id OAA14867; Thu, 6 Jun 2002 14:28:26 -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 g56LOib29995;
	Thu, 6 Jun 2002 16:24:44 -0500 (CDT)
Received: from email.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 QAA13542;
	Thu, 6 Jun 2002 16:25:05 -0500 (CDT)
Message-ID: <3CFFD3B8.8B2EC16D@email.mot.com>
Date: Thu, 06 Jun 2002 16:27:20 -0500
From: Qiaobing Xie <qxie1@email.mot.com>
X-Mailer: Mozilla 4.77 [en] (Win98; U)
X-Accept-Language: en
MIME-Version: 1.0
To: Phillip Conrad <conrad@acm.org>
CC: Randall Stewart <randall@stewart.chicago.il.us>, bidulock@openss7.org,
        Qiaobing Xie <Qiaobing.Xie@motorola.com>,
        "Armando L. Caro Jr." <acaro@mail.eecis.udel.edu>, tsvwg@ietf.org,
        Michael Tuexen <michael.tuexen@icn.siemens.de>,
        "Michael A. Ramalho" <mramalho@cisco.com>
Subject: Re: [Tsvwg] More on PRSCTP-00
References: <Pine.GSO.4.33.0205311736060.8809-100000@ren.eecis.udel.edu>
	 <3CF80F65.58E43813@email.mot.com> <20020531195333.D11468@openss7.org>
	 <3CFBC11C.1EFA747@email.mot.com> <20020603144342.A30667@openss7.org>
	 <3CFC144D.65697A41@email.mot.com> <20020603215952.E30667@openss7.org>
	 <3CFD57D8.5637E431@email.mot.com> <20020604195020.A16020@openss7.org>
	 <5.1.0.14.2.20020605155749.0236b050@joda.cis.temple.edu> <5.1.0.14.2.20020605230911.025cca18@mail.earthlink.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

Hi, Phill,

Phillip Conrad wrote:
> 
...
> Let me see if understand your point.. I think you are saying that from a
> holistic standpoint, the overall system may be simpler if we introduce
> forward SSNs, because as you, Armando and Brian have all pointed out, the
> additional fields can actually make implementation of the sender and
> receiver more straightforward, and more efficient... i.e., simpler.   I can
> accept this line of reasoning.

To add more bits to the wire in exchange for some implementation ease is
a valid point. I will always buy this argument, as long as we don't go
to the extreme.

....
> The objection that Brian raises, and others (included myself) have
> concurred with, is that the FORWARD TSN chunk, as specified in the current
> draft, creates the possibility that this principle may be negated. As
> Brian's example illustrates, the loss of a TSN unrelated to stream j may
> delay delivery of packets on stream j, because when a FORWARD TSN moves the
> cum ack point forward, there is an ambiguity about which streams the
> "skipped" packets belong to. The FORWARD SSN corrects this ambiguity.

I have tried very hard but failed to draft up an "all considered"
scenario with 2 streams to show how one missing TSN from one stream can
unfairly block the delivery of a queued, ready-to-be-delivered message
from the other stream. Maybe this ambiguity only manifests itself with
at least 3 streams involved? Probably another research topic :-)

....
> I am glad that Qiaobing is being rigorous and asking us to consider very
> carefully whether a change in the design is needed, and inspiring us to
> make our arguments very precise.  Qiaobing, don't let us bully you... as
> the film "Twelve Angry Men" illustrates.. just because someone is "on their
> own", doesn't mean they are wrong.   In that film, a jury that starts out
> 11-to-1 in favor of a guilty verdict, eventually reaches a unanimous
> verdict of "not guilty."  We may disagree on "taste" or "philosophy", but
> if we are in disagreement about things that we should be able to logically
> reason out, then we probably need to continue the dialogue until we have
> true consensus.

Thank you, Phill! I am truly really surprised to hear this kind of nice
words, especially on a mailing list like this one in IETF :-)

regards,
-Qiaobing

> 
> Thanks,
> Phill

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



From daemon@optimus.ietf.org  Thu Jun  6 20:12:28 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 UAA18842
	for <tsvwg-archive@odin.ietf.org>; Thu, 6 Jun 2002 20:12:28 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id UAA09052
	for tsvwg-archive@odin.ietf.org; Thu, 6 Jun 2002 20:13:00 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id TAA07116;
	Thu, 6 Jun 2002 19:42:02 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id TAA07085
	for <tsvwg@optimus.ietf.org>; Thu, 6 Jun 2002 19:42:00 -0400 (EDT)
Received: from bfgbhome.inetint.com (IDENT:root@tnt-dal-42-242.dallas.net [209.44.42.242])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id TAA16295
	for <tsvwg@ietf.org>; Thu, 6 Jun 2002 19:41:27 -0400 (EDT)
Received: (from brian@localhost)
	by bfgbhome.inetint.com (8.9.3/8.9.3) id SAA22183;
	Thu, 6 Jun 2002 18:41:51 -0500
Date: Thu, 6 Jun 2002 18:41:51 -0500
From: "Brian F. G. Bidulock" <bidulock@openss7.org>
To: randall@cisco.com, tsvwg@ietf.org
Message-ID: <20020606184151.D20729@openss7.org>
Reply-To: bidulock@openss7.org
Mime-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
User-Agent: Mutt/1.2i
In-Reply-To: <no.id>; from randall@cisco.com on Thu, Jun 06, 2002 at 11:05:07PM +0000
Organization: http://www.openss7.org/
Dsn-Notification-To: <bidulock@openss7.org>
Content-Transfer-Encoding: 8bit
Subject: [Tsvwg] Re: your mail
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,

Here's an attempt to explain:  view from the sender's side:

  TSN  98  SID 1  SSN 1 --->
  TSN  99  SID 1  SSN 2 --->
  TSN 100  SID 1  SSN 3 --->
  TSN 101  SID 2  SSN 1 --->
  TSN 102  SID 3  SSN 1 --->
  TSN 103  SID 4  SSN 1 --->
  TSN 104  SID 5  SSN 1 --->
  <-- SACK (CUM ACK 99, GAP ACK for 101, 102, 103, 104)
  (SID 1 SSN 3 expires)
  
  Now the draft tells use when moving the TSN forward, we
  include gap-acked TSNs, so we send:

  FORWARD TSN 104 -->
  <-- SACK (CUM ACK 99, GAP ACK none) (101 thru 104 were discarded)

  Now, it could happen that the data in streams 2 thru 5 were being
  sent "reliably".  They have been lost to the receiver (it discarded
  them when reneging) and are also lost to the sender.  Not terribly
  reliable, is it?

My problem is with C2.

Now, this effect is also true within a stream as between streams (because
each individual SSN within a stream could be sent "reliably" or not).

The proposed solution is that C2 be reworded to advance the
Adv.Peer.Ack.Pnt only over "abandonned" TSNs.  So, in the above
example, the FORWARD TSN would be:

  FORWARD TSN 100 --->

If there are no renegs, the peer should immediately reply with

  <-- SACK (CUM ACK 104)

anyway (this was Armando's point earlier in this thread), so
performance is not impaired and reliability is restored.

--brian


On Thu, 06 Jun 2002, randall@cisco.com wrote:
> 
> "Brian F. G. Bidulock" wrote:
> > > Hmm... I had not thought in detail on this case...
> > >
> > > Now so I send
> > >
> > > Last SACK
> > > cum-ack=90
> > > Gap=92-94
> > > Gap=96-99
> > >
> > > and I now (as the sender) want
> > > to skip to 101..
> > > FWD-TSN(101)------->
> > >
> > >                    but in the mean time my peer is doing
> > >                       <----SACK(90) ... NO GAPS..
> > >
> > >
> > > If I do the above (as the current draft states) when the
> > > SACK arrives the SACK is the SAME point as the CUM-ACK... so
> > > It is not old or behind... I don't strike anything... and I would
> > > in theory go to put a mark on the 92-94 and the 96-99 .. but they
> > > are already marked as being skipped...
> > 
> > What if the SSNs in these TSNs are reliable?  They have been
> > dropped and yet they have not been successfully delivered to
> > the peer.  This compromises the reliability of all other "reliable"
> > messages.
> > 
> 
> Brian:
> 
> I am a bit confused.. it is probably just late in the day and also
> the classes I am in :0
> 
> So could you help me with this ... I want to understand (and I am not
> :<)
> 
> So in my example...
> 
> Are you saying something before 101 is NOT to be skipped i.e. reliable?
> 
> If so I would not do a fwd-tsn to 101 but to an earlier number....
> 
> If you could illustrate this with the above example it might help
> me understand what you mean...
> 
> 
> R
> 
> -- 
> Randall R. Stewart
> randall@stewart.chicago.il.us 815-342-5222 (cell phone)

-- 
Brian F. G. Bidulock    ¦ The reasonable man adapts himself to the ¦
bidulock@openss7.org    ¦ world; the unreasonable one persists in  ¦
http://www.openss7.org/ ¦ trying  to adapt the  world  to himself. ¦
                        ¦ Therefore  all  progress  depends on the ¦
                        ¦ unreasonable man. -- George Bernard Shaw ¦

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



From daemon@optimus.ietf.org  Thu Jun  6 23:10:57 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 XAA00023
	for <tsvwg-archive@odin.ietf.org>; Thu, 6 Jun 2002 23:10:57 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id XAA18953
	for tsvwg-archive@odin.ietf.org; Thu, 6 Jun 2002 23:11:28 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id WAA17100;
	Thu, 6 Jun 2002 22:32:24 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id WAA17073
	for <tsvwg@optimus.ietf.org>; Thu, 6 Jun 2002 22:32:22 -0400 (EDT)
Received: from motgate4.mot.com (motgate4.mot.com [144.189.100.102])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id WAA27273
	for <tsvwg@ietf.org>; Thu, 6 Jun 2002 22:31:50 -0400 (EDT)
Received: [from pobox3.mot.com (pobox3.mot.com [10.64.251.242]) by motgate4.mot.com (motgate4 2.1) with ESMTP id TAA04899 for <tsvwg@ietf.org>; Thu, 6 Jun 2002 19:32:17 -0700 (MST)]
Received: [from relay1.cig.mot.com (relay1.cig.mot.com [136.182.15.23]) by pobox3.mot.com (MOT-pobox3 2.0) with ESMTP id TAA14875 for <tsvwg@ietf.org>; Thu, 6 Jun 2002 19:31:56 -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 g572WCb16726;
	Thu, 6 Jun 2002 21:32:12 -0500 (CDT)
Received: from email.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 VAA16840;
	Thu, 6 Jun 2002 21:32:34 -0500 (CDT)
Message-ID: <3D001BC8.51B30A14@email.mot.com>
Date: Thu, 06 Jun 2002 21:34:48 -0500
From: Qiaobing Xie <qxie1@email.mot.com>
X-Mailer: Mozilla 4.77 [en] (Win98; U)
X-Accept-Language: en
MIME-Version: 1.0
To: bidulock@openss7.org
CC: randall@cisco.com, tsvwg@ietf.org
Subject: Re: [Tsvwg] Re: your mail
References: <20020606184151.D20729@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

...
> The proposed solution is that C2 be reworded to advance the
> Adv.Peer.Ack.Pnt only over "abandonned" TSNs.  So, in the above
> example, the FORWARD TSN would be:
> 
>   FORWARD TSN 100 --->
> 

This looks like a fine solution to me.

-Qiaobing

> If there are no renegs, the peer should immediately reply with
> 
>   <-- SACK (CUM ACK 104)
> 
> anyway (this was Armando's point earlier in this thread), so
> performance is not impaired and reliability is restored.
> 
> --brian

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



From daemon@optimus.ietf.org  Fri Jun  7 09:22:39 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 JAA25693
	for <tsvwg-archive@odin.ietf.org>; Fri, 7 Jun 2002 09:22:39 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id JAA26343
	for tsvwg-archive@odin.ietf.org; Fri, 7 Jun 2002 09:23:11 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id IAA24400;
	Fri, 7 Jun 2002 08:47:14 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id IAA24372
	for <tsvwg@optimus.ietf.org>; Fri, 7 Jun 2002 08:47:12 -0400 (EDT)
Received: from irmgard.exp-math.uni-essen.de (irmgard.exp-math.uni-essen.de [132.252.150.18])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id IAA24995
	for <tsvwg@ietf.org>; Fri, 7 Jun 2002 08:46:40 -0400 (EDT)
Received: from there (kappes.exp-math.uni-essen.de [132.252.150.151])
	by irmgard.exp-math.uni-essen.de (8.12.2/8.12.2) with SMTP id g57ClAMr027226
	for <tsvwg@ietf.org>; Fri, 7 Jun 2002 14:47:10 +0200
Message-Id: <200206071247.g57ClAMr027226@irmgard.exp-math.uni-essen.de>
Content-Type: text/plain;
  charset="iso-8859-1"
From: Thomas Dreibholz <dreibh@exp-math.uni-essen.de>
Organization: University of Essen, Institute for Experimental Mathematics
To: tsvwg@ietf.org
Date: Fri, 7 Jun 2002 14:45:56 +0200
X-Mailer: KMail [version 1.3.2]
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
Content-Transfer-Encoding: 8bit
Subject: [Tsvwg] Comments to SCTP Socket API 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: 8bit

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

Hi all,

Some comments to draft-ietf-tsvwg-sctpsocket-04.txt:

- - Appendix A and B use the setsockopt() parameters SCTP_RECVDATAIOEVNT, 
SCTP_RECVASSOCEVNT, SCTP_RECVSENDFAILEVNT, ... to subscribe for the events to 
be received. Version 4 defines the option SCTP_SET_EVENTS to do these 
settings, the old options seem to have been removed.
- - Section 5.3.1.6, SCTP_ADAPTION_INDICATION:
sctp_adaption_event uses the prefix sai_ for its fields. Therefore, 
sse_assoc_id should be replaced by sai_assoc_id.


Best regards
- -- 
=======================================================================
 Dipl.-Inform. Thomas Dreibholz

 University of Essen,                            Room ES210
 Inst. for Experimental Mathematics              Ellernstraße 29
 Computer Networking Technology Group            D-45326 Essen/Germany
- -----------------------------------------------------------------------
 E-Mail:     dreibh@exp-math.uni-essen.de
 Homepage:   http://www.exp-math.uni-essen.de/~dreibh
=======================================================================

- -----BEGIN PGP PUBLIC KEY BLOCK-----
Version: GnuPG v1.0.6 (GNU/Linux)
Comment: Weitere Infos: siehe http://www.gnupg.org
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=g36Q
- -----END PGP PUBLIC KEY BLOCK-----
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.6 (GNU/Linux)
Comment: For info see http://www.gnupg.org

iD8DBQE9AKsE32BbsHYPLWURAkqSAJkBjgQy2XL4bRRxgrgDlAGQrt634ACgh+QZ
n2FgowQ9cqIqDtFzoMVM6F0=
=HPok
-----END PGP SIGNATURE-----

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



From daemon@optimus.ietf.org  Fri Jun  7 09:58:46 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 JAA26495
	for <tsvwg-archive@odin.ietf.org>; Fri, 7 Jun 2002 09:58:46 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id JAA28212
	for tsvwg-archive@odin.ietf.org; Fri, 7 Jun 2002 09:59:18 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id JAA26296;
	Fri, 7 Jun 2002 09:21:31 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id JAA26269
	for <tsvwg@optimus.ietf.org>; Fri, 7 Jun 2002 09:21:29 -0400 (EDT)
Received: from irmgard.exp-math.uni-essen.de (irmgard.exp-math.uni-essen.de [132.252.150.18])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id JAA25660
	for <tsvwg@ietf.org>; Fri, 7 Jun 2002 09:20:56 -0400 (EDT)
Received: from there (kappes.exp-math.uni-essen.de [132.252.150.151])
	by irmgard.exp-math.uni-essen.de (8.12.2/8.12.2) with SMTP id g57DLQMr034324
	for <tsvwg@ietf.org>; Fri, 7 Jun 2002 15:21:26 +0200
Message-Id: <200206071321.g57DLQMr034324@irmgard.exp-math.uni-essen.de>
Content-Type: text/plain;
  charset="iso-8859-1"
From: Thomas Dreibholz <dreibh@exp-math.uni-essen.de>
Organization: University of Essen, Institute for Experimental Mathematics
To: tsvwg@ietf.org
Date: Fri, 7 Jun 2002 15:20:13 +0200
X-Mailer: KMail [version 1.3.2]
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
Content-Transfer-Encoding: 8bit
Subject: [Tsvwg] Another comment to SCTP Socket API 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: 8bit

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

Hi all,

Another comment to draft-ietf-tsvwg-sctpsocket-04.txt:
- - 5.3.1.7 SCTP_PARTIAL_DELIVERY_EVENT
pdapi_type should be SCTP_PARTIAL_DELIVERY_EVENT, and not 
SCTP_ADAPTION_INDICATION.


Best regards
- -- 
=======================================================================
 Dipl.-Inform. Thomas Dreibholz

 University of Essen,                            Room ES210
 Inst. for Experimental Mathematics              Ellernstraße 29
 Computer Networking Technology Group            D-45326 Essen/Germany
- -----------------------------------------------------------------------
 E-Mail:     dreibh@exp-math.uni-essen.de
 Homepage:   http://www.exp-math.uni-essen.de/~dreibh
=======================================================================

- -----BEGIN PGP PUBLIC KEY BLOCK-----
Version: GnuPG v1.0.6 (GNU/Linux)
Comment: Weitere Infos: siehe http://www.gnupg.org
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=g36Q
- -----END PGP PUBLIC KEY BLOCK-----
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.6 (GNU/Linux)
Comment: For info see http://www.gnupg.org

iD8DBQE9ALMN32BbsHYPLWURApKcAKCueq4UH4jGoRpc38HoTgH/QADIiwCfTQtD
zS6bqlF6WSRhr8pacibJvfg=
=VarD
-----END PGP SIGNATURE-----

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



From daemon@optimus.ietf.org  Fri Jun  7 10:51:39 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 KAA29066
	for <tsvwg-archive@odin.ietf.org>; Fri, 7 Jun 2002 10:51:38 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id KAA02593
	for tsvwg-archive@odin.ietf.org; Fri, 7 Jun 2002 10:52:11 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id KAA00296;
	Fri, 7 Jun 2002 10:28:48 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id NAA19609
	for <tsvwg@optimus.ietf.org>; Mon, 3 Jun 2002 13:37:45 -0400 (EDT)
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 NAA25052
	for <tsvwg@ietf.org>; Mon, 3 Jun 2002 13:37:14 -0400 (EDT)
Received: from mira-sjc5-2.cisco.com (IDENT:mirapoint@mira-sjc5-2.cisco.com [171.71.163.16])
	by sj-msg-core-3.cisco.com (8.12.2/8.12.2) with ESMTP id g53HauuF021793;
	Mon, 3 Jun 2002 10:36:56 -0700 (PDT)
Received: from stewart.chicago.il.us (dhcp-171-69-54-231.cisco.com [171.69.54.231])
	by mira-sjc5-2.cisco.com (Mirapoint)
	with ESMTP id ACO13315;
	Mon, 3 Jun 2002 10:37:11 -0700 (PDT)
Message-ID: <3CFBA941.D82F5FF4@stewart.chicago.il.us>
Date: Mon, 03 Jun 2002 12:37:06 -0500
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: bidulock@openss7.org
CC: tsvwg@ietf.org
Subject: Re: [Tsvwg] More on PRSCTP-00
References: <20020530113915.A27056@openss7.org>
Content-Type: text/plain; charset=iso-8859-1
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

I agree here.. .maybe the FWD-TSN chunk should
report the new TSN and
all stream/sequence numbers dropped... this
would elimanate this...

I proposed this a long time back but Q did not like it :>

I think your reason here may be a good justification.. I will
catch up on this email thread when I get a chance :-0

R

"Brian F. G. Bidulock" wrote:
> 
> Randall,
> 
> More problem with PRSCTP:
> 
> Because FORWARD TSN is done on a TSN basis only, it reintroduces
> Head-Of-Line blocking: if the SSNs following a drop arrive at the
> receive out of order on an ordered stream, it is not possible for
> the receive to determine what is the next expected SSN in the
> stream because the FORWARD TSN contains no information about which
> SSNs were dropped.  Therefore, after a drop reported in an FORWARD
> TSN, every stream is blocked from delivering to the ULP until all
> missing TSNs have arrived.  This is HOL blocking again.  I would
> think that the FORWARD TSN procedure should reduce delay in delivery
> to the ULP rather than increase it.
> 
> One solution to this is to report dropped SSNs.
> 
> --brian
> 
> --
> Brian F. G. Bidulock    ¦ The reasonable man adapts himself to the ¦
> bidulock@openss7.org    ¦ world; the unreasonable one persists in  ¦
> http://www.openss7.org/ ¦ trying  to adapt the  world  to himself. ¦
>                         ¦ Therefore  all  progress  depends on the ¦
>                         ¦ unreasonable man. -- George Bernard Shaw ¦
> 
> _______________________________________________
> 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  Fri Jun  7 10:53:44 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 KAA29142
	for <tsvwg-archive@odin.ietf.org>; Fri, 7 Jun 2002 10:53:44 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id KAA02721
	for tsvwg-archive@odin.ietf.org; Fri, 7 Jun 2002 10:54:17 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id KAA00306;
	Fri, 7 Jun 2002 10:28:49 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id NAA18721
	for <tsvwg@optimus.ietf.org>; Mon, 3 Jun 2002 13:24:33 -0400 (EDT)
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 NAA24712
	for <tsvwg@ietf.org>; Mon, 3 Jun 2002 13:24:04 -0400 (EDT)
Received: from mira-sjc5-2.cisco.com (IDENT:mirapoint@mira-sjc5-2.cisco.com [171.71.163.16])
	by sj-msg-core-3.cisco.com (8.12.2/8.12.2) with ESMTP id g53HNluF014934;
	Mon, 3 Jun 2002 10:23:47 -0700 (PDT)
Received: from stewart.chicago.il.us (dhcp-171-69-54-231.cisco.com [171.69.54.231])
	by mira-sjc5-2.cisco.com (Mirapoint)
	with ESMTP id ACO12771;
	Mon, 3 Jun 2002 10:24:01 -0700 (PDT)
Message-ID: <3CFBA62B.C3F756EA@stewart.chicago.il.us>
Date: Mon, 03 Jun 2002 12:23:56 -0500
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: bidulock@openss7.org
CC: tsvwg@ietf.org
References: <20020601125459.A26453@openss7.org>
Content-Type: text/plain; charset=iso-8859-1
Content-Transfer-Encoding: 8bit
Content-Transfer-Encoding: 8bit
Subject: [Tsvwg] Re: Cause code conflict.
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

Yes.. I think what we will probably do is move
the add IP up to a 0x0100 range for now.. Michael Tuexen said
something about wanting to propose a new cause code of the
implementors guide... so moving to a different range would
probably be best....

In the end IANA will figure out what the "official" number
is down the road... they may take the docuements valule that
we suggest.. but then again they may change it in the final
RFC process.... don't know...




R


"Brian F. G. Bidulock" wrote:
> 
> Randall,
> 
> There is a conflict in cause code values between the IG and ADD-IP:
> 
> draft-ietf-tsvwg-addip-sctp-05.txt: Section 3.3
> 
>     Cause Code
>     Value          Cause Code
>     ---------      ----------------
>     0x000B          Request to Delete Last Remaining IP Address.
> 
> draft-ietf-tsvwg-sctpimpguide-05.txt: Section 2.6.2
> 
>       Cause Code
>       Value           Cause Code
>       ---------      ----------------
>       ...
>       11              Restart of an association with new addresses
> 
> Please move one of these cause codes to remove the conflict.
> 
> --brian
> 
> --
> Brian F. G. Bidulock    ¦ The reasonable man adapts himself to the ¦
> bidulock@openss7.org    ¦ world; the unreasonable one persists in  ¦
> http://www.openss7.org/ ¦ trying  to adapt the  world  to himself. ¦
>                         ¦ Therefore  all  progress  depends on the ¦
>                         ¦ unreasonable man. -- George Bernard Shaw ¦

-- 
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  Fri Jun  7 10:53:49 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 KAA29162
	for <tsvwg-archive@odin.ietf.org>; Fri, 7 Jun 2002 10:53:49 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id KAA02736
	for tsvwg-archive@odin.ietf.org; Fri, 7 Jun 2002 10:54:22 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id KAA29765;
	Fri, 7 Jun 2002 10:23:02 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id TAA11051
	for <tsvwg@ns.ietf.org>; Fri, 31 May 2002 19:01:41 -0400 (EDT)
Received: from motgate.mot.com (motgate.mot.com [129.188.136.100])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id TAA06405
	for <tsvwg@ietf.org>; Fri, 31 May 2002 19:01:15 -0400 (EDT)
Received: [from pobox3.mot.com (pobox3.mot.com [10.64.251.242]) by motgate.mot.com (motgate 2.1) with ESMTP id QAA24255; Fri, 31 May 2002 16:01:38 -0700 (MST)]
Received: [from relay1.cig.mot.com (relay1.cig.mot.com [136.182.15.23]) by pobox3.mot.com (MOT-pobox3 2.0) with ESMTP id QAA25872; Fri, 31 May 2002 16:01:27 -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 g4VN1Yb08547;
	Fri, 31 May 2002 18:01:34 -0500 (CDT)
Received: from email.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 SAA03567;
	Fri, 31 May 2002 18:01:44 -0500 (CDT)
Message-ID: <3CF80138.366C854A@email.mot.com>
Date: Fri, 31 May 2002 18:03:20 -0500
From: Qiaobing Xie <qxie1@email.mot.com>
X-Mailer: Mozilla 4.77 [en] (Win98; U)
X-Accept-Language: en
MIME-Version: 1.0
To: bidulock@openss7.org
CC: tsvwg@ietf.org, Randall Stewart <randall@stewart.chicago.il.us>
Subject: Re: [Tsvwg] PRSTCP-00
References: <20020530040537.A21085@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

Hi, Brian,

Since I was the one who formulated these steps. I will try to answer
your questions.

"Brian F. G. Bidulock" wrote:
...
> 2) Page 11:
> 
>    "The data sender MUST then perform the following additional steps:"
> 
>    And
> 
>    "C2) Try to further advance..."
> 
>    C2) shows the Adv.Ack.Pt advancing over gap-acked TSNs.  What
>    happens if the peer is in the process of a renege on those
>    gap-acked TSNs?  When the SACK reneging on the gap-acked TSNs
>    arrives, the reneged TSNs will be behind the Adv.Ack.Pt.  Also,
>    when the peer receives the FORWARD TSN, it will be forced to
>    update its cumulative ack point beyond the reneged TSNs.  These
>    reneged TSN might be fully reliable TSNs which have now been
>    dropped.

Unless the renege is being misused or abused, there is no change that
the above scenario can happen. Remember that the renege is only used
when the receiver runs out of buffer when it is in the middle of
re-assembling a fragmented message *and* not all fragments are received
*and* the arrived ones are out of order. So when a TSN is reneged, it
must be true that there is at least another missing TSN that is earlier
than the reneged TSN *and* belongs to the same message. Since these two
TSNs belong to the same message, on the sender side, it would be either
that the missing TSN will stop the advance of Adv.Ack.Pt, or the reneged
TSN will be "dropped" as well (since the sender should never "drop" half
a message). Either case, there will be problem.

> 
>    Considering reneging, I don't think that the PRSCTP sender should
>    advance the Adv.Ack.Pt past gap-acked TSNs.  It should wait until
>    those TSNs are cumulatively acked (no longer renegable -- is that
>    a word?) before advancing the Adv.Ack.Pt further.

ditto.

-Qiaobing


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



From daemon@optimus.ietf.org  Fri Jun  7 10:54:17 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 KAA29194
	for <tsvwg-archive@odin.ietf.org>; Fri, 7 Jun 2002 10:54:17 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id KAA02751
	for tsvwg-archive@odin.ietf.org; Fri, 7 Jun 2002 10:54:50 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id KAA29650;
	Fri, 7 Jun 2002 10:22:54 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id IAA22885
	for <tsvwg@optimus.ietf.org>; Thu, 30 May 2002 08:35:39 -0400 (EDT)
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 IAA01506
	for <tsvwg@ietf.org>; Thu, 30 May 2002 08:35:11 -0400 (EDT)
Received: from mira-sjc5-2.cisco.com (IDENT:mirapoint@mira-sjc5-2.cisco.com [171.71.163.16])
	by sj-msg-core-1.cisco.com (8.12.2/8.12.2) with ESMTP id g4UCZ2Hs000342;
	Thu, 30 May 2002 05:35:02 -0700 (PDT)
Received: from stewart.chicago.il.us (hrn2-dhcp-161-44-87-122.cisco.com [161.44.87.122])
	by mira-sjc5-2.cisco.com (Mirapoint)
	with ESMTP id ACN37101;
	Thu, 30 May 2002 05:35:02 -0700 (PDT)
Message-ID: <3CF61C6D.35FD292C@stewart.chicago.il.us>
Date: Thu, 30 May 2002 07:34:54 -0500
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: bidulock@openss7.org
CC: tsvwg@ietf.org, conrad@acm.org
References: <20020530040537.A21085@openss7.org>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-Transfer-Encoding: 7bit
Subject: [Tsvwg] Re: PRSTCP-00
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

"Brian F. G. Bidulock" wrote:
> 
> Randall,
> 
> Two comments and a question:
> 
> 1) My compliments on an excellent draft.

Thanks... this was not only my work .. but the excellent 
ideas and help of my co-authors as well... 

> 
> 2) Page 11:
> 
>    "The data sender MUST then perform the following additional steps:"
> 
>    And
> 
>    "C2) Try to further advance..."
> 
>    C2) shows the Adv.Ack.Pt advancing over gap-acked TSNs.  What
>    happens if the peer is in the process of a renege on those
>    gap-acked TSNs?  When the SACK reneging on the gap-acked TSNs
>    arrives, the reneged TSNs will be behind the Adv.Ack.Pt.  Also,
>    when the peer receives the FORWARD TSN, it will be forced to
>    update its cumulative ack point beyond the reneged TSNs.  These
>    reneged TSN might be fully reliable TSNs which have now been
>    dropped.

Hmm... I had not thought in detail on this case...

Now so I send

Last SACK
cum-ack=90
Gap=92-94
Gap=96-99

and I now (as the sender) want
to skip to 101..
FWD-TSN(101)------->

                   but in the mean time my peer is doing
                      <----SACK(90) ... NO GAPS..


If I do the above (as the current draft states) when the
SACK arrives the SACK is the SAME point as the CUM-ACK... so
It is not old or behind... I don't strike anything... and I would
in theory go to put a mark on the 92-94 and the 96-99 .. but they
are already marked as being skipped...

Hmmm... I am not sure I see a problem... what I think would
happen is that we would see 90, try to advance the cum-ack
and once again generate a 

FWD-TSN(101)-------------> 

in response to the SACK(90)...

Am I missing something?

> 
>    Considering reneging, I don't think that the PRSCTP sender should
>    advance the Adv.Ack.Pt past gap-acked TSNs.  It should wait until
>    those TSNs are cumulatively acked (no longer renegable -- is that
>    a word?) before advancing the Adv.Ack.Pt further.

I am not sure why.. maybe I miss something? can you explain?

> 
> Q) I understand that it is only one possible PRSCTP user interface
>    methodology, but what is the use of a transport-based "timed
>    reliability"?  I agree that application timed reliability (from
>    transport user to transport user) might be useful, but this
>    concept of "timed reliability" does not take into account the
>    amount of time that data waits queued in the peer's receive queue
>    before being delivered to the application.  As such, it only
>    really times the amount of time in the send and retransmission
>    queues which is related to transmit congestion only: when an

True... if the peer is NOT READING ... this time is not
accounted for...


>    association becomes congested, data with the shortest lifetimes
>    will be dropped.  It also appears to mean that data with the
>    longer lifetimes have higher priority in times of congestion.
>    With a mix of partial and full reliability data, all the partial
>    data will be dropped in times of congestion and all the full
>    reliability data will queue indefinitely waiting for the congestion
>    to abate.  What has this to do with "reliability"?  It appears
>    more reminicent of the Unix "nice" command.  Unfortunately,
>    setting the priority of a message by setting its lifetime
>    parameter would require a deep understanding of the queueing
>    model and the traffic distribution of all other messages in all
>    other streams.  It seems unlikely that a useful value could be
>    determined except for the most narrow of applications.
> 

I don't know if I agree here... I am not strongly married to
this particular service but I do think it is useful. I can see
sending a message that is going to be useless in 4 seconds from
the time I send it.. so unless I get it ACK'd by the
peer there is NO chance the peer can use it..

If the peer is not reading and I have this timelness constraint
I see this as more of a bug in the reader than a issue with
it not reading...

I think there are quite a few uses for this service but I
will let my co-authors comment on this.. 

I like a buffer constrained model myself...

>    Is there a good real-world example of where transport-based
>    "timed reliability" would be most useful?  The previous concept
>    of limiting the number of retransmissions appeared far more
>    useful for such things as RTP/SCTP (e.g. number of retransmission
>    set to zero).

I will let Phil respond to this... he may have some other ideas.

I will be writting a service here shortly on buffer contrained
use of PR-SCTP... more targeted for ipfix...


> 
> Well, suppose thats really a bunch of comments and a bunch of
> questiosns...
> 


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  Fri Jun  7 10:55:45 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 KAA29243
	for <tsvwg-archive@odin.ietf.org>; Fri, 7 Jun 2002 10:55:44 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id KAA02816
	for tsvwg-archive@odin.ietf.org; Fri, 7 Jun 2002 10:56:17 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id KAA00349;
	Fri, 7 Jun 2002 10:28:53 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id TAA09405
	for <tsvwg@ns.ietf.org>; Mon, 3 Jun 2002 19:37:08 -0400 (EDT)
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 TAA04331
	for <tsvwg@ietf.org>; Mon, 3 Jun 2002 19:36:37 -0400 (EDT)
Received: from mira-sjc5-2.cisco.com (IDENT:mirapoint@mira-sjc5-2.cisco.com [171.71.163.16])
	by sj-msg-core-1.cisco.com (8.12.2/8.12.2) with ESMTP id g53NaaHs008280;
	Mon, 3 Jun 2002 16:36:36 -0700 (PDT)
Received: from stewart.chicago.il.us (dhcp-171-69-54-231.cisco.com [171.69.54.231])
	by mira-sjc5-2.cisco.com (Mirapoint)
	with ESMTP id ACO26368;
	Mon, 3 Jun 2002 16:36:34 -0700 (PDT)
Message-ID: <3CFBFD7D.AAA5632F@stewart.chicago.il.us>
Date: Mon, 03 Jun 2002 18:36:29 -0500
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: "Armando L. Caro Jr." <acaro@mail.eecis.udel.edu>
CC: "Brian F. G. Bidulock" <bidulock@openss7.org>, tsvwg@ietf.org
Subject: Re: [Tsvwg] SCTP IG (SACKs)
References: <Pine.GSO.4.33.0206031620140.23103-100000@ren.eecis.udel.edu>
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

Armando/Brian:

Being bored by a process lecture.. I have a few moments :>

I see no reason to make this change... because .. as I said Lixia
already provided for this...

Page 82 of RFC2960 right before 6.8:

" 
   The maximum number of Gap Ack Blocks that can be reported within a
   single SACK chunk is limited by the current path MTU.  When a single
   SACK can not cover all the Gap Ack Blocks needed to be reported due
   to the MTU limitation, the endpoint MUST send only one SACK,
   reporting the Gap Ack Blocks from the lowest to highest TSNs, within
   the size limit set by the MTU, and leave the remaining highest TSN
   numbers unacknowledged.

"

the MUST in the impl-guide just makes it so you MUST send Gap ack
blocks.
The above wording still applies and is NOT effected by this change..


We may want to add a bit to the above to state that duplicates are
dropped first before gap ack blocks... that is not specified but
I think it is implied already...


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  Fri Jun  7 10:57:29 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 KAA29316
	for <tsvwg-archive@odin.ietf.org>; Fri, 7 Jun 2002 10:57:29 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id KAA02924
	for tsvwg-archive@odin.ietf.org; Fri, 7 Jun 2002 10:58:02 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id KAA29708;
	Fri, 7 Jun 2002 10:22:58 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id RAA06772
	for <tsvwg@ns.ietf.org>; Fri, 31 May 2002 17:27:31 -0400 (EDT)
Received: from motgate4.mot.com (motgate4.mot.com [144.189.100.102])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id RAA05237
	for <tsvwg@ietf.org>; Fri, 31 May 2002 17:27:06 -0400 (EDT)
Received: [from pobox4.mot.com (pobox4.mot.com [10.64.251.243]) by motgate4.mot.com (motgate4 2.1) with ESMTP id OAA13771; Fri, 31 May 2002 14:27:30 -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 OAA16834; Fri, 31 May 2002 14:27:30 -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 g4VLRRb02355;
	Fri, 31 May 2002 16:27:27 -0500 (CDT)
Received: from email.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 QAA03507;
	Fri, 31 May 2002 16:27:47 -0500 (CDT)
Message-ID: <3CF7EB37.E3B806D8@email.mot.com>
Date: Fri, 31 May 2002 16:29:27 -0500
From: Qiaobing Xie <qxie1@email.mot.com>
X-Mailer: Mozilla 4.77 [en] (Win98; U)
X-Accept-Language: en
MIME-Version: 1.0
To: Ivan.Arias-Rodriguez@nokia.com
CC: randall@stewart.chicago.il.us, tsvwg@ietf.org
Subject: Re: [Tsvwg] Comments about prsctp-00
References: <58AEDE27D48F884285007ABF827FF2635980AF@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

Hi, Ivan,

...snip...
> 
> > Yes, agree with Randy here. To report gaps or a full list of "dropped"
> > TSNs in FORWARD TSN chunk will not be that useful because the gaps from
> > sender's view is not always the same as that from the receiver's view
> > and what really matters is what the receiver sees (this is
> > fundamentally  different from a SACK where what is reported in a SACK is the
> > receiver's view).
> >
> > For example, when the sender reports a list of "dropped" TNSs, some of
> > the "dropped" TSNs might have already been correctly received
> > and passed on to the UPL. So whether the receiver will find the additional gap
> > information useful is unclear.
> 
>         Well, you are just telling the receiver which TSNs will not retransmitted anymore... If you already got them, it does not matter. If you are still missing any of those dropped TSNs, then you should discard the user messages with at least one TSN dropped...
> 
>         But ignoring those gaps would not hurt you either (you would be exactly as we are now)...

This is exactly my point: reporting gaps is just an optimization that
*may* help the receiver *occassionally*. But the cost of doing that is
increased design and implementation complexity on both ends. In my
opinion this is hardly justifiable. Adding the following negotiations
makes the proposition even worse.

> 
>         I know you won't like this, but you could even state if you support gaps (either at the sender or receiver side) inside the Forward-TSN-Supported parameter... Thus:
> 
>         - If the INIT sender supports everything, you may send gaps if you support them at the sender side, and state in your INIT ACK if you want the INIT sender to send you gaps.
> 
>         - If the INIT sender supports gaps at the sender side, you may state in your INIT ACK if you want it to send you those gaps. But you won't send gaps as it does not support them.
> 
>         - If the INIT sender supports gaps at the receiver side, the INIT ACK sender may send them...
> 
>         It would be another level of negotiation... :-/

regards,
-Qiaobing


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



From daemon@optimus.ietf.org  Fri Jun  7 10:58:12 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 KAA29381
	for <tsvwg-archive@odin.ietf.org>; Fri, 7 Jun 2002 10:58:12 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id KAA02952
	for tsvwg-archive@odin.ietf.org; Fri, 7 Jun 2002 10:58:44 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id KAA29834;
	Fri, 7 Jun 2002 10:23:08 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id TAA11171
	for <tsvwg@ns.ietf.org>; Fri, 31 May 2002 19:05:54 -0400 (EDT)
Received: from motgate3.mot.com (motgate3.mot.com [144.189.100.103])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id TAA06467
	for <tsvwg@ietf.org>; Fri, 31 May 2002 19:05:28 -0400 (EDT)
Received: [from pobox.mot.com (pobox.mot.com [129.188.137.100]) by motgate3.mot.com (motgate3 2.1) with ESMTP id QAA27211; Fri, 31 May 2002 16:05:41 -0700 (MST)]
Received: [from relay1.cig.mot.com (relay1.cig.mot.com [136.182.15.23]) by pobox.mot.com (MOT-pobox 2.0) with ESMTP id QAA14754; Fri, 31 May 2002 16:05:53 -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 g4VN5ob08789;
	Fri, 31 May 2002 18:05:50 -0500 (CDT)
Received: from email.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 SAA03576;
	Fri, 31 May 2002 18:06:10 -0500 (CDT)
Message-ID: <3CF80246.37DFA47F@email.mot.com>
Date: Fri, 31 May 2002 18:07:50 -0500
From: Qiaobing Xie <qxie1@email.mot.com>
X-Mailer: Mozilla 4.77 [en] (Win98; U)
X-Accept-Language: en
MIME-Version: 1.0
To: "Armando L. Caro Jr." <acaro@mail.eecis.udel.edu>
CC: tsvwg@ietf.org
Subject: Re: [Tsvwg] PRSTCP-00
References: <Pine.GSO.4.33.0205311718360.8809-100000@ren.eecis.udel.edu>
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

"Armando L. Caro Jr." wrote:
....
> >    Considering reneging, I don't think that the PRSCTP sender should
> >    advance the Adv.Ack.Pt past gap-acked TSNs.  It should wait until
> >    those TSNs are cumulatively acked (no longer renegable -- is that
> >    a word?) before advancing the Adv.Ack.Pt further.
> 
> Yes, I agree with you. There shouldn't be a change in performance
> either. The receiver sees the cum ack change, moves it even further
> if possible, and conveys the info to the sender via a SACK.

Not necessarily... I am not convinced at all that there is a conflict
between PR-SCTP dropping TSNs and the renege function. See my response
to Brian sent a min ago.

-Qiaobing


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



From daemon@optimus.ietf.org  Fri Jun  7 10:59:13 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 KAA29445
	for <tsvwg-archive@odin.ietf.org>; Fri, 7 Jun 2002 10:59:13 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id KAA02992
	for tsvwg-archive@odin.ietf.org; Fri, 7 Jun 2002 10:59:45 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id KAA29840;
	Fri, 7 Jun 2002 10:23:10 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id TAA11372
	for <tsvwg@ns.ietf.org>; Fri, 31 May 2002 19:09:22 -0400 (EDT)
Received: from motgate3.mot.com (motgate3.mot.com [144.189.100.103])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id TAA06536
	for <tsvwg@ietf.org>; Fri, 31 May 2002 19:08:55 -0400 (EDT)
Received: [from pobox.mot.com (pobox.mot.com [129.188.137.100]) by motgate3.mot.com (motgate3 2.1) with ESMTP id QAA28632; Fri, 31 May 2002 16:09:05 -0700 (MST)]
Received: [from relay1.cig.mot.com (relay1.cig.mot.com [136.182.15.23]) by pobox.mot.com (MOT-pobox 2.0) with ESMTP id QAA15591; Fri, 31 May 2002 16:09:17 -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 g4VN9Eb08939;
	Fri, 31 May 2002 18:09:14 -0500 (CDT)
Received: from email.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 SAA03582;
	Fri, 31 May 2002 18:09:34 -0500 (CDT)
Message-ID: <3CF80312.9E8F2A3F@email.mot.com>
Date: Fri, 31 May 2002 18:11:14 -0500
From: Qiaobing Xie <qxie1@email.mot.com>
X-Mailer: Mozilla 4.77 [en] (Win98; U)
X-Accept-Language: en
MIME-Version: 1.0
To: bidulock@openss7.org, tsvwg@ietf.org
CC: Randall Stewart <randall@stewart.chicago.il.us>
Subject: Re: [Tsvwg] PRSTCP-00
References: <20020530040537.A21085@openss7.org> <3CF80138.366C854A@email.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

> Unless the renege is being misused or abused, there is no change that
> the above scenario can happen. Remember that the renege is only used
> when the receiver runs out of buffer when it is in the middle of
> re-assembling a fragmented message *and* not all fragments are received
> *and* the arrived ones are out of order. So when a TSN is reneged, it
> must be true that there is at least another missing TSN that is earlier
> than the reneged TSN *and* belongs to the same message. Since these two
> TSNs belong to the same message, on the sender side, it would be either
> that the missing TSN will stop the advance of Adv.Ack.Pt, or the reneged
> TSN will be "dropped" as well (since the sender should never "drop" half
> a message). Either case, there will be problem.
                                   ^^^^^^^^^^^^^
                              ...will be *no* problem...


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



From daemon@optimus.ietf.org  Fri Jun  7 10:59:36 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 KAA29483
	for <tsvwg-archive@odin.ietf.org>; Fri, 7 Jun 2002 10:59:32 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id LAA03038
	for tsvwg-archive@odin.ietf.org; Fri, 7 Jun 2002 11:00:04 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id KAA00983;
	Fri, 7 Jun 2002 10:31:46 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id TAA03551
	for <tsvwg@optimus.ietf.org>; Thu, 6 Jun 2002 19:03:38 -0400 (EDT)
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 TAA07891
	for <tsvwg@ietf.org>; Thu, 6 Jun 2002 19:03:02 -0400 (EDT)
Received: from mira-sjc5-2.cisco.com (IDENT:mirapoint@mira-sjc5-2.cisco.com [171.71.163.16])
	by sj-msg-core-1.cisco.com (8.12.2/8.12.2) with ESMTP id g56N32Hs023755;
	Thu, 6 Jun 2002 16:03:02 -0700 (PDT)
Received: from stewart.chicago.il.us (dhcp-171-69-54-231.cisco.com [171.69.54.231])
	by mira-sjc5-2.cisco.com (Mirapoint)
	with ESMTP id ACP20014;
	Thu, 6 Jun 2002 16:03:01 -0700 (PDT)
Message-ID: <3CFFEA20.8F5DF28A@stewart.chicago.il.us>
Date: Thu, 06 Jun 2002 18:02:56 -0500
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: bidulock@openss7.org
CC: tsvwg@ietf.org
Subject: Re: [Tsvwg] Re: [Fwd: PRSTCP-00]
References: <3CFF9108.FA4C1000@stewart.chicago.il.us> <20020606164317.A20729@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

"Brian F. G. Bidulock" wrote:
> > Hmm... I had not thought in detail on this case...
> >
> > Now so I send
> >
> > Last SACK
> > cum-ack=90
> > Gap=92-94
> > Gap=96-99
> >
> > and I now (as the sender) want
> > to skip to 101..
> > FWD-TSN(101)------->
> >
> >                    but in the mean time my peer is doing
> >                       <----SACK(90) ... NO GAPS..
> >
> >
> > If I do the above (as the current draft states) when the
> > SACK arrives the SACK is the SAME point as the CUM-ACK... so
> > It is not old or behind... I don't strike anything... and I would
> > in theory go to put a mark on the 92-94 and the 96-99 .. but they
> > are already marked as being skipped...
> 
> What if the SSNs in these TSNs are reliable?  They have been
> dropped and yet they have not been successfully delivered to
> the peer.  This compromises the reliability of all other "reliable"
> messages.
> 

Brian:

I am a bit confused.. it is probably just late in the day and also
the classes I am in :0

So could you help me with this ... I want to understand (and I am not
:<)

So in my example...

Are you saying something before 101 is NOT to be skipped i.e. reliable?

If so I would not do a fwd-tsn to 101 but to an earlier number....

If you could illustrate this with the above example it might help
me understand what you mean...


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  Fri Jun  7 11:00:08 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 LAA29535
	for <tsvwg-archive@odin.ietf.org>; Fri, 7 Jun 2002 11:00:08 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id LAA03276
	for tsvwg-archive@odin.ietf.org; Fri, 7 Jun 2002 11:00:40 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id KAA00989;
	Fri, 7 Jun 2002 10:31:47 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id MAA18657
	for <tsvwg@ns.ietf.org>; Wed, 5 Jun 2002 12:45:35 -0400 (EDT)
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 MAA06100
	for <tsvwg@ietf.org>; Wed, 5 Jun 2002 12:45:05 -0400 (EDT)
Received: from mira-sjc5-2.cisco.com (IDENT:mirapoint@mira-sjc5-2.cisco.com [171.71.163.16])
	by sj-msg-core-4.cisco.com (8.12.2/8.12.2) with ESMTP id g55GiuIZ027615;
	Wed, 5 Jun 2002 09:44:58 -0700 (PDT)
Received: from stewart.chicago.il.us (dhcp-171-69-54-231.cisco.com [171.69.54.231])
	by mira-sjc5-2.cisco.com (Mirapoint)
	with ESMTP id ACO73613;
	Wed, 5 Jun 2002 09:44:54 -0700 (PDT)
Message-ID: <3CFE4001.BCE81C14@stewart.chicago.il.us>
Date: Wed, 05 Jun 2002 11:44:49 -0500
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: bidulock@openss7.org, Qiaobing Xie <qxie1@email.mot.com>
CC: Qiaobing Xie <Qiaobing.Xie@motorola.com>,
        "Armando L. Caro Jr." <acaro@mail.eecis.udel.edu>, tsvwg@ietf.org,
        Michael Tuexen <michael.tuexen@icn.siemens.de>,
        "Michael A. Ramalho" <mramalho@cisco.com>,
        Phil Conrad <conrad@joda.cis.temple.edu>
Subject: Re: [Tsvwg] More on PRSCTP-00
References: <Pine.GSO.4.33.0205311736060.8809-100000@ren.eecis.udel.edu> <3CF80F65.58E43813@email.mot.com> <20020531195333.D11468@openss7.org> <3CFBC11C.1EFA747@email.mot.com> <20020603144342.A30667@openss7.org> <3CFC144D.65697A41@email.mot.com> <20020603215952.E30667@openss7.org> <3CFD57D8.5637E431@email.mot.com> <20020604195020.A16020@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

Brian/Qiaobing:

I have read through this interesting debate. Qiaobing and
I had a major discussion on this way back when it was
considered U-SCTP. Renee Revis first pointed this out to me.. but
from a different persepective i.e. the receiver being able to
efficently process stuck TSN's ... 

After a lengthy debate I agreed to not have the SSN/SID listed
in the FWD TSN... I did NOT think of the HOL blocking issue.. or
I would have used it as a issue (I was the one arguing for the
SSN/SID in the FWD-TSN :->).

After watching this debate.. I think I have come to the conclusion that
Brian is correct :) 

Michael R. is here... I will discuss it with him.. I want to get
the other co-author's take on this one... If the others agree with
Brian (and me :>) then I think we will put this in.... also I would
like to hear what the rest of the WG feels on this???


I think it is reasonable and not that much overhead to include JUST
the dropped streams NEXT expected SSN... You would be adding 4 bytes
for EVERY stream dropping one or more TSN's... note that this does
NOT mean 4 bytes per TSN but 4 bytes per where one or more TSN's were
dropped..


I think this is reasonable.. it improves processing AND avoids HOL
blocking..


Michael T/Phil? what do you think... Michael R.. I will talk to you at
the
next class break..



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  Fri Jun  7 11:00:29 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 LAA29555
	for <tsvwg-archive@odin.ietf.org>; Fri, 7 Jun 2002 11:00:28 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id LAA03340
	for tsvwg-archive@odin.ietf.org; Fri, 7 Jun 2002 11:01:01 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id KAA01148;
	Fri, 7 Jun 2002 10:31:59 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id TAA09476
	for <tsvwg@ns.ietf.org>; Wed, 5 Jun 2002 19:44:10 -0400 (EDT)
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 TAA28096
	for <tsvwg@ietf.org>; Wed, 5 Jun 2002 19:43:38 -0400 (EDT)
Received: from mira-sjc5-2.cisco.com (IDENT:mirapoint@mira-sjc5-2.cisco.com [171.71.163.16])
	by sj-msg-core-1.cisco.com (8.12.2/8.12.2) with ESMTP id g55NhYHs019560;
	Wed, 5 Jun 2002 16:43:34 -0700 (PDT)
Received: from stewart.chicago.il.us (dhcp-171-69-54-231.cisco.com [171.69.54.231])
	by mira-sjc5-2.cisco.com (Mirapoint)
	with ESMTP id ACO89642;
	Wed, 5 Jun 2002 16:43:33 -0700 (PDT)
Message-ID: <3CFEA220.DE96B441@stewart.chicago.il.us>
Date: Wed, 05 Jun 2002 18:43:28 -0500
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: bidulock@openss7.org
CC: Phillip Conrad <conrad@acm.org>, Qiaobing Xie <qxie1@email.mot.com>,
        Qiaobing Xie <Qiaobing.Xie@motorola.com>,
        "Armando L. Caro Jr." <acaro@mail.eecis.udel.edu>, tsvwg@ietf.org,
        Michael Tuexen <michael.tuexen@icn.siemens.de>,
        "Michael A. Ramalho" <mramalho@cisco.com>
Subject: Re: [Tsvwg] More on PRSCTP-00
References: <3CF80F65.58E43813@email.mot.com> <20020531195333.D11468@openss7.org> <3CFBC11C.1EFA747@email.mot.com> <20020603144342.A30667@openss7.org> <3CFC144D.65697A41@email.mot.com> <20020603215952.E30667@openss7.org> <3CFD57D8.5637E431@email.mot.com> <20020604195020.A16020@openss7.org> <5.1.0.14.2.20020605155749.0236b050@joda.cis.temple.edu> <3CFE8964.81F6314F@stewart.chicago.il.us> <20020605180715.E2012@openss7.org>
Content-Type: text/plain; charset=iso-8859-1
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

"Brian F. G. Bidulock" wrote:
> 
> Randall,
> 
> On Wed, 05 Jun 2002, Randall Stewart wrote:
> 
> > A few comments..
> >
> > Phillip Conrad wrote:
> > >
> > > Dear Colleagues:
> > >       I agree with Randy that Brian has raised an important issue for us to
> > > consider, and I think has explained it quite well with a very clear,
> > > illustrative example.
> > >
> > > The question is one of a tradeoff between two designs:
> > >
> > > (A) current design: simpler, fewer bits on the wire, but may (re)-introduce
> > > an HOL blocking problem
> >
> > I must disagree with your "it is simpler" ... it is NOT. The question is
> > where is the burden. Having implemented pr-sctp in the kernel of freebsd
> > NOT having the stream number/SSN in the FWD-TSN means I must keep
> > an extra set of linked lists and such to track any stream that has
> > pending stuck data. When a fwd-tsn arrives I must HUNT all of these
> > streams and look at each inbound queue.
> >
> > The counter to this is the sender (in Brian and my earlier proposal
> > that was to Q only) must track this info to send with the FWD-TSN.
> > Actually
> > less of a burden IMHO...
> 
> I agree.  I discovered this situation while implementing PR-SCTP in the
> Linux Kernel.  The normal receiver can rely on the fact that the next
> expected SSN is the last fully received SSN plus 1.  PR-SCTP without
> a forward SSN causes the receive to have to update this expected value
> with the next SSN in the reorder queue whenever there are no TSN gaps.
> This even might require that when TSN gaps are filled that the receiver
> walk all inbound stream structures to update the next expected SSN.
> In an independent implementation, I have arrived at the same conclusion:
> that the reciever has more work to do for FORWARD TSN without a forward
> SSN.
> 

Yeah.. I basically had to make a special linked list set that the
stream queues could be placed on.. a bit more efficient but a
reall hassel to maintian a "wheel" (as I call it) of streams with
pending data on it... 

> In fact the receiver has this work to do if it has ever received a
> FORWARD TSN.  This is because it is not possible for the receiver to
> know a priori which streams could be affected by this scenairo.  Therefore,
> the load on the receiver is increased even when this case does not occur.
> It is the mere possibility of its occurence which introduces the load.
> (Each time gaps close, the next expected SSN for all streams need to be
> re-examined whether there are SSN gaps or not.)

Exactly.. I constantly must place stream q;s on and off the :"wheel" no
matter if a fwd-tsn is EVER received...

> 
> On the sender side, the sender must always track the last sent SSN and
> assign the next SSN (++SSN).  Indeed the sender when generating a FORWARD
> TSN chunk would need to traverse outbound stream structures with dropped
> TSNs to formulate Forward TSNs, but this is likely to be a far shorter
> list than at the receiver and non-PR streams would never need to be
> involved.

And it must continue to KEEP this information until it is sure the
peer has received and moved the cum-ack forward..

What I end up doing on the sender side is I hold just the bookkeeping
structures of the TSN (stream seq/stream no/tsn ... etc) but I
drop the mbuf holding the data so I gain a bit of help in mbuf
utilization.

But I can't get rid of the chunk structure UNTIL the cum-ack says he
got my FWD-TSN and accepted it...

So there is NO additional info to save on the sender side and only 
do I save what I must... on the receiver side I have a LOT of extra
bookkeeping.... 


In fact with your confirmation I would claim just the opposite of
Phil's statement... that doing it the way it is NOW documented takes
MORE overhead.. only by adding the SSN/SID can you reduce this..

By the way Renee also pointed out the extra overhead for the IOS
implementation as well :>... so this makes three different
implementations
NOT 2.... 

I plan on adding this to the next pr-sctp unless Michael Tuexen
violently
agrees with Qiaobing ... 

I still await Michaels answer????


> 
> --brian
> 
> > >
> > > (B) alternative design: more complex, may require larger Forward TSN
> > > chunks, but avoids re-introducing an HOL blocking problem, and thus
> > > improves performance in some situations.
> >
> >
> > see above it is NOT more complex...
> >
> > >
> > > The question of whether to choose A or B depends on whether you believe
> > > that the situation Brian describes is a "corner case" or not.   So I've
> > > walked through Brian's example, and have determined that I don't think it
> > > is a corner case at all.   So, I'm inclined to agree with Randy and Brian
> > > that we should consider a change.
> > >
> > > But I'd like to invite others to walk through my line of reasoning to test
> > > whether I've reasoned this correctly.
> >
> >
> > I agree it is NOT a corner case..
> >
> > >
> > > My rationale that Brian's example is NOT a corner case:
> > > ========================================
> > >
> > > In Brian's example, messages are being sent round-robin on a number of
> > > streams.  This seems like a fairly reasonable use of SCTP streams, do that
> > > aspect does not strike me as unusual.
> >
> > I agree :>
> >
> > >
> > > Now for the next part of this explanation, I'm going to have to infer
> > > certain things, so, Brian, please correct me if I infer wrongly.   *One*
> > > possibility for the sequence of events leading to the
> > > situation Brian describes is that there is a burst loss of data chunks
> > > TSN110 through TSN119 (ten chunks, but possibly less than ten packets..
> > > perhaps even only ONE packet if the chunks are small enough to be
> > > bundled!)... and then data chunks TSN120 through TSN128 make it through
> > > OK.   This does not strike me as particularly unusual; a loss of a single
> > > packet followed by a successful packet is sufficient to cause this situation.
> > >
> >
> > Or the other possiblility would be a set of bundled chunks in one packet
> > being lost.. either way is the same result I would think..
> >
> > > The next thing that happens is that TSNs 110 through 118 are "dropped" in
> > > the technical sense of that word; see the defn in the current draft.  This
> > > means that the sender gives up on sending them, and issues a forward
> > > TSN.   However, no Forward TSN is issued for TSN119.  This may be because
> > > the messages on stream 9 are treated as Reliable.   This is NOT an unusual
> > > situation; it is perfectly reasonably to assume that an application might
> > > have "k" streams with data that can be dropped, and "n - k" streams of data
> > > that must be treated as reliable.  In the case of Brian's example, n is 10,
> > > and k is 9, but I think that his example will still produce similar results
> > > even for other values of "k" and "n".
> > >
> >
> > Note.. streams themselves are not reliable or partially reliable. With
> > the new draft it is the message itself which is sent with the
> > reliablilty
> > set to "fully reliable". One possilbe implementation would be to make
> > things correlate to a particular stream.. but this is just an
> > implementation
> > option..
> >
> > > Bottom line:
> > > ========
> > >
> > > I don't think its a corner case at all.   We have all the PR streams
> > > potentially being HOL blocked by loss of a single packet.  I think we need
> > > to fix this.
> >
> > I agree..
> >
> > >
> > > The solution of forward SSNs is one viable solution I think.  There may be
> > > others, but that's as good as any for right now I think.   There may be
> > > another solution as well, but let me do some writing and thinking about it
> > > off line before I muddy the waters with it here... just to make sure it is
> > > a viable alternative.
> >
> > agreed
> >
> >
> > >
> > > Randy... I'm willing to make the text changes and send a proposed version
> > > around to the co-authors for comment if there is consensus to move in that
> > > direction.   Let me know.
> > >
> >
> > I will get it out this weekend if I may.. there are some interesting
> > things that I have thought about that must be covered... I had something
> > like this in an old u-sctp draft (during Q and I's debate earlier :>)
> >
> >
> >
> > >
> > > (Side note: I was worried that the "technical definition" of the word
> > > "dropped" would cause confusion, and it already has.   Perhaps should use
> > > another word in the draft to describe the operation where the sender
> > > decides to no longer transmit and/or retransmit a specific data chunk.  How
> > > about the term "abandoned" in place of "dropped"?)
> >
> >
> > Sounds good..
> >
> > note, Michael R also agrees .. or seems to .. we will talk more..
> >
> > R
> >
> >
> > >
> > > /ptc
> > >
> > > At 11:44 AM 6/5/2002 -0500, Randall Stewart wrote:
> > > >Brian/Qiaobing:
> > > >
> > > >I have read through this interesting debate. Qiaobing and
> > > >I had a major discussion on this way back when it was
> > > >considered U-SCTP. Renee Revis first pointed this out to me.. but
> > > >from a different persepective i.e. the receiver being able to
> > > >efficently process stuck TSN's ...
> > > >
> > > >After a lengthy debate I agreed to not have the SSN/SID listed
> > > >in the FWD TSN... I did NOT think of the HOL blocking issue.. or
> > > >I would have used it as a issue (I was the one arguing for the
> > > >SSN/SID in the FWD-TSN :->).
> > > >
> > > >After watching this debate.. I think I have come to the conclusion that
> > > >Brian is correct :)
> > > >
> > > >Michael R. is here... I will discuss it with him.. I want to get
> > > >the other co-author's take on this one... If the others agree with
> > > >Brian (and me :>) then I think we will put this in.... also I would
> > > >like to hear what the rest of the WG feels on this???
> > > >
> > > >
> > > >I think it is reasonable and not that much overhead to include JUST
> > > >the dropped streams NEXT expected SSN... You would be adding 4 bytes
> > > >for EVERY stream dropping one or more TSN's... note that this does
> > > >NOT mean 4 bytes per TSN but 4 bytes per where one or more TSN's were
> > > >dropped..
> > > >
> > > >
> > > >I think this is reasonable.. it improves processing AND avoids HOL
> > > >blocking..
> > > >
> > > >
> > > >Michael T/Phil? what do you think... Michael R.. I will talk to you at
> > > >the
> > > >next class break..
> > > >
> > > >
> > > >
> > > >R
> > > >
> > > >--
> > > >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)
> 
> --
> Brian F. G. Bidulock    ¦ The reasonable man adapts himself to the ¦
> bidulock@openss7.org    ¦ world; the unreasonable one persists in  ¦
> http://www.openss7.org/ ¦ trying  to adapt the  world  to himself. ¦
>                         ¦ Therefore  all  progress  depends on the ¦
>                         ¦ unreasonable man. -- George Bernard Shaw ¦

-- 
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  Fri Jun  7 11:02:43 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 LAA29669
	for <tsvwg-archive@odin.ietf.org>; Fri, 7 Jun 2002 11:02:42 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id LAA03685
	for tsvwg-archive@odin.ietf.org; Fri, 7 Jun 2002 11:03:15 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id KAA01073;
	Fri, 7 Jun 2002 10:31:53 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id NAA18814
	for <tsvwg@optimus.ietf.org>; Mon, 3 Jun 2002 13:26:32 -0400 (EDT)
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 NAA24775
	for <tsvwg@ietf.org>; Mon, 3 Jun 2002 13:26:01 -0400 (EDT)
Received: from mira-sjc5-2.cisco.com (IDENT:mirapoint@mira-sjc5-2.cisco.com [171.71.163.16])
	by sj-msg-core-4.cisco.com (8.12.2/8.12.2) with ESMTP id g53HPrIb018600;
	Mon, 3 Jun 2002 10:25:55 -0700 (PDT)
Received: from stewart.chicago.il.us (dhcp-171-69-54-231.cisco.com [171.69.54.231])
	by mira-sjc5-2.cisco.com (Mirapoint)
	with ESMTP id ACO12842;
	Mon, 3 Jun 2002 10:25:52 -0700 (PDT)
Message-ID: <3CFBA69B.FB3FE86E@stewart.chicago.il.us>
Date: Mon, 03 Jun 2002 12:25:47 -0500
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: bidulock@openss7.org
CC: tsvwg@ietf.org
Subject: Re: [Tsvwg] SCTP IG (SACKs)
References: <20020601152128.B28346@openss7.org>
Content-Type: text/plain; charset=iso-8859-1
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

Brian:

Don't have time to check now but I think this case is
already covered... it tells you to truncate the highest
gaps off...

I think Lixia wrote on this issue in the original RFC...

I am in a class and have been away from email for quite a bit
so I am backlogged.... I will have a look when things calm down
for me a bit :>

R

"Brian F. G. Bidulock" wrote:
> 
> Randall,
> 
> draft-ietf-tsvwg-sctpimguide-05.txt Sction 2.7 places a MUST on
> sending gap reports.  Before when this was a SHOULD, it was obvious
> that gap ack blocks in a SACK which exceeded an MTU could be
> truncated.  Now it is not so clear.  What should an implementation
> do if the number of gap ack blocks in a SACK exceeds the path MTU?
> Truncate?  Force IP fragmentation?
> 
> --brian
> 
> --
> Brian F. G. Bidulock    ¦ The reasonable man adapts himself to the ¦
> bidulock@openss7.org    ¦ world; the unreasonable one persists in  ¦
> http://www.openss7.org/ ¦ trying  to adapt the  world  to himself. ¦
>                         ¦ Therefore  all  progress  depends on the ¦
>                         ¦ unreasonable man. -- George Bernard Shaw ¦
> 
> _______________________________________________
> 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  Fri Jun  7 11:51:30 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 LAA01653
	for <tsvwg-archive@odin.ietf.org>; Fri, 7 Jun 2002 11:51:26 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id LAA06028
	for tsvwg-archive@odin.ietf.org; Fri, 7 Jun 2002 11:51:57 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id KAA01276;
	Fri, 7 Jun 2002 10:32:12 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id MAA18693
	for <tsvwg@ns.ietf.org>; Wed, 5 Jun 2002 12:47:04 -0400 (EDT)
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 MAA06142
	for <tsvwg@ietf.org>; Wed, 5 Jun 2002 12:46:35 -0400 (EDT)
Received: from mira-sjc5-2.cisco.com (IDENT:mirapoint@mira-sjc5-2.cisco.com [171.71.163.16])
	by sj-msg-core-1.cisco.com (8.12.2/8.12.2) with ESMTP id g55GkTHs013833;
	Wed, 5 Jun 2002 09:46:29 -0700 (PDT)
Received: from stewart.chicago.il.us (dhcp-171-69-54-231.cisco.com [171.69.54.231])
	by mira-sjc5-2.cisco.com (Mirapoint)
	with ESMTP id ACO73677;
	Wed, 5 Jun 2002 09:46:29 -0700 (PDT)
Message-ID: <3CFE4060.BF983A37@stewart.chicago.il.us>
Date: Wed, 05 Jun 2002 11:46:25 -0500
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: bidulock@openss7.org
CC: "Armando L. Caro Jr." <acaro@mail.eecis.udel.edu>,
        Qiaobing Xie <qxie1@email.mot.com>,
        Qiaobing Xie <Qiaobing.Xie@motorola.com>, tsvwg@ietf.org
Subject: Re: [Tsvwg] More on PRSCTP-00
References: <20020603144342.A30667@openss7.org> <Pine.GSO.4.33.0206031603240.23103-100000@ren.eecis.udel.edu> <20020603153829.C30667@openss7.org>
Content-Type: text/plain; charset=iso-8859-1
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

"Brian F. G. Bidulock" wrote:
> 
> Armando,
> 
> Yes!  Maybe the "??" field could be used to indicate the total
> number of dropped TSNs (of course maxing out at 0xffff) reported
> by this FORWARD TSN chunk.  The receiver might be able to use that
> to determine whether it has accounted for all drops with the
> provided Forward SSNs.
> 
> The procedure would be upon receipt of a FORWARD TSN chunk to perform
> the cummulative ack adjustments already provided in the draft, however,
> the next expected stream sequence number for each inbound stream would
> be adjusted forward to the SSN contained in the Forward SSNs.

After delivering every stuck SID that could be delivered ... right?

R

> 
> --brian
> 
> On Mon, 03 Jun 2002, Armando L. Caro Jr. wrote:
> 
> > On Mon, 3 Jun 2002, Brian F. G. Bidulock wrote:
> >
> > > Sending a Forward-SSN for only those streams that drop ordered packets
> > > within the cumm-ack to Forward-TSN range would require far less data
> > > in the FORWARD TSN, yet provide only what is required to deliver around
> > > drops, takes 4 bytes less data than your triplet approach, and could still
> > > be truncated when required.
> >
> > So you are proposing a new forward tsn chunk like the following?
> >
> >          0                   1                   2                   3
> >          0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
> >         +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
> >         |1 1 0 0 0 0 0 0|  Chunk Flags  |0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0|
> >         +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
> >         |                      New Cumulative TSN                       |
> >         +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
> >         |      Num of Forward SSNs = N  |         ??                    |
> >         +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
> >         |      Stream Identifier S      |   Stream Sequence Number n    |
> >         +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
> >         /                                                               /
> >         \                              ...                              \
> >         /                                                               /
> >         +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
> >         |      Stream Identifier S      |   Stream Sequence Number n    |
> >         +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
> >
> > > I would say MUST, because this is as important to SCTP's ability to
> > > avoid this interstream blocking as are GAP reports.
> >
> > I think MUST is good, otherwise PR-SCTP may not provided much better
> > performance than plain SCTP.
> >
> > ~armando
> >
> > 0--                                                --0
> > | Armando L. Caro Jr.     |       acaro@cis.udel.edu |
> > | University of Delaware  |  www.cis.udel.edu/~acaro |
> > 0--                                                --0
> >
> >
> >
> > _______________________________________________
> > tsvwg mailing list
> > tsvwg@ietf.org
> > https://www1.ietf.org/mailman/listinfo/tsvwg
> 
> --
> Brian F. G. Bidulock    ¦ The reasonable man adapts himself to the ¦
> bidulock@openss7.org    ¦ world; the unreasonable one persists in  ¦
> http://www.openss7.org/ ¦ trying  to adapt the  world  to himself. ¦
>                         ¦ Therefore  all  progress  depends on the ¦
>                         ¦ unreasonable man. -- George Bernard Shaw ¦
> 
> _______________________________________________
> 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  Fri Jun  7 11:52:12 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 LAA01693
	for <tsvwg-archive@odin.ietf.org>; Fri, 7 Jun 2002 11:52:12 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id LAA06117
	for tsvwg-archive@odin.ietf.org; Fri, 7 Jun 2002 11:52:43 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id KAA01186;
	Fri, 7 Jun 2002 10:32:05 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id TAA09255
	for <tsvwg@ns.ietf.org>; Wed, 5 Jun 2002 19:35:41 -0400 (EDT)
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 TAA27615
	for <tsvwg@ietf.org>; Wed, 5 Jun 2002 19:35:10 -0400 (EDT)
Received: from mira-sjc5-2.cisco.com (IDENT:mirapoint@mira-sjc5-2.cisco.com [171.71.163.16])
	by sj-msg-core-2.cisco.com (8.12.2/8.12.2) with ESMTP id g55NZ5PI004271;
	Wed, 5 Jun 2002 16:35:05 -0700 (PDT)
Received: from stewart.chicago.il.us (dhcp-171-69-54-231.cisco.com [171.69.54.231])
	by mira-sjc5-2.cisco.com (Mirapoint)
	with ESMTP id ACO89398;
	Wed, 5 Jun 2002 16:35:05 -0700 (PDT)
Message-ID: <3CFEA024.394BC38F@stewart.chicago.il.us>
Date: Wed, 05 Jun 2002 18:35:01 -0500
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 <qxie1@email.mot.com>
CC: Phillip Conrad <conrad@acm.org>, bidulock@openss7.org,
        Qiaobing Xie <Qiaobing.Xie@motorola.com>,
        "Armando L. Caro Jr." <acaro@mail.eecis.udel.edu>, tsvwg@ietf.org,
        Michael Tuexen <michael.tuexen@icn.siemens.de>,
        "Michael A. Ramalho" <mramalho@cisco.com>
Subject: Re: [Tsvwg] More on PRSCTP-00
References: <Pine.GSO.4.33.0205311736060.8809-100000@ren.eecis.udel.edu>
		 <3CF80F65.58E43813@email.mot.com> <20020531195333.D11468@openss7.org>
		 <3CFBC11C.1EFA747@email.mot.com> <20020603144342.A30667@openss7.org>
		 <3CFC144D.65697A41@email.mot.com> <20020603215952.E30667@openss7.org>
		 <3CFD57D8.5637E431@email.mot.com> <20020604195020.A16020@openss7.org> <5.1.0.14.2.20020605155749.0236b050@joda.cis.temple.edu> <3CFE8A04.24188859@email.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:
> 
> Phil,
> 
> Please take my little challenges below as a way to help you making your
> case tighter :-)
> 
> Phillip Conrad wrote:
> ...
> > The next thing that happens is that TSNs 110 through 118 are "dropped" in
> > the technical sense of that word; see the defn in the current draft.  This
> > means that the sender gives up on sending them, and issues a forward
> > TSN.   However, no Forward TSN is issued for TSN119.  This may be because
> > the messages on stream 9 are treated as Reliable.   This is NOT an unusual
> > situation;
> 
> What unusual here is that you seem to imply that you send the Fwd-TSN
> without retransmitting the *reliable but lost* TSN119. Remember, TSN119
> is lost *together* with the other 8 TSNs, and the decision point of
> sending Fwd-TSN is *deliberately* crafted to be the same as that for
> normal retransmission.
> 


You make a rather larger assumption here Q... i.e. that TSN119 was
lost at the SAME time as the others... this does NOT have to be
the case... it would be if you had a single bundled packet loss.. but
I could also see it be due to multiple packet losses in the network.
This is very very common... and they do NOT have to be at the same
time.... so say the FR algorithm may trigger on the ones up to 118 a
SACK
or so AHEAD of 119's...

I have ALWAYS thought that having the streams info of dropped streams
in the FWD-TSN would make it MUCH easier to implement for the reciever
and not very complex for the sender (since most of the state is
already present anyway)... you know this... but the HOL blocking issue
gives a much more concrete reason to do it...


> Therefore, what should be usual is that when the sender reached the
> conclusion that it needs to send the Fwd-TSN for the 8 "dropped" TSNs,
> it MUST also have reached the conclusion at the same moment to
> retransmit TSN119 (furthermore, I'd think in more implementations the
> Fwd-TSN will also be piggybacked with the retrans TSN119).
> 

No, I don't think you can make this assumption.. see above... only
if TSN119 was bundled with the rest... 

> The stream blocking analysis (by Brian) only holds if the Fwd-TSN
> arrives without the retransmitted TSN119. But if both arrive at the
> same, would the analysis still hold?
> 

of course not.. since then everything is resolved..




> > ...We have all the PR streams
> > potentially being HOL blocked by loss of a single packet.  I think we need
> > to fix this.
> 
> Interesting point. But we seem to accept the same thing (streams
> potentially being HOL blocked by loss of a single packet) in base SCTP.
> How come?
> 
But a single loss can never block more than ONE stream.. I think
this is the difference... 

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  Fri Jun  7 12:04:33 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 MAA02168
	for <tsvwg-archive@odin.ietf.org>; Fri, 7 Jun 2002 12:04:33 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id MAA08631
	for tsvwg-archive@odin.ietf.org; Fri, 7 Jun 2002 12:05:04 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id KAA01346;
	Fri, 7 Jun 2002 10:32:20 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id QAA00193
	for <tsvwg@ns.ietf.org>; Wed, 5 Jun 2002 16:39:56 -0400 (EDT)
Received: from smtp.comcast.net (smtp.comcast.net [24.153.64.2])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id QAA18434
	for <tsvwg@ietf.org>; Wed, 5 Jun 2002 16:39:24 -0400 (EDT)
Received: from livingroom.acm.org
 (pcp01481250pcs.ncstl01.de.comcast.net [68.82.71.168])
 by mtaout06.icomcast.net
 (iPlanet Messaging Server 5.1 HotFix 0.8 (built May 13 2002))
 with ESMTP id <0GX900ICL2MKHG@mtaout06.icomcast.net> for tsvwg@ietf.org; Wed,
 05 Jun 2002 16:37:35 -0400 (EDT)
Date: Wed, 05 Jun 2002 16:37:17 -0400
From: Phillip Conrad <conrad@acm.org>
Subject: Re: [Tsvwg] More on PRSCTP-00
In-reply-to: <3CFE4001.BCE81C14@stewart.chicago.il.us>
X-Sender: phillipconrad@mail.comcast.net
To: Randall Stewart <randall@stewart.chicago.il.us>, bidulock@openss7.org,
        Qiaobing Xie <qxie1@email.mot.com>
Cc: Qiaobing Xie <Qiaobing.Xie@motorola.com>,
        "Armando L. Caro Jr." <acaro@mail.eecis.udel.edu>, tsvwg@ietf.org,
        Michael Tuexen <michael.tuexen@icn.siemens.de>,
        "Michael A. Ramalho" <mramalho@cisco.com>
Message-id: <5.1.0.14.2.20020605155749.0236b050@joda.cis.temple.edu>
MIME-version: 1.0
X-Mailer: QUALCOMM Windows Eudora Version 5.1
Content-type: text/plain; charset=us-ascii; format=flowed
Content-transfer-encoding: 7BIT
References: <Pine.GSO.4.33.0205311736060.8809-100000@ren.eecis.udel.edu>
 <3CF80F65.58E43813@email.mot.com> <20020531195333.D11468@openss7.org>
 <3CFBC11C.1EFA747@email.mot.com> <20020603144342.A30667@openss7.org>
 <3CFC144D.65697A41@email.mot.com> <20020603215952.E30667@openss7.org>
 <3CFD57D8.5637E431@email.mot.com> <20020604195020.A16020@openss7.org>
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

Dear Colleagues:
      I agree with Randy that Brian has raised an important issue for us to 
consider, and I think has explained it quite well with a very clear, 
illustrative example.

The question is one of a tradeoff between two designs:

(A) current design: simpler, fewer bits on the wire, but may (re)-introduce 
an HOL blocking problem

(B) alternative design: more complex, may require larger Forward TSN 
chunks, but avoids re-introducing an HOL blocking problem, and thus 
improves performance in some situations.

The question of whether to choose A or B depends on whether you believe 
that the situation Brian describes is a "corner case" or not.   So I've 
walked through Brian's example, and have determined that I don't think it 
is a corner case at all.   So, I'm inclined to agree with Randy and Brian 
that we should consider a change.

But I'd like to invite others to walk through my line of reasoning to test 
whether I've reasoned this correctly.

My rationale that Brian's example is NOT a corner case:
========================================

In Brian's example, messages are being sent round-robin on a number of 
streams.  This seems like a fairly reasonable use of SCTP streams, do that 
aspect does not strike me as unusual.

Now for the next part of this explanation, I'm going to have to infer 
certain things, so, Brian, please correct me if I infer wrongly.   *One* 
possibility for the sequence of events leading to the
situation Brian describes is that there is a burst loss of data chunks 
TSN110 through TSN119 (ten chunks, but possibly less than ten packets.. 
perhaps even only ONE packet if the chunks are small enough to be 
bundled!)... and then data chunks TSN120 through TSN128 make it through 
OK.   This does not strike me as particularly unusual; a loss of a single 
packet followed by a successful packet is sufficient to cause this situation.

The next thing that happens is that TSNs 110 through 118 are "dropped" in 
the technical sense of that word; see the defn in the current draft.  This 
means that the sender gives up on sending them, and issues a forward 
TSN.   However, no Forward TSN is issued for TSN119.  This may be because 
the messages on stream 9 are treated as Reliable.   This is NOT an unusual 
situation; it is perfectly reasonably to assume that an application might 
have "k" streams with data that can be dropped, and "n - k" streams of data 
that must be treated as reliable.  In the case of Brian's example, n is 10, 
and k is 9, but I think that his example will still produce similar results 
even for other values of "k" and "n".

Bottom line:
========

I don't think its a corner case at all.   We have all the PR streams 
potentially being HOL blocked by loss of a single packet.  I think we need 
to fix this.

The solution of forward SSNs is one viable solution I think.  There may be 
others, but that's as good as any for right now I think.   There may be 
another solution as well, but let me do some writing and thinking about it 
off line before I muddy the waters with it here... just to make sure it is 
a viable alternative.

Randy... I'm willing to make the text changes and send a proposed version 
around to the co-authors for comment if there is consensus to move in that 
direction.   Let me know.

Cheers,
Phill

(Side note: I was worried that the "technical definition" of the word 
"dropped" would cause confusion, and it already has.   Perhaps should use 
another word in the draft to describe the operation where the sender 
decides to no longer transmit and/or retransmit a specific data chunk.  How 
about the term "abandoned" in place of "dropped"?)

/ptc

At 11:44 AM 6/5/2002 -0500, Randall Stewart wrote:
>Brian/Qiaobing:
>
>I have read through this interesting debate. Qiaobing and
>I had a major discussion on this way back when it was
>considered U-SCTP. Renee Revis first pointed this out to me.. but
>from a different persepective i.e. the receiver being able to
>efficently process stuck TSN's ...
>
>After a lengthy debate I agreed to not have the SSN/SID listed
>in the FWD TSN... I did NOT think of the HOL blocking issue.. or
>I would have used it as a issue (I was the one arguing for the
>SSN/SID in the FWD-TSN :->).
>
>After watching this debate.. I think I have come to the conclusion that
>Brian is correct :)
>
>Michael R. is here... I will discuss it with him.. I want to get
>the other co-author's take on this one... If the others agree with
>Brian (and me :>) then I think we will put this in.... also I would
>like to hear what the rest of the WG feels on this???
>
>
>I think it is reasonable and not that much overhead to include JUST
>the dropped streams NEXT expected SSN... You would be adding 4 bytes
>for EVERY stream dropping one or more TSN's... note that this does
>NOT mean 4 bytes per TSN but 4 bytes per where one or more TSN's were
>dropped..
>
>
>I think this is reasonable.. it improves processing AND avoids HOL
>blocking..
>
>
>Michael T/Phil? what do you think... Michael R.. I will talk to you at
>the
>next class break..
>
>
>
>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  Fri Jun  7 12:07:26 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 MAA02236
	for <tsvwg-archive@odin.ietf.org>; Fri, 7 Jun 2002 12:07:26 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id MAA08756
	for tsvwg-archive@odin.ietf.org; Fri, 7 Jun 2002 12:07:58 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id KAA01317;
	Fri, 7 Jun 2002 10:32:18 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id RAA04780
	for <tsvwg@ns.ietf.org>; Wed, 5 Jun 2002 17:58:44 -0400 (EDT)
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 RAA22305
	for <tsvwg@ietf.org>; Wed, 5 Jun 2002 17:58:13 -0400 (EDT)
Received: from mira-sjc5-2.cisco.com (IDENT:mirapoint@mira-sjc5-2.cisco.com [171.71.163.16])
	by sj-msg-core-1.cisco.com (8.12.2/8.12.2) with ESMTP id g55Lw2Hs007301;
	Wed, 5 Jun 2002 14:58:02 -0700 (PDT)
Received: from stewart.chicago.il.us (dhcp-171-69-54-231.cisco.com [171.69.54.231])
	by mira-sjc5-2.cisco.com (Mirapoint)
	with ESMTP id ACO85861;
	Wed, 5 Jun 2002 14:58:00 -0700 (PDT)
Message-ID: <3CFE8964.81F6314F@stewart.chicago.il.us>
Date: Wed, 05 Jun 2002 16:57:56 -0500
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: Phillip Conrad <conrad@acm.org>
CC: bidulock@openss7.org, Qiaobing Xie <qxie1@email.mot.com>,
        Qiaobing Xie <Qiaobing.Xie@motorola.com>,
        "Armando L. Caro Jr." <acaro@mail.eecis.udel.edu>, tsvwg@ietf.org,
        Michael Tuexen <michael.tuexen@icn.siemens.de>,
        "Michael A. Ramalho" <mramalho@cisco.com>
Subject: Re: [Tsvwg] More on PRSCTP-00
References: <Pine.GSO.4.33.0205311736060.8809-100000@ren.eecis.udel.edu>
	 <3CF80F65.58E43813@email.mot.com> <20020531195333.D11468@openss7.org>
	 <3CFBC11C.1EFA747@email.mot.com> <20020603144342.A30667@openss7.org>
	 <3CFC144D.65697A41@email.mot.com> <20020603215952.E30667@openss7.org>
	 <3CFD57D8.5637E431@email.mot.com> <20020604195020.A16020@openss7.org> <5.1.0.14.2.20020605155749.0236b050@joda.cis.temple.edu>
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

A few comments..

Phillip Conrad wrote:
> 
> Dear Colleagues:
>       I agree with Randy that Brian has raised an important issue for us to
> consider, and I think has explained it quite well with a very clear,
> illustrative example.
> 
> The question is one of a tradeoff between two designs:
> 
> (A) current design: simpler, fewer bits on the wire, but may (re)-introduce
> an HOL blocking problem

I must disagree with your "it is simpler" ... it is NOT. The question is
where is the burden. Having implemented pr-sctp in the kernel of freebsd
NOT having the stream number/SSN in the FWD-TSN means I must keep
an extra set of linked lists and such to track any stream that has
pending stuck data. When a fwd-tsn arrives I must HUNT all of these
streams and look at each inbound queue.

The counter to this is the sender (in Brian and my earlier proposal
that was to Q only) must track this info to send with the FWD-TSN.
Actually
less of a burden IMHO...

> 
> (B) alternative design: more complex, may require larger Forward TSN
> chunks, but avoids re-introducing an HOL blocking problem, and thus
> improves performance in some situations.


see above it is NOT more complex... 

> 
> The question of whether to choose A or B depends on whether you believe
> that the situation Brian describes is a "corner case" or not.   So I've
> walked through Brian's example, and have determined that I don't think it
> is a corner case at all.   So, I'm inclined to agree with Randy and Brian
> that we should consider a change.
> 
> But I'd like to invite others to walk through my line of reasoning to test
> whether I've reasoned this correctly.


I agree it is NOT a corner case..

> 
> My rationale that Brian's example is NOT a corner case:
> ========================================
> 
> In Brian's example, messages are being sent round-robin on a number of
> streams.  This seems like a fairly reasonable use of SCTP streams, do that
> aspect does not strike me as unusual.

I agree :>

> 
> Now for the next part of this explanation, I'm going to have to infer
> certain things, so, Brian, please correct me if I infer wrongly.   *One*
> possibility for the sequence of events leading to the
> situation Brian describes is that there is a burst loss of data chunks
> TSN110 through TSN119 (ten chunks, but possibly less than ten packets..
> perhaps even only ONE packet if the chunks are small enough to be
> bundled!)... and then data chunks TSN120 through TSN128 make it through
> OK.   This does not strike me as particularly unusual; a loss of a single
> packet followed by a successful packet is sufficient to cause this situation.
> 

Or the other possiblility would be a set of bundled chunks in one packet
being lost.. either way is the same result I would think..

> The next thing that happens is that TSNs 110 through 118 are "dropped" in
> the technical sense of that word; see the defn in the current draft.  This
> means that the sender gives up on sending them, and issues a forward
> TSN.   However, no Forward TSN is issued for TSN119.  This may be because
> the messages on stream 9 are treated as Reliable.   This is NOT an unusual
> situation; it is perfectly reasonably to assume that an application might
> have "k" streams with data that can be dropped, and "n - k" streams of data
> that must be treated as reliable.  In the case of Brian's example, n is 10,
> and k is 9, but I think that his example will still produce similar results
> even for other values of "k" and "n".
> 

Note.. streams themselves are not reliable or partially reliable. With
the new draft it is the message itself which is sent with the
reliablilty
set to "fully reliable". One possilbe implementation would be to make
things correlate to a particular stream.. but this is just an
implementation
option..

> Bottom line:
> ========
> 
> I don't think its a corner case at all.   We have all the PR streams
> potentially being HOL blocked by loss of a single packet.  I think we need
> to fix this.

I agree..

> 
> The solution of forward SSNs is one viable solution I think.  There may be
> others, but that's as good as any for right now I think.   There may be
> another solution as well, but let me do some writing and thinking about it
> off line before I muddy the waters with it here... just to make sure it is
> a viable alternative.

agreed


> 
> Randy... I'm willing to make the text changes and send a proposed version
> around to the co-authors for comment if there is consensus to move in that
> direction.   Let me know.
> 

I will get it out this weekend if I may.. there are some interesting
things that I have thought about that must be covered... I had something
like this in an old u-sctp draft (during Q and I's debate earlier :>)



> 
> (Side note: I was worried that the "technical definition" of the word
> "dropped" would cause confusion, and it already has.   Perhaps should use
> another word in the draft to describe the operation where the sender
> decides to no longer transmit and/or retransmit a specific data chunk.  How
> about the term "abandoned" in place of "dropped"?)


Sounds good..

note, Michael R also agrees .. or seems to .. we will talk more..

R


> 
> /ptc
> 
> At 11:44 AM 6/5/2002 -0500, Randall Stewart wrote:
> >Brian/Qiaobing:
> >
> >I have read through this interesting debate. Qiaobing and
> >I had a major discussion on this way back when it was
> >considered U-SCTP. Renee Revis first pointed this out to me.. but
> >from a different persepective i.e. the receiver being able to
> >efficently process stuck TSN's ...
> >
> >After a lengthy debate I agreed to not have the SSN/SID listed
> >in the FWD TSN... I did NOT think of the HOL blocking issue.. or
> >I would have used it as a issue (I was the one arguing for the
> >SSN/SID in the FWD-TSN :->).
> >
> >After watching this debate.. I think I have come to the conclusion that
> >Brian is correct :)
> >
> >Michael R. is here... I will discuss it with him.. I want to get
> >the other co-author's take on this one... If the others agree with
> >Brian (and me :>) then I think we will put this in.... also I would
> >like to hear what the rest of the WG feels on this???
> >
> >
> >I think it is reasonable and not that much overhead to include JUST
> >the dropped streams NEXT expected SSN... You would be adding 4 bytes
> >for EVERY stream dropping one or more TSN's... note that this does
> >NOT mean 4 bytes per TSN but 4 bytes per where one or more TSN's were
> >dropped..
> >
> >
> >I think this is reasonable.. it improves processing AND avoids HOL
> >blocking..
> >
> >
> >Michael T/Phil? what do you think... Michael R.. I will talk to you at
> >the
> >next class break..
> >
> >
> >
> >R
> >
> >--
> >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@optimus.ietf.org  Fri Jun  7 12:14:01 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 MAA02562
	for <tsvwg-archive@odin.ietf.org>; Fri, 7 Jun 2002 12:14:01 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id MAA09219
	for tsvwg-archive@odin.ietf.org; Fri, 7 Jun 2002 12:14:32 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id KAA01234;
	Fri, 7 Jun 2002 10:32:07 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id DAA11684
	for <tsvwg@optimus.ietf.org>; Thu, 6 Jun 2002 03:50:28 -0400 (EDT)
Received: from web20605.mail.yahoo.com (web20605.mail.yahoo.com [216.136.226.163])
	by ietf.org (8.9.1a/8.9.1a) with SMTP id DAA01142
	for <tsvwg@ietf.org>; Thu, 6 Jun 2002 03:49:57 -0400 (EDT)
Message-ID: <20020606075027.559.qmail@web20605.mail.yahoo.com>
Received: from [61.1.171.56] by web20605.mail.yahoo.com via HTTP; Thu, 06 Jun 2002 00:50:27 PDT
Date: Thu, 6 Jun 2002 00:50:27 -0700 (PDT)
From: from dillip <dillip55@yahoo.com>
To: tsvwg@ietf.org
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Subject: [Tsvwg] please give me yours feedback.
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 Sir ! 
   I have some query on SCTP. Please help me regarding
this issue .

   I gone through RFC#2960 for implementing SCTP stack
and found some problem related to Gap-Ack-Ack . I am
writing details of the problem below here and please
reply me back . 

Accoridng to RFC:::
3.3.4 Selective Acknowledgement (SACK) (3):
   Gap Ack Block Start: 16 bits (unsigned integer)

      Indicates the Start offset TSN for this Gap Ack
Block.  To
      calculate the actual TSN number the Cumulative
TSN Ack is added to
      this offset number.  This calculated TSN
identifies the first TSN
      in this Gap Ack Block that has been received.

   Gap Ack Block End:  16 bits (unsigned integer)
   Indicates the End offset TSN for this Gap Ack
Block.  To calculate the actual TSN number the
Cumulative TSN Ack is added to this offset number. 
This calculated TSN identifies the TSN of the last
DATA chunk received in this Gap Ack Block.

Scenario-1>>>>>>>>
------------------
                        ----------
                        | TSN=17 |
                        ----------
                        |        | <- still missing
                        ----------
                        | TSN=15 |
                        ----------
                        | TSN=14 |
                        ----------
                        |        | <- still missing
                        ----------
                        | TSN=12 |
                        ----------
                        | TSN=11 |
                        ----------
                        | TSN=10 |
                        ----------

   then, the parameter part of the SACK MUST be
constructed as follows (assuming the new a_rwnd is set
to 4660 by the sender):

                  +--------------------------------+
                  |   Cumulative TSN Ack = 12      |
                  +--------------------------------+
                  |        a_rwnd = 4660           |
                  +----------------+---------------+
                  | num of block=2 | num of dup=0  |
                  +----------------+---------------+
                  |block #1 strt=2 |block #1 end=3 |
                  +----------------+---------------+
                  |block #2 strt=5 |block #2 end=5 |
                  +----------------+---------------+

	Here the actual TSN number should calculated by :::
		actual_tsn = cumulative_tsn + offset_number ;
	If this is correct then for the above scenario result
will be :::
                  +--------------------------------+
                  |   Cumulative TSN Ack = 12      |
                  +--------------------------------+
                  |        a_rwnd = 4660           |
                  +----------------+---------------+
                  | num of block=2 | num of dup=0  |
                  +----------------+---------------+
                  |block #1 strt=1 |block #1 end=1 |
                  +----------------+---------------+
                  |block #2 strt=4 |block #2 end=4 |
                  +----------------+---------------+
		
Scenario-2>>>>>>>>
------------------
	If the scenario looks::::::::::::::
                        ----------
                        | TSN=17 |
                        ----------
                        |        | <- still missing
                        ----------
                        | TSN=15 |
                        ----------
                        | 	   |
                        ----------
                        |        | <- still missing
                        ----------
                        | TSN=12 |
                        ----------
                        | TSN=11 |
                        ----------
	If this is correct then for the above scenario result
will be :::
                  +--------------------------------+
                  |   Cumulative TSN Ack = 12      |
                  +--------------------------------+
                  |        a_rwnd = 4660           |
                  +----------------+---------------+
                  | num of block=2 | num of dup=0  |
                  +----------------+---------------+
                  |block #1 strt=1 |block #1 end=2 |
                  +----------------+---------------+
                  |block #2 strt=4 |block #2 end=4 |
                  +----------------+---------------+
		

	Is it correct or I need to folow some other algorithm
to calculate the gaps ??

Regards , 
Dillip


=====
regards , 
DP 
Alliance Computer Telephony Inc .
Tel +91 6784 30418  (O)
    +91 6782 365252 (R)

__________________________________________________
Do You Yahoo!?
Yahoo! - Official partner of 2002 FIFA World Cup
http://fifaworldcup.yahoo.com


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



From daemon@optimus.ietf.org  Fri Jun  7 12:22:39 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 MAA03022
	for <tsvwg-archive@odin.ietf.org>; Fri, 7 Jun 2002 12:22:39 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id MAA09608
	for tsvwg-archive@odin.ietf.org; Fri, 7 Jun 2002 12:23:10 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id KAA01252;
	Fri, 7 Jun 2002 10:32:09 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id AAA22758
	for <tsvwg@optimus.ietf.org>; Thu, 6 Jun 2002 00:16:25 -0400 (EDT)
Received: from smtp.comcast.net (smtp.comcast.net [24.153.64.2])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id AAA14016
	for <tsvwg@ietf.org>; Thu, 6 Jun 2002 00:15:55 -0400 (EDT)
Received: from livingroom.acm.org
 (pcp01481250pcs.ncstl01.de.comcast.net [68.82.71.168])
 by mtaout06.icomcast.net
 (iPlanet Messaging Server 5.1 HotFix 0.8 (built May 13 2002))
 with ESMTP id <0GX900056NTX9Q@mtaout06.icomcast.net> for tsvwg@ietf.org; Thu,
 06 Jun 2002 00:15:35 -0400 (EDT)
Date: Thu, 06 Jun 2002 00:15:27 -0400
From: Phillip Conrad <conrad@acm.org>
Subject: Re: [Tsvwg] More on PRSCTP-00
In-reply-to: <3CFE8964.81F6314F@stewart.chicago.il.us>
X-Sender: phillipconrad@mail.comcast.net
To: Randall Stewart <randall@stewart.chicago.il.us>
Cc: bidulock@openss7.org, Qiaobing Xie <qxie1@email.mot.com>,
        Qiaobing Xie <Qiaobing.Xie@motorola.com>,
        "Armando L. Caro Jr." <acaro@mail.eecis.udel.edu>, tsvwg@ietf.org,
        Michael Tuexen <michael.tuexen@icn.siemens.de>,
        "Michael A. Ramalho" <mramalho@cisco.com>
Message-id: <5.1.0.14.2.20020605230911.025cca18@mail.earthlink.net>
MIME-version: 1.0
X-Mailer: QUALCOMM Windows Eudora Version 5.1
Content-type: text/plain; charset=us-ascii; format=flowed
Content-transfer-encoding: 7BIT
References: <Pine.GSO.4.33.0205311736060.8809-100000@ren.eecis.udel.edu>
 <3CF80F65.58E43813@email.mot.com> <20020531195333.D11468@openss7.org>
 <3CFBC11C.1EFA747@email.mot.com> <20020603144342.A30667@openss7.org>
 <3CFC144D.65697A41@email.mot.com> <20020603215952.E30667@openss7.org>
 <3CFD57D8.5637E431@email.mot.com> <20020604195020.A16020@openss7.org>
 <5.1.0.14.2.20020605155749.0236b050@joda.cis.temple.edu>
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

Replying to several notes at once here... first a reply to some of Randy's 
comments.. then a follow up on a conversation between Qiaobing and Brian.

At 04:57 PM 6/5/2002 -0500, Randall Stewart wrote:
>Phillip Conrad wrote:
> > (A) current design: simpler, fewer bits on the wire, but may (re)-introduce
> > an HOL blocking problem
>I must disagree with your "it is simpler" ... it is NOT. The question is
>where is the burden. Having implemented pr-sctp in the kernel of freebsd
>NOT having the stream number/SSN in the FWD-TSN means I must keep
>an extra set of linked lists and such to track any stream that has
>pending stuck data. When a fwd-tsn arrives I must HUNT all of these
>streams and look at each inbound queue.

...

Let me see if understand your point.. I think you are saying that from a 
holistic standpoint, the overall system may be simpler if we introduce 
forward SSNs, because as you, Armando and Brian have all pointed out, the 
additional fields can actually make implementation of the sender and 
receiver more straightforward, and more efficient... i.e., simpler.   I can 
accept this line of reasoning.

I should have been more precise in my language:  what I intended to convey 
is that the "on-the-wire protocol is simpler"... i.e., fewer fields to 
describe, parse and process.    But as your argument illustrates, when 
considering simplicity/complexity, we need to think of the system as a 
whole, not just one aspect of it.
...

> > Now for the next part of this explanation, I'm going to have to infer
> > certain things, so, Brian, please correct me if I infer wrongly.   *One*
> > possibility for the sequence of events leading to the
> > situation Brian describes is that there is a burst loss of data chunks
> > TSN110 through TSN119 (ten chunks, but possibly less than ten packets..
> > perhaps even only ONE packet if the chunks are small enough to be
> > bundled!)... and then data chunks TSN120 through TSN128 make it through
> > OK.   This does not strike me as particularly unusual; a loss of a single
> > packet followed by a successful packet is sufficient to cause this 
> situation.
> >
>
>Or the other possiblility would be a set of bundled chunks in one packet
>being lost.. either way is the same result I would think..

Hmmm... I think you and I are saying the same thing here.    The point is 
that 10 chunks might mean ten packets were dropped in a burst, or it might 
mean only one packet was dropped that contained 10 bundled chunks.  From 
the standpoint of chunk processing in SCTP, it does really matter... and 
neither of these is a particularly far fetched scenario.

> > The next thing that happens is that TSNs 110 through 118 are "dropped" in
> > the technical sense of that word; see the defn in the current draft.  This
> > means that the sender gives up on sending them, and issues a forward
> > TSN.   However, no Forward TSN is issued for TSN119.  This may be because
> > the messages on stream 9 are treated as Reliable.   This is NOT an unusual
> > situation; it is perfectly reasonably to assume that an application might
> > have "k" streams with data that can be dropped, and "n - k" streams of data
> > that must be treated as reliable.  In the case of Brian's example, n is 10,
> > and k is 9, but I think that his example will still produce similar results
> > even for other values of "k" and "n".
> >
>
>Note.. streams themselves are not reliable or partially reliable. With
>the new draft it is the message itself which is sent with the
>reliablilty
>set to "fully reliable". One possilbe implementation would be to make
>things correlate to a particular stream.. but this is just an
>implementation
>option..

Understood... As you point out, the current spec (unlike the
previous U-SCTP spec) does not designate particular streams as reliable or
unreliable; instead indvidual messages are designated this way on the basis
of their lifetime parameters.

What I'm talking about here is just an application scenario
for purposes of illustration.   As part of the argument that  Brian's scenario
is not a corner case, we need to argue that the scenario he describes can
arise is some reasonably normal usage of SCTP.

So for purposes of  illustration only, it is reasonable
to assume  that a given application might "so happen" to use
"k" out of "n" streams for messages with  lifetime parameters
that can  expire, and "n-k" streams for messages that are
fully reliable, not mixing the two.

This makes describing the scenario easier, and
also seems like something that an application might likely do.    For those
that like things to be more "concrete" in a "PS" to this note, I describe a
specific application scenario to motivate this situation.

Now turning to another discussion:

At 05:00 PM 6/5/2002 -0500, Qiaobing Xie wrote:
>Phil,
>
>Please take my little challenges below as a way to help you making your
>case tighter :-)

Qiaobing... thanks for this... I take your comments in that spirit.  And I 
think you've found some weaknesses in my argument that can be tightened up.

>On Wed, 05 Jun 2002, Qiaobing Xie wrote:
> > Again, what's really funny is that we suddenly start to see this HOL
> > blocking caused by the loss of a single SCTP packet (presumably
> > containing multiple TSNs from different streams) as an issue. We seem to
> > be able to live with it in the base SCTP for a long time.

At 05:15 PM 6/5/2002 -0500, Brian F. G. Bidulock responded:
>But these TSNs were also dropped.  It is the further loss of a single
>TSN which causes the blocking.  It is not the case that the loss of a
>single TSN from one stream blocks other streams, reagardless of the history
>leading up to the loss of the TSN.

I think Brian has it right here.. but I just want to clarify to make sure I 
understand.   Qiaobing is correct in asserting that in the base SCTP 
protocol, it is true that if packets from multiple streams are bundled, the 
loss of a single packet can impact multiple streams.  This is true in the 
base SCTP.

However, there is a property of the base SCTP protocol that can be 
expressed as follows: the loss of a chunk on stream i does NOT delay 
delivery of a chunk on stream j, where i!=j.   Only the loss of lower 
numbered chunks on stream j can affect the delivery of a chunk on stream 
j.   Thus, the base SCTP protocol avoids a particular kind of HOL 
blocking.   (Now this principle is not strictly true even in the base SCTP, 
if one considers the effects of rwnd and cwnd... but at least from the 
standpoint of "ordered delivery", this principle holds.)

The objection that Brian raises, and others (included myself) have 
concurred with, is that the FORWARD TSN chunk, as specified in the current 
draft, creates the possibility that this principle may be negated.    As 
Brian's example illustrates, the loss of a TSN unrelated to stream j may 
delay delivery of packets on stream j, because when a FORWARD TSN moves the 
cum ack point forward, there is an ambiguity about which streams the 
"skipped" packets belong to.   The FORWARD SSN corrects this ambiguity.

I am glad that Qiaobing is being rigorous and asking us to consider very 
carefully whether a change in the design is needed, and inspiring us to 
make our arguments very precise.  Qiaobing, don't let us bully you... as 
the film "Twelve Angry Men" illustrates.. just because someone is "on their 
own", doesn't mean they are wrong.   In that film, a jury that starts out 
11-to-1 in favor of a guilty verdict, eventually reaches a unanimous 
verdict of "not guilty."  We may disagree on "taste" or "philosophy", but 
if we are in disagreement about things that we should be able to logically 
reason out, then we probably need to continue the dialogue until we have 
true consensus.

Thanks,
Phill



*PS: For those who need a specific application scenario, consider a 
real-time application in manufacturing process control that is sending one 
stream of data that is critical to archive (perhaps emissions reports that 
must be kept for the EPA and thus must be sent reliably) along with some 
other streams of data (temperature, pressure, etc)  where timeliness is 
more crucial than having a complete record of the data...   suppose that 
the received data are being graphed in real time, and if a few intermediate 
points are skipped, the person reading the chart can still make sense of 
the data stream.    One could imagine a similar setup for remote control of 
an robotic vehicle doing toxic waste cleanup, mine sweeping, etc.



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



From daemon@optimus.ietf.org  Fri Jun  7 14:18:24 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 OAA07266
	for <tsvwg-archive@odin.ietf.org>; Fri, 7 Jun 2002 14:18:15 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id OAA20535
	for tsvwg-archive@odin.ietf.org; Fri, 7 Jun 2002 14:18:46 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id NAA18735;
	Fri, 7 Jun 2002 13:49:20 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id NAA18707
	for <tsvwg@optimus.ietf.org>; Fri, 7 Jun 2002 13:49:18 -0400 (EDT)
Received: from ilsa.franken.de (postfix@ilsa.franken.de [193.175.24.42])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id NAA05517
	for <tsvwg@ietf.org>; Fri, 7 Jun 2002 13:48:46 -0400 (EDT)
Received: from localhost (pD9E4F1B9.dip.t-dialin.net [217.228.241.185])
	by ilsa.franken.de (Postfix) with ESMTP
	id E18A9245CF; Fri,  7 Jun 2002 19:49:07 +0200 (CEST)
Date: Fri, 7 Jun 2002 19:48:52 +0200
Subject: Re: [Tsvwg] More on PRSCTP-00
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Mime-Version: 1.0 (Apple Message framework v482)
Cc: bidulock@openss7.org, Phillip Conrad <conrad@acm.org>,
        Qiaobing Xie <qxie1@email.mot.com>,
        Qiaobing Xie <Qiaobing.Xie@motorola.com>,
        "Armando L. Caro Jr." <acaro@mail.eecis.udel.edu>, tsvwg@ietf.org,
        "Michael A. Ramalho" <mramalho@cisco.com>
To: Randall Stewart <randall@stewart.chicago.il.us>
From: Michael Tuexen <Michael.Tuexen@lurchi.franken.de>
In-Reply-To: <3CFEA220.DE96B441@stewart.chicago.il.us>
Message-Id: <D4966AFE-7A3E-11D6-9E0D-0030654C1AB6@lurchi.franken.de>
X-Mailer: Apple Mail (2.482)
Content-Transfer-Encoding: 8bit
X-MIME-Autoconverted: from quoted-printable to 8bit by optimus.ietf.org id NAA18708
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

Dear all,

reading the thread I would like to make some remarks:

1. I agree that using SSN in the FWD chunk avoids HOL (which is good).
2. Brians example show how HOL can occur if no SSN info in the FWD chunk
    is used. He only looks at the receiver.

Brians example are always based on the assumption that there is a GAP
in the TSN space at the receiver side and the receiver gets a FWD TSN
which does not fill the whole gap. Then the HOL happens at the receiver.

I'm just wondering how often this will happen. What are the reasons that
a FWD chunk will be send: RTX Timer expiration or fast retransmission.
In both cases I would send a FWD chunk and bundle the remaining DATA 
chunks
(if possible). So there is a good chance that the whole gap is filled by
the FWD chunk and the retransmitted DATA chunks.
I'm not sure how likely Brians scenario is taking the sender into 
account.

So I think one has to take the senders behavior into account when 
discussing
this issue.

Best regards
Michael

On Thursday, June 6, 2002, at 01:43 AM, Randall Stewart wrote:

> "Brian F. G. Bidulock" wrote:
>>
>> Randall,
>>
>> On Wed, 05 Jun 2002, Randall Stewart wrote:
>>
>>> A few comments..
>>>
>>> Phillip Conrad wrote:
>>>>
>>>> Dear Colleagues:
>>>>       I agree with Randy that Brian has raised an important issue 
>>>> for us to
>>>> consider, and I think has explained it quite well with a very clear,
>>>> illustrative example.
>>>>
>>>> The question is one of a tradeoff between two designs:
>>>>
>>>> (A) current design: simpler, fewer bits on the wire, but may 
>>>> (re)-introduce
>>>> an HOL blocking problem
>>>
>>> I must disagree with your "it is simpler" ... it is NOT. The question 
>>> is
>>> where is the burden. Having implemented pr-sctp in the kernel of 
>>> freebsd
>>> NOT having the stream number/SSN in the FWD-TSN means I must keep
>>> an extra set of linked lists and such to track any stream that has
>>> pending stuck data. When a fwd-tsn arrives I must HUNT all of these
>>> streams and look at each inbound queue.
>>>
>>> The counter to this is the sender (in Brian and my earlier proposal
>>> that was to Q only) must track this info to send with the FWD-TSN.
>>> Actually
>>> less of a burden IMHO...
>>
>> I agree.  I discovered this situation while implementing PR-SCTP in the
>> Linux Kernel.  The normal receiver can rely on the fact that the next
>> expected SSN is the last fully received SSN plus 1.  PR-SCTP without
>> a forward SSN causes the receive to have to update this expected value
>> with the next SSN in the reorder queue whenever there are no TSN gaps.
>> This even might require that when TSN gaps are filled that the receiver
>> walk all inbound stream structures to update the next expected SSN.
>> In an independent implementation, I have arrived at the same 
>> conclusion:
>> that the reciever has more work to do for FORWARD TSN without a forward
>> SSN.
>>
>
> Yeah.. I basically had to make a special linked list set that the
> stream queues could be placed on.. a bit more efficient but a
> reall hassel to maintian a "wheel" (as I call it) of streams with
> pending data on it...
>
>> In fact the receiver has this work to do if it has ever received a
>> FORWARD TSN.  This is because it is not possible for the receiver to
>> know a priori which streams could be affected by this scenairo.  
>> Therefore,
>> the load on the receiver is increased even when this case does not 
>> occur.
>> It is the mere possibility of its occurence which introduces the load.
>> (Each time gaps close, the next expected SSN for all streams need to be
>> re-examined whether there are SSN gaps or not.)
>
> Exactly.. I constantly must place stream q;s on and off the :"wheel" no
> matter if a fwd-tsn is EVER received...
>
>>
>> On the sender side, the sender must always track the last sent SSN and
>> assign the next SSN (++SSN).  Indeed the sender when generating a 
>> FORWARD
>> TSN chunk would need to traverse outbound stream structures with 
>> dropped
>> TSNs to formulate Forward TSNs, but this is likely to be a far shorter
>> list than at the receiver and non-PR streams would never need to be
>> involved.
>
> And it must continue to KEEP this information until it is sure the
> peer has received and moved the cum-ack forward..
>
> What I end up doing on the sender side is I hold just the bookkeeping
> structures of the TSN (stream seq/stream no/tsn ... etc) but I
> drop the mbuf holding the data so I gain a bit of help in mbuf
> utilization.
>
> But I can't get rid of the chunk structure UNTIL the cum-ack says he
> got my FWD-TSN and accepted it...
>
> So there is NO additional info to save on the sender side and only
> do I save what I must... on the receiver side I have a LOT of extra
> bookkeeping....
>
>
> In fact with your confirmation I would claim just the opposite of
> Phil's statement... that doing it the way it is NOW documented takes
> MORE overhead.. only by adding the SSN/SID can you reduce this..
>
> By the way Renee also pointed out the extra overhead for the IOS
> implementation as well :>... so this makes three different
> implementations
> NOT 2....
>
> I plan on adding this to the next pr-sctp unless Michael Tuexen
> violently
> agrees with Qiaobing ...
>
> I still await Michaels answer????
>
>
>>
>> --brian
>>
>>>>
>>>> (B) alternative design: more complex, may require larger Forward TSN
>>>> chunks, but avoids re-introducing an HOL blocking problem, and thus
>>>> improves performance in some situations.
>>>
>>>
>>> see above it is NOT more complex...
>>>
>>>>
>>>> The question of whether to choose A or B depends on whether you 
>>>> believe
>>>> that the situation Brian describes is a "corner case" or not.   So 
>>>> I've
>>>> walked through Brian's example, and have determined that I don't 
>>>> think it
>>>> is a corner case at all.   So, I'm inclined to agree with Randy and 
>>>> Brian
>>>> that we should consider a change.
>>>>
>>>> But I'd like to invite others to walk through my line of reasoning 
>>>> to test
>>>> whether I've reasoned this correctly.
>>>
>>>
>>> I agree it is NOT a corner case..
>>>
>>>>
>>>> My rationale that Brian's example is NOT a corner case:
>>>> ========================================
>>>>
>>>> In Brian's example, messages are being sent round-robin on a number 
>>>> of
>>>> streams.  This seems like a fairly reasonable use of SCTP streams, 
>>>> do that
>>>> aspect does not strike me as unusual.
>>>
>>> I agree :>
>>>
>>>>
>>>> Now for the next part of this explanation, I'm going to have to infer
>>>> certain things, so, Brian, please correct me if I infer wrongly.   
>>>> *One*
>>>> possibility for the sequence of events leading to the
>>>> situation Brian describes is that there is a burst loss of data 
>>>> chunks
>>>> TSN110 through TSN119 (ten chunks, but possibly less than ten 
>>>> packets..
>>>> perhaps even only ONE packet if the chunks are small enough to be
>>>> bundled!)... and then data chunks TSN120 through TSN128 make it 
>>>> through
>>>> OK.   This does not strike me as particularly unusual; a loss of a 
>>>> single
>>>> packet followed by a successful packet is sufficient to cause this 
>>>> situation.
>>>>
>>>
>>> Or the other possiblility would be a set of bundled chunks in one 
>>> packet
>>> being lost.. either way is the same result I would think..
>>>
>>>> The next thing that happens is that TSNs 110 through 118 are 
>>>> "dropped" in
>>>> the technical sense of that word; see the defn in the current 
>>>> draft.  This
>>>> means that the sender gives up on sending them, and issues a forward
>>>> TSN.   However, no Forward TSN is issued for TSN119.  This may be 
>>>> because
>>>> the messages on stream 9 are treated as Reliable.   This is NOT an 
>>>> unusual
>>>> situation; it is perfectly reasonably to assume that an application 
>>>> might
>>>> have "k" streams with data that can be dropped, and "n - k" streams 
>>>> of data
>>>> that must be treated as reliable.  In the case of Brian's example, n 
>>>> is 10,
>>>> and k is 9, but I think that his example will still produce similar 
>>>> results
>>>> even for other values of "k" and "n".
>>>>
>>>
>>> Note.. streams themselves are not reliable or partially reliable. With
>>> the new draft it is the message itself which is sent with the
>>> reliablilty
>>> set to "fully reliable". One possilbe implementation would be to make
>>> things correlate to a particular stream.. but this is just an
>>> implementation
>>> option..
>>>
>>>> Bottom line:
>>>> ========
>>>>
>>>> I don't think its a corner case at all.   We have all the PR streams
>>>> potentially being HOL blocked by loss of a single packet.  I think 
>>>> we need
>>>> to fix this.
>>>
>>> I agree..
>>>
>>>>
>>>> The solution of forward SSNs is one viable solution I think.  There 
>>>> may be
>>>> others, but that's as good as any for right now I think.   There may 
>>>> be
>>>> another solution as well, but let me do some writing and thinking 
>>>> about it
>>>> off line before I muddy the waters with it here... just to make sure 
>>>> it is
>>>> a viable alternative.
>>>
>>> agreed
>>>
>>>
>>>>
>>>> Randy... I'm willing to make the text changes and send a proposed 
>>>> version
>>>> around to the co-authors for comment if there is consensus to move 
>>>> in that
>>>> direction.   Let me know.
>>>>
>>>
>>> I will get it out this weekend if I may.. there are some interesting
>>> things that I have thought about that must be covered... I had 
>>> something
>>> like this in an old u-sctp draft (during Q and I's debate earlier :>)
>>>
>>>
>>>
>>>>
>>>> (Side note: I was worried that the "technical definition" of the word
>>>> "dropped" would cause confusion, and it already has.   Perhaps 
>>>> should use
>>>> another word in the draft to describe the operation where the sender
>>>> decides to no longer transmit and/or retransmit a specific data 
>>>> chunk.  How
>>>> about the term "abandoned" in place of "dropped"?)
>>>
>>>
>>> Sounds good..
>>>
>>> note, Michael R also agrees .. or seems to .. we will talk more..
>>>
>>> R
>>>
>>>
>>>>
>>>> /ptc
>>>>
>>>> At 11:44 AM 6/5/2002 -0500, Randall Stewart wrote:
>>>>> Brian/Qiaobing:
>>>>>
>>>>> I have read through this interesting debate. Qiaobing and
>>>>> I had a major discussion on this way back when it was
>>>>> considered U-SCTP. Renee Revis first pointed this out to me.. but
>>>>> from a different persepective i.e. the receiver being able to
>>>>> efficently process stuck TSN's ...
>>>>>
>>>>> After a lengthy debate I agreed to not have the SSN/SID listed
>>>>> in the FWD TSN... I did NOT think of the HOL blocking issue.. or
>>>>> I would have used it as a issue (I was the one arguing for the
>>>>> SSN/SID in the FWD-TSN :->).
>>>>>
>>>>> After watching this debate.. I think I have come to the conclusion 
>>>>> that
>>>>> Brian is correct :)
>>>>>
>>>>> Michael R. is here... I will discuss it with him.. I want to get
>>>>> the other co-author's take on this one... If the others agree with
>>>>> Brian (and me :>) then I think we will put this in.... also I would
>>>>> like to hear what the rest of the WG feels on this???
>>>>>
>>>>>
>>>>> I think it is reasonable and not that much overhead to include JUST
>>>>> the dropped streams NEXT expected SSN... You would be adding 4 bytes
>>>>> for EVERY stream dropping one or more TSN's... note that this does
>>>>> NOT mean 4 bytes per TSN but 4 bytes per where one or more TSN's 
>>>>> were
>>>>> dropped..
>>>>>
>>>>>
>>>>> I think this is reasonable.. it improves processing AND avoids HOL
>>>>> blocking..
>>>>>
>>>>>
>>>>> Michael T/Phil? what do you think... Michael R.. I will talk to you 
>>>>> at
>>>>> the
>>>>> next class break..
>>>>>
>>>>>
>>>>>
>>>>> R
>>>>>
>>>>> --
>>>>> 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)
>>
>> --
>> Brian F. G. Bidulock    ¦ The reasonable man adapts himself to the ¦
>> bidulock@openss7.org    ¦ world; the unreasonable one persists in  ¦
>> http://www.openss7.org/ ¦ trying  to adapt the  world  to himself. ¦
>>                         ¦ Therefore  all  progress  depends on the ¦
>>                         ¦ unreasonable man. -- George Bernard Shaw ¦
>
> --
> 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
>
>
Michael.Tuexen@micmac.franken.de


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



From daemon@optimus.ietf.org  Fri Jun  7 17:01:19 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 RAA12566
	for <tsvwg-archive@odin.ietf.org>; Fri, 7 Jun 2002 17:01:11 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id RAA06250
	for tsvwg-archive@odin.ietf.org; Fri, 7 Jun 2002 17:01:43 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id QAA03989;
	Fri, 7 Jun 2002 16:40:16 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id QAA03895
	for <tsvwg@optimus.ietf.org>; Fri, 7 Jun 2002 16:40:12 -0400 (EDT)
Received: from bfgbhome.inetint.com (IDENT:root@tnt-dal-42-211.dallas.net [209.44.42.211])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id QAA11902
	for <tsvwg@ietf.org>; Fri, 7 Jun 2002 16:39:35 -0400 (EDT)
Received: (from brian@localhost)
	by bfgbhome.inetint.com (8.9.3/8.9.3) id PAA06478;
	Fri, 7 Jun 2002 15:39:11 -0500
Date: Fri, 7 Jun 2002 15:39:11 -0500
From: "Brian F. G. Bidulock" <bidulock@openss7.org>
To: Qiaobing Xie <Qiaobing.Xie@motorola.com>
Cc: Michael Tuexen <Michael.Tuexen@lurchi.franken.de>,
        Randall Stewart <randall@stewart.chicago.il.us>,
        Phillip Conrad <conrad@acm.org>,
        "Armando L. Caro Jr." <acaro@mail.eecis.udel.edu>, tsvwg@ietf.org,
        "Michael A. Ramalho" <mramalho@cisco.com>
Subject: Re: [Tsvwg] More on PRSCTP-00
Message-ID: <20020607153911.D2907@openss7.org>
Reply-To: bidulock@openss7.org
References: <D4966AFE-7A3E-11D6-9E0D-0030654C1AB6@lurchi.franken.de> <3D010E5E.D51114EF@motorola.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
User-Agent: Mutt/1.2i
In-Reply-To: <3D010E5E.D51114EF@motorola.com>; from Qiaobing.Xie@motorola.com on Fri, Jun 07, 2002 at 02:49:50PM -0500
Organization: http://www.openss7.org/
Dsn-Notification-To: <bidulock@openss7.org>
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

Michael, Qiaobing,

Here is the sender side of the example.  It is quite reasonable
that this could occur.  The receiver is blocked on streams 1
through 9 until retransmission of TSN 119 is successful.  This
is the sender side of the same example.

    TSN 101  SID 1   SSN 1   |
    TSN 102  SID 2   SSN 1   |
    TSN 103  SID 3   SSN 1   |
    TSN 104  SID 4   SSN 1   |   SCTP Packet
    TSN 105  SID 5   SSN 1   +--------------------->
    TSN 106  SID 6   SSN 1   |
    TSN 107  SID 7   SSN 1   |
    TSN 108  SID 8   SSN 1   |
    TSN 109  SID 9   SSN 1   |

    TSN 110  SID 1   SSN 2   |
    TSN 111  SID 2   SSN 2   |
    TSN 112  SID 3   SSN 2   |
    TSN 113  SID 4   SSN 2   |   SCTP Packet
    TSN 114  SID 5   SSN 2   +----------X
    TSN 115  SID 6   SSN 2   |         Lost
    TSN 116  SID 7   SSN 2   |
    TSN 117  SID 8   SSN 2   |
    TSN 118  SID 9   SSN 2   |
                                 SCTP Packet
    TSN 119  SID 0   SSN X   |----------X
                                       Lost

    (Timed reliability expiry on TSNs 110 to 118)

    FORWARD TSN 118          |--------------------->

    TSN 120  SID 1   SSN 3   |
    TSN 121  SID 2   SSN 3   |
    TSN 122  SID 3   SSN 3   |
    TSN 123  SID 4   SSN 3   |   SCTP Packet
    TSN 124  SID 5   SSN 3   +--------------------->
    TSN 125  SID 6   SSN 3   |
    TSN 126  SID 7   SSN 3   |
    TSN 127  SID 8   SSN 3   |
    TSN 128  SID 9   SSN 3   |

    (Receiver is now blocked on streams 1 through 9.)

    (Some time later, retrans timeout on TSN 119)

                                 SCTP Packet
    TSN 119  SID 0   SSN X   |--------------------->


Q.E.D.

--brian

On Fri, 07 Jun 2002, Qiaobing Xie wrote:

> Michael,
> 
> Right on! You summarized my previous point beautifully. I start to
> realize that lately my arguments tend to become lengthy and off the
> point... maybe a sign of getting old :<
> 
> So far I kind of buy Randy's point better that "adding some bits in
> exchange for ease of implementation", but still yet to be convinced
> about the HOL blocking argument made by Brian. 
> 
> I've asked for an "all-considered" example of the HOL blocking, but
> haven't seen one yet.
> 
> regards,
> -Qiaobing
> 
> Michael Tuexen wrote:
> > 
> > Dear all,
> > 
> > reading the thread I would like to make some remarks:
> > 
> > 1. I agree that using SSN in the FWD chunk avoids HOL (which is good).
> > 2. Brians example show how HOL can occur if no SSN info in the FWD chunk
> >     is used. He only looks at the receiver.
> > 
> > Brians example are always based on the assumption that there is a GAP
> > in the TSN space at the receiver side and the receiver gets a FWD TSN
> > which does not fill the whole gap. Then the HOL happens at the receiver.
> > 
> > I'm just wondering how often this will happen. What are the reasons that
> > a FWD chunk will be send: RTX Timer expiration or fast retransmission.
> > In both cases I would send a FWD chunk and bundle the remaining DATA
> > chunks
> > (if possible). So there is a good chance that the whole gap is filled by
> > the FWD chunk and the retransmitted DATA chunks.
> > I'm not sure how likely Brians scenario is taking the sender into
> > account.
> > 
> > So I think one has to take the senders behavior into account when
> > discussing
> > this issue.
> > 
> > Best regards
> > Michael
> > 
> > On Thursday, June 6, 2002, at 01:43 AM, Randall Stewart wrote:
> > 
> > > "Brian F. G. Bidulock" wrote:
> > >>
> > >> Randall,
> > >>
> > >> On Wed, 05 Jun 2002, Randall Stewart wrote:
> > >>
> > >>> A few comments..
> > >>>
> > >>> Phillip Conrad wrote:
> > >>>>
> > >>>> Dear Colleagues:
> > >>>>       I agree with Randy that Brian has raised an important issue
> > >>>> for us to
> > >>>> consider, and I think has explained it quite well with a very clear,
> > >>>> illustrative example.
> > >>>>
> > >>>> The question is one of a tradeoff between two designs:
> > >>>>
> > >>>> (A) current design: simpler, fewer bits on the wire, but may
> > >>>> (re)-introduce
> > >>>> an HOL blocking problem
> > >>>
> > >>> I must disagree with your "it is simpler" ... it is NOT. The question
> > >>> is
> > >>> where is the burden. Having implemented pr-sctp in the kernel of
> > >>> freebsd
> > >>> NOT having the stream number/SSN in the FWD-TSN means I must keep
> > >>> an extra set of linked lists and such to track any stream that has
> > >>> pending stuck data. When a fwd-tsn arrives I must HUNT all of these
> > >>> streams and look at each inbound queue.
> > >>>
> > >>> The counter to this is the sender (in Brian and my earlier proposal
> > >>> that was to Q only) must track this info to send with the FWD-TSN.
> > >>> Actually
> > >>> less of a burden IMHO...
> > >>
> > >> I agree.  I discovered this situation while implementing PR-SCTP in the
> > >> Linux Kernel.  The normal receiver can rely on the fact that the next
> > >> expected SSN is the last fully received SSN plus 1.  PR-SCTP without
> > >> a forward SSN causes the receive to have to update this expected value
> > >> with the next SSN in the reorder queue whenever there are no TSN gaps.
> > >> This even might require that when TSN gaps are filled that the receiver
> > >> walk all inbound stream structures to update the next expected SSN.
> > >> In an independent implementation, I have arrived at the same
> > >> conclusion:
> > >> that the reciever has more work to do for FORWARD TSN without a forward
> > >> SSN.
> > >>
> > >
> > > Yeah.. I basically had to make a special linked list set that the
> > > stream queues could be placed on.. a bit more efficient but a
> > > reall hassel to maintian a "wheel" (as I call it) of streams with
> > > pending data on it...
> > >
> > >> In fact the receiver has this work to do if it has ever received a
> > >> FORWARD TSN.  This is because it is not possible for the receiver to
> > >> know a priori which streams could be affected by this scenairo.
> > >> Therefore,
> > >> the load on the receiver is increased even when this case does not
> > >> occur.
> > >> It is the mere possibility of its occurence which introduces the load.
> > >> (Each time gaps close, the next expected SSN for all streams need to be
> > >> re-examined whether there are SSN gaps or not.)
> > >
> > > Exactly.. I constantly must place stream q;s on and off the :"wheel" no
> > > matter if a fwd-tsn is EVER received...
> > >
> > >>
> > >> On the sender side, the sender must always track the last sent SSN and
> > >> assign the next SSN (++SSN).  Indeed the sender when generating a
> > >> FORWARD
> > >> TSN chunk would need to traverse outbound stream structures with
> > >> dropped
> > >> TSNs to formulate Forward TSNs, but this is likely to be a far shorter
> > >> list than at the receiver and non-PR streams would never need to be
> > >> involved.
> > >
> > > And it must continue to KEEP this information until it is sure the
> > > peer has received and moved the cum-ack forward..
> > >
> > > What I end up doing on the sender side is I hold just the bookkeeping
> > > structures of the TSN (stream seq/stream no/tsn ... etc) but I
> > > drop the mbuf holding the data so I gain a bit of help in mbuf
> > > utilization.
> > >
> > > But I can't get rid of the chunk structure UNTIL the cum-ack says he
> > > got my FWD-TSN and accepted it...
> > >
> > > So there is NO additional info to save on the sender side and only
> > > do I save what I must... on the receiver side I have a LOT of extra
> > > bookkeeping....
> > >
> > >
> > > In fact with your confirmation I would claim just the opposite of
> > > Phil's statement... that doing it the way it is NOW documented takes
> > > MORE overhead.. only by adding the SSN/SID can you reduce this..
> > >
> > > By the way Renee also pointed out the extra overhead for the IOS
> > > implementation as well :>... so this makes three different
> > > implementations
> > > NOT 2....
> > >
> > > I plan on adding this to the next pr-sctp unless Michael Tuexen
> > > violently
> > > agrees with Qiaobing ...
> > >
> > > I still await Michaels answer????
> > >
> > >
> > >>
> > >> --brian
> > >>
> > >>>>
> > >>>> (B) alternative design: more complex, may require larger Forward TSN
> > >>>> chunks, but avoids re-introducing an HOL blocking problem, and thus
> > >>>> improves performance in some situations.
> > >>>
> > >>>
> > >>> see above it is NOT more complex...
> > >>>
> > >>>>
> > >>>> The question of whether to choose A or B depends on whether you
> > >>>> believe
> > >>>> that the situation Brian describes is a "corner case" or not.   So
> > >>>> I've
> > >>>> walked through Brian's example, and have determined that I don't
> > >>>> think it
> > >>>> is a corner case at all.   So, I'm inclined to agree with Randy and
> > >>>> Brian
> > >>>> that we should consider a change.
> > >>>>
> > >>>> But I'd like to invite others to walk through my line of reasoning
> > >>>> to test
> > >>>> whether I've reasoned this correctly.
> > >>>
> > >>>
> > >>> I agree it is NOT a corner case..
> > >>>
> > >>>>
> > >>>> My rationale that Brian's example is NOT a corner case:
> > >>>> ========================================
> > >>>>
> > >>>> In Brian's example, messages are being sent round-robin on a number
> > >>>> of
> > >>>> streams.  This seems like a fairly reasonable use of SCTP streams,
> > >>>> do that
> > >>>> aspect does not strike me as unusual.
> > >>>
> > >>> I agree :>
> > >>>
> > >>>>
> > >>>> Now for the next part of this explanation, I'm going to have to infer
> > >>>> certain things, so, Brian, please correct me if I infer wrongly.
> > >>>> *One*
> > >>>> possibility for the sequence of events leading to the
> > >>>> situation Brian describes is that there is a burst loss of data
> > >>>> chunks
> > >>>> TSN110 through TSN119 (ten chunks, but possibly less than ten
> > >>>> packets..
> > >>>> perhaps even only ONE packet if the chunks are small enough to be
> > >>>> bundled!)... and then data chunks TSN120 through TSN128 make it
> > >>>> through
> > >>>> OK.   This does not strike me as particularly unusual; a loss of a
> > >>>> single
> > >>>> packet followed by a successful packet is sufficient to cause this
> > >>>> situation.
> > >>>>
> > >>>
> > >>> Or the other possiblility would be a set of bundled chunks in one
> > >>> packet
> > >>> being lost.. either way is the same result I would think..
> > >>>
> > >>>> The next thing that happens is that TSNs 110 through 118 are
> > >>>> "dropped" in
> > >>>> the technical sense of that word; see the defn in the current
> > >>>> draft.  This
> > >>>> means that the sender gives up on sending them, and issues a forward
> > >>>> TSN.   However, no Forward TSN is issued for TSN119.  This may be
> > >>>> because
> > >>>> the messages on stream 9 are treated as Reliable.   This is NOT an
> > >>>> unusual
> > >>>> situation; it is perfectly reasonably to assume that an application
> > >>>> might
> > >>>> have "k" streams with data that can be dropped, and "n - k" streams
> > >>>> of data
> > >>>> that must be treated as reliable.  In the case of Brian's example, n
> > >>>> is 10,
> > >>>> and k is 9, but I think that his example will still produce similar
> > >>>> results
> > >>>> even for other values of "k" and "n".
> > >>>>
> > >>>
> > >>> Note.. streams themselves are not reliable or partially reliable. With
> > >>> the new draft it is the message itself which is sent with the
> > >>> reliablilty
> > >>> set to "fully reliable". One possilbe implementation would be to make
> > >>> things correlate to a particular stream.. but this is just an
> > >>> implementation
> > >>> option..
> > >>>
> > >>>> Bottom line:
> > >>>> ========
> > >>>>
> > >>>> I don't think its a corner case at all.   We have all the PR streams
> > >>>> potentially being HOL blocked by loss of a single packet.  I think
> > >>>> we need
> > >>>> to fix this.
> > >>>
> > >>> I agree..
> > >>>
> > >>>>
> > >>>> The solution of forward SSNs is one viable solution I think.  There
> > >>>> may be
> > >>>> others, but that's as good as any for right now I think.   There may
> > >>>> be
> > >>>> another solution as well, but let me do some writing and thinking
> > >>>> about it
> > >>>> off line before I muddy the waters with it here... just to make sure
> > >>>> it is
> > >>>> a viable alternative.
> > >>>
> > >>> agreed
> > >>>
> > >>>
> > >>>>
> > >>>> Randy... I'm willing to make the text changes and send a proposed
> > >>>> version
> > >>>> around to the co-authors for comment if there is consensus to move
> > >>>> in that
> > >>>> direction.   Let me know.
> > >>>>
> > >>>
> > >>> I will get it out this weekend if I may.. there are some interesting
> > >>> things that I have thought about that must be covered... I had
> > >>> something
> > >>> like this in an old u-sctp draft (during Q and I's debate earlier :>)
> > >>>
> > >>>
> > >>>
> > >>>>
> > >>>> (Side note: I was worried that the "technical definition" of the word
> > >>>> "dropped" would cause confusion, and it already has.   Perhaps
> > >>>> should use
> > >>>> another word in the draft to describe the operation where the sender
> > >>>> decides to no longer transmit and/or retransmit a specific data
> > >>>> chunk.  How
> > >>>> about the term "abandoned" in place of "dropped"?)
> > >>>
> > >>>
> > >>> Sounds good..
> > >>>
> > >>> note, Michael R also agrees .. or seems to .. we will talk more..
> > >>>
> > >>> R
> > >>>
> > >>>
> > >>>>
> > >>>> /ptc
> > >>>>
> > >>>> At 11:44 AM 6/5/2002 -0500, Randall Stewart wrote:
> > >>>>> Brian/Qiaobing:
> > >>>>>
> > >>>>> I have read through this interesting debate. Qiaobing and
> > >>>>> I had a major discussion on this way back when it was
> > >>>>> considered U-SCTP. Renee Revis first pointed this out to me.. but
> > >>>>> from a different persepective i.e. the receiver being able to
> > >>>>> efficently process stuck TSN's ...
> > >>>>>
> > >>>>> After a lengthy debate I agreed to not have the SSN/SID listed
> > >>>>> in the FWD TSN... I did NOT think of the HOL blocking issue.. or
> > >>>>> I would have used it as a issue (I was the one arguing for the
> > >>>>> SSN/SID in the FWD-TSN :->).
> > >>>>>
> > >>>>> After watching this debate.. I think I have come to the conclusion
> > >>>>> that
> > >>>>> Brian is correct :)
> > >>>>>
> > >>>>> Michael R. is here... I will discuss it with him.. I want to get
> > >>>>> the other co-author's take on this one... If the others agree with
> > >>>>> Brian (and me :>) then I think we will put this in.... also I would
> > >>>>> like to hear what the rest of the WG feels on this???
> > >>>>>
> > >>>>>
> > >>>>> I think it is reasonable and not that much overhead to include JUST
> > >>>>> the dropped streams NEXT expected SSN... You would be adding 4 bytes
> > >>>>> for EVERY stream dropping one or more TSN's... note that this does
> > >>>>> NOT mean 4 bytes per TSN but 4 bytes per where one or more TSN's
> > >>>>> were
> > >>>>> dropped..
> > >>>>>
> > >>>>>
> > >>>>> I think this is reasonable.. it improves processing AND avoids HOL
> > >>>>> blocking..
> > >>>>>
> > >>>>>
> > >>>>> Michael T/Phil? what do you think... Michael R.. I will talk to you
> > >>>>> at
> > >>>>> the
> > >>>>> next class break..
> > >>>>>
> > >>>>>
> > >>>>>
> > >>>>> R
> > >>>>>
> > >>>>> --
> > >>>>> 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)
> > >>
> > >> --
> > >> Brian F. G. Bidulock    ¦ The reasonable man adapts himself to the ¦
> > >> bidulock@openss7.org    ¦ world; the unreasonable one persists in  ¦
> > >> http://www.openss7.org/ ¦ trying  to adapt the  world  to himself. ¦
> > >>                         ¦ Therefore  all  progress  depends on the ¦
> > >>                         ¦ unreasonable man. -- George Bernard Shaw ¦
> > >
> > > --
> > > 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
> > >
> > >
> > Michael.Tuexen@micmac.franken.de

-- 
Brian F. G. Bidulock    ¦ The reasonable man adapts himself to the ¦
bidulock@openss7.org    ¦ world; the unreasonable one persists in  ¦
http://www.openss7.org/ ¦ trying  to adapt the  world  to himself. ¦
                        ¦ Therefore  all  progress  depends on the ¦
                        ¦ unreasonable man. -- George Bernard Shaw ¦

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



From daemon@optimus.ietf.org  Fri Jun  7 20:26:30 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 UAA17795
	for <tsvwg-archive@odin.ietf.org>; Fri, 7 Jun 2002 20:26:30 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id UAA17285
	for tsvwg-archive@odin.ietf.org; Fri, 7 Jun 2002 20:27:00 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id UAA16066;
	Fri, 7 Jun 2002 20:05:38 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id MAA04068
	for <tsvwg@optimus.ietf.org>; Fri, 7 Jun 2002 12:54:19 -0400 (EDT)
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 MAA03900
	for <tsvwg@ietf.org>; Fri, 7 Jun 2002 12:53:46 -0400 (EDT)
Received: from mira-sjc5-2.cisco.com (IDENT:mirapoint@mira-sjc5-2.cisco.com [171.71.163.16])
	by sj-msg-core-1.cisco.com (8.12.2/8.12.2) with ESMTP id g57GreHs028972;
	Fri, 7 Jun 2002 09:53:40 -0700 (PDT)
Received: from stewart.chicago.il.us (dhcp-171-69-54-231.cisco.com [171.69.54.231])
	by mira-sjc5-2.cisco.com (Mirapoint)
	with ESMTP id ACP36196;
	Fri, 7 Jun 2002 09:53:38 -0700 (PDT)
Message-ID: <3D00E50E.24856650@stewart.chicago.il.us>
Date: Fri, 07 Jun 2002 11:53:34 -0500
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: from dillip <dillip55@yahoo.com>
CC: tsvwg@ietf.org
Subject: Re: [Tsvwg] please give me yours feedback.
References: <20020606075027.559.qmail@web20605.mail.yahoo.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

DP:

As I said to you in private... comments below..


from dillip wrote:
> 
> Dear Sir !
>    I have some query on SCTP. Please help me regarding
> this issue .
> 
>    I gone through RFC#2960 for implementing SCTP stack
> and found some problem related to Gap-Ack-Ack . I am
> writing details of the problem below here and please
> reply me back .
> 
> Accoridng to RFC:::
> 3.3.4 Selective Acknowledgement (SACK) (3):
>    Gap Ack Block Start: 16 bits (unsigned integer)
> 
>       Indicates the Start offset TSN for this Gap Ack
> Block.  To
>       calculate the actual TSN number the Cumulative
> TSN Ack is added to
>       this offset number.  This calculated TSN
> identifies the first TSN
>       in this Gap Ack Block that has been received.
> 
>    Gap Ack Block End:  16 bits (unsigned integer)
>    Indicates the End offset TSN for this Gap Ack
> Block.  To calculate the actual TSN number the
> Cumulative TSN Ack is added to this offset number.
> This calculated TSN identifies the TSN of the last
> DATA chunk received in this Gap Ack Block.
> 
> Scenario-1>>>>>>>>
> ------------------
>                         ----------
>                         | TSN=17 |
>                         ----------
>                         |        | <- still missing
>                         ----------
>                         | TSN=15 |
>                         ----------
>                         | TSN=14 |
>                         ----------
>                         |        | <- still missing
>                         ----------
>                         | TSN=12 |
>                         ----------
>                         | TSN=11 |
>                         ----------
>                         | TSN=10 |
>                         ----------
> 
>    then, the parameter part of the SACK MUST be
> constructed as follows (assuming the new a_rwnd is set
> to 4660 by the sender):
> 
>                   +--------------------------------+
>                   |   Cumulative TSN Ack = 12      |
>                   +--------------------------------+
>                   |        a_rwnd = 4660           |
>                   +----------------+---------------+
>                   | num of block=2 | num of dup=0  |
>                   +----------------+---------------+
>                   |block #1 strt=2 |block #1 end=3 |
>                   +----------------+---------------+
>                   |block #2 strt=5 |block #2 end=5 |
>                   +----------------+---------------+
> 

The above SACK is correct... it is what would appear on the wire..


>         Here the actual TSN number should calculated by :::
>                 actual_tsn = cumulative_tsn + offset_number ;
>         If this is correct then for the above scenario result
> will be :::
>                   +--------------------------------+
>                   |   Cumulative TSN Ack = 12      |
>                   +--------------------------------+
>                   |        a_rwnd = 4660           |
>                   +----------------+---------------+
>                   | num of block=2 | num of dup=0  |
>                   +----------------+---------------+
>                   |block #1 strt=1 |block #1 end=1 |
>                   +----------------+---------------+
>                   |block #2 strt=4 |block #2 end=4 |
>                   +----------------+---------------+


I have NO idea what the aboe is... it IS NOT what is on the wire... but
instead the conclusion of what one receiving the above SACK would reach.

The key to remember is we send gaps showing what we have RECEIVED not
what is missing.. the reciever of the SACK must figure out what is
missing..

> 
> Scenario-2>>>>>>>>
> ------------------
>         If the scenario looks::::::::::::::
>                         ----------
>                         | TSN=17 |
>                         ----------
>                         |        | <- still missing
>                         ----------
>                         | TSN=15 |
>                         ----------
>                         |          |
>                         ----------
>                         |        | <- still missing
>                         ----------
>                         | TSN=12 |
>                         ----------
>                         | TSN=11 |
>                         ----------
>         If this is correct then for the above scenario result
> will be :::
>                   +--------------------------------+
>                   |   Cumulative TSN Ack = 12      |
>                   +--------------------------------+
>                   |        a_rwnd = 4660           |
>                   +----------------+---------------+
>                   | num of block=2 | num of dup=0  |
>                   +----------------+---------------+
>                   |block #1 strt=1 |block #1 end=2 |
>                   +----------------+---------------+
>                   |block #2 strt=4 |block #2 end=4 |
>                   +----------------+---------------+
> 

The ABOVE IS INCORRECT... it is NOT what would be on the wire... 

> 
>         Is it correct or I need to folow some other algorithm
> to calculate the gaps ??
> 

For the receiver of data .. yes you MUST use an algorithm to
calculate what to send in the SACK. The algorithm I used
in both the user space ref impl or the kernel space ref impl
available at:

http://www.sctp.org

is I take a bit (or byte) array with a array offset TSN number

So to find out if a TSN is received you take the base TSN of the
array subtract it from the TSN you wish to check... go to that
bit and see if it is a one or a zero... Note in the user
space implemention I burned more memory and did this on a byte basis...

I am sure there are other methods as well...

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  Fri Jun  7 20:26:35 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 UAA17812
	for <tsvwg-archive@odin.ietf.org>; Fri, 7 Jun 2002 20:26:35 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id UAA17299
	for tsvwg-archive@odin.ietf.org; Fri, 7 Jun 2002 20:27:05 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id UAA16033;
	Fri, 7 Jun 2002 20:05:36 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id MAA10874
	for <tsvwg@optimus.ietf.org>; Fri, 7 Jun 2002 12:43:33 -0400 (EDT)
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 MAA03720
	for <tsvwg@ietf.org>; Fri, 7 Jun 2002 12:43:01 -0400 (EDT)
Received: from mira-sjc5-2.cisco.com (IDENT:mirapoint@mira-sjc5-2.cisco.com [171.71.163.16])
	by sj-msg-core-1.cisco.com (8.12.2/8.12.2) with ESMTP id g57Gh2Hs022384;
	Fri, 7 Jun 2002 09:43:02 -0700 (PDT)
Received: from stewart.chicago.il.us (dhcp-171-69-54-231.cisco.com [171.69.54.231])
	by mira-sjc5-2.cisco.com (Mirapoint)
	with ESMTP id ACP35803;
	Fri, 7 Jun 2002 09:43:01 -0700 (PDT)
Message-ID: <3D00E291.BF6AF3EC@stewart.chicago.il.us>
Date: Fri, 07 Jun 2002 11:42:57 -0500
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 <qxie1@email.mot.com>
CC: bidulock@openss7.org, tsvwg@ietf.org
Subject: Re: [Tsvwg] Re: your mail
References: <20020606184151.D20729@openss7.org> <3D001BC8.51B30A14@email.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 sounds fine to me too... we will get it into the next
version as well as the other word changes...

Note to brian: my cisco account is rrs@cisco.com NOT randall@cisco.com

R

Qiaobing Xie wrote:
> 
> ...
> > The proposed solution is that C2 be reworded to advance the
> > Adv.Peer.Ack.Pnt only over "abandonned" TSNs.  So, in the above
> > example, the FORWARD TSN would be:
> >
> >   FORWARD TSN 100 --->
> >
> 
> This looks like a fine solution to me.
> 
> -Qiaobing
> 
> > If there are no renegs, the peer should immediately reply with
> >
> >   <-- SACK (CUM ACK 104)
> >
> > anyway (this was Armando's point earlier in this thread), so
> > performance is not impaired and reliability is restored.
> >
> > --brian
> 
> _______________________________________________
> 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  Fri Jun  7 20:26:52 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 UAA17830
	for <tsvwg-archive@odin.ietf.org>; Fri, 7 Jun 2002 20:26:52 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id UAA17313
	for tsvwg-archive@odin.ietf.org; Fri, 7 Jun 2002 20:27:22 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id UAA16086;
	Fri, 7 Jun 2002 20:05:39 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id PAA01313
	for <tsvwg@optimus.ietf.org>; Fri, 7 Jun 2002 15:47:20 -0400 (EDT)
Received: from motgate.mot.com (motgate.mot.com [129.188.136.100])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id PAA10566
	for <tsvwg@ietf.org>; Fri, 7 Jun 2002 15:46:48 -0400 (EDT)
Received: [from pobox4.mot.com (pobox4.mot.com [10.64.251.243]) by motgate.mot.com (motgate 2.1) with ESMTP id MAA13418; Fri, 7 Jun 2002 12:47:12 -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 MAA28705; Fri, 7 Jun 2002 12:47:11 -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 g57Jl8b13320;
	Fri, 7 Jun 2002 14:47:08 -0500 (CDT)
Received: from motorola.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 OAA28365;
	Fri, 7 Jun 2002 14:47:29 -0500 (CDT)
Message-ID: <3D010E5E.D51114EF@motorola.com>
Date: Fri, 07 Jun 2002 14:49:50 -0500
From: Qiaobing Xie <Qiaobing.Xie@motorola.com>
X-Mailer: Mozilla 4.77 [en] (Win98; U)
X-Accept-Language: en
MIME-Version: 1.0
To: Michael Tuexen <Michael.Tuexen@lurchi.franken.de>
CC: Randall Stewart <randall@stewart.chicago.il.us>, bidulock@openss7.org,
        Phillip Conrad <conrad@acm.org>,
        "Armando L. Caro Jr." <acaro@mail.eecis.udel.edu>, tsvwg@ietf.org,
        "Michael A. Ramalho" <mramalho@cisco.com>
Subject: Re: [Tsvwg] More on PRSCTP-00
References: <D4966AFE-7A3E-11D6-9E0D-0030654C1AB6@lurchi.franken.de>
Content-Type: text/plain; charset=iso-8859-1
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

Michael,

Right on! You summarized my previous point beautifully. I start to
realize that lately my arguments tend to become lengthy and off the
point... maybe a sign of getting old :<

So far I kind of buy Randy's point better that "adding some bits in
exchange for ease of implementation", but still yet to be convinced
about the HOL blocking argument made by Brian. 

I've asked for an "all-considered" example of the HOL blocking, but
haven't seen one yet.

regards,
-Qiaobing

Michael Tuexen wrote:
> 
> Dear all,
> 
> reading the thread I would like to make some remarks:
> 
> 1. I agree that using SSN in the FWD chunk avoids HOL (which is good).
> 2. Brians example show how HOL can occur if no SSN info in the FWD chunk
>     is used. He only looks at the receiver.
> 
> Brians example are always based on the assumption that there is a GAP
> in the TSN space at the receiver side and the receiver gets a FWD TSN
> which does not fill the whole gap. Then the HOL happens at the receiver.
> 
> I'm just wondering how often this will happen. What are the reasons that
> a FWD chunk will be send: RTX Timer expiration or fast retransmission.
> In both cases I would send a FWD chunk and bundle the remaining DATA
> chunks
> (if possible). So there is a good chance that the whole gap is filled by
> the FWD chunk and the retransmitted DATA chunks.
> I'm not sure how likely Brians scenario is taking the sender into
> account.
> 
> So I think one has to take the senders behavior into account when
> discussing
> this issue.
> 
> Best regards
> Michael
> 
> On Thursday, June 6, 2002, at 01:43 AM, Randall Stewart wrote:
> 
> > "Brian F. G. Bidulock" wrote:
> >>
> >> Randall,
> >>
> >> On Wed, 05 Jun 2002, Randall Stewart wrote:
> >>
> >>> A few comments..
> >>>
> >>> Phillip Conrad wrote:
> >>>>
> >>>> Dear Colleagues:
> >>>>       I agree with Randy that Brian has raised an important issue
> >>>> for us to
> >>>> consider, and I think has explained it quite well with a very clear,
> >>>> illustrative example.
> >>>>
> >>>> The question is one of a tradeoff between two designs:
> >>>>
> >>>> (A) current design: simpler, fewer bits on the wire, but may
> >>>> (re)-introduce
> >>>> an HOL blocking problem
> >>>
> >>> I must disagree with your "it is simpler" ... it is NOT. The question
> >>> is
> >>> where is the burden. Having implemented pr-sctp in the kernel of
> >>> freebsd
> >>> NOT having the stream number/SSN in the FWD-TSN means I must keep
> >>> an extra set of linked lists and such to track any stream that has
> >>> pending stuck data. When a fwd-tsn arrives I must HUNT all of these
> >>> streams and look at each inbound queue.
> >>>
> >>> The counter to this is the sender (in Brian and my earlier proposal
> >>> that was to Q only) must track this info to send with the FWD-TSN.
> >>> Actually
> >>> less of a burden IMHO...
> >>
> >> I agree.  I discovered this situation while implementing PR-SCTP in the
> >> Linux Kernel.  The normal receiver can rely on the fact that the next
> >> expected SSN is the last fully received SSN plus 1.  PR-SCTP without
> >> a forward SSN causes the receive to have to update this expected value
> >> with the next SSN in the reorder queue whenever there are no TSN gaps.
> >> This even might require that when TSN gaps are filled that the receiver
> >> walk all inbound stream structures to update the next expected SSN.
> >> In an independent implementation, I have arrived at the same
> >> conclusion:
> >> that the reciever has more work to do for FORWARD TSN without a forward
> >> SSN.
> >>
> >
> > Yeah.. I basically had to make a special linked list set that the
> > stream queues could be placed on.. a bit more efficient but a
> > reall hassel to maintian a "wheel" (as I call it) of streams with
> > pending data on it...
> >
> >> In fact the receiver has this work to do if it has ever received a
> >> FORWARD TSN.  This is because it is not possible for the receiver to
> >> know a priori which streams could be affected by this scenairo.
> >> Therefore,
> >> the load on the receiver is increased even when this case does not
> >> occur.
> >> It is the mere possibility of its occurence which introduces the load.
> >> (Each time gaps close, the next expected SSN for all streams need to be
> >> re-examined whether there are SSN gaps or not.)
> >
> > Exactly.. I constantly must place stream q;s on and off the :"wheel" no
> > matter if a fwd-tsn is EVER received...
> >
> >>
> >> On the sender side, the sender must always track the last sent SSN and
> >> assign the next SSN (++SSN).  Indeed the sender when generating a
> >> FORWARD
> >> TSN chunk would need to traverse outbound stream structures with
> >> dropped
> >> TSNs to formulate Forward TSNs, but this is likely to be a far shorter
> >> list than at the receiver and non-PR streams would never need to be
> >> involved.
> >
> > And it must continue to KEEP this information until it is sure the
> > peer has received and moved the cum-ack forward..
> >
> > What I end up doing on the sender side is I hold just the bookkeeping
> > structures of the TSN (stream seq/stream no/tsn ... etc) but I
> > drop the mbuf holding the data so I gain a bit of help in mbuf
> > utilization.
> >
> > But I can't get rid of the chunk structure UNTIL the cum-ack says he
> > got my FWD-TSN and accepted it...
> >
> > So there is NO additional info to save on the sender side and only
> > do I save what I must... on the receiver side I have a LOT of extra
> > bookkeeping....
> >
> >
> > In fact with your confirmation I would claim just the opposite of
> > Phil's statement... that doing it the way it is NOW documented takes
> > MORE overhead.. only by adding the SSN/SID can you reduce this..
> >
> > By the way Renee also pointed out the extra overhead for the IOS
> > implementation as well :>... so this makes three different
> > implementations
> > NOT 2....
> >
> > I plan on adding this to the next pr-sctp unless Michael Tuexen
> > violently
> > agrees with Qiaobing ...
> >
> > I still await Michaels answer????
> >
> >
> >>
> >> --brian
> >>
> >>>>
> >>>> (B) alternative design: more complex, may require larger Forward TSN
> >>>> chunks, but avoids re-introducing an HOL blocking problem, and thus
> >>>> improves performance in some situations.
> >>>
> >>>
> >>> see above it is NOT more complex...
> >>>
> >>>>
> >>>> The question of whether to choose A or B depends on whether you
> >>>> believe
> >>>> that the situation Brian describes is a "corner case" or not.   So
> >>>> I've
> >>>> walked through Brian's example, and have determined that I don't
> >>>> think it
> >>>> is a corner case at all.   So, I'm inclined to agree with Randy and
> >>>> Brian
> >>>> that we should consider a change.
> >>>>
> >>>> But I'd like to invite others to walk through my line of reasoning
> >>>> to test
> >>>> whether I've reasoned this correctly.
> >>>
> >>>
> >>> I agree it is NOT a corner case..
> >>>
> >>>>
> >>>> My rationale that Brian's example is NOT a corner case:
> >>>> ========================================
> >>>>
> >>>> In Brian's example, messages are being sent round-robin on a number
> >>>> of
> >>>> streams.  This seems like a fairly reasonable use of SCTP streams,
> >>>> do that
> >>>> aspect does not strike me as unusual.
> >>>
> >>> I agree :>
> >>>
> >>>>
> >>>> Now for the next part of this explanation, I'm going to have to infer
> >>>> certain things, so, Brian, please correct me if I infer wrongly.
> >>>> *One*
> >>>> possibility for the sequence of events leading to the
> >>>> situation Brian describes is that there is a burst loss of data
> >>>> chunks
> >>>> TSN110 through TSN119 (ten chunks, but possibly less than ten
> >>>> packets..
> >>>> perhaps even only ONE packet if the chunks are small enough to be
> >>>> bundled!)... and then data chunks TSN120 through TSN128 make it
> >>>> through
> >>>> OK.   This does not strike me as particularly unusual; a loss of a
> >>>> single
> >>>> packet followed by a successful packet is sufficient to cause this
> >>>> situation.
> >>>>
> >>>
> >>> Or the other possiblility would be a set of bundled chunks in one
> >>> packet
> >>> being lost.. either way is the same result I would think..
> >>>
> >>>> The next thing that happens is that TSNs 110 through 118 are
> >>>> "dropped" in
> >>>> the technical sense of that word; see the defn in the current
> >>>> draft.  This
> >>>> means that the sender gives up on sending them, and issues a forward
> >>>> TSN.   However, no Forward TSN is issued for TSN119.  This may be
> >>>> because
> >>>> the messages on stream 9 are treated as Reliable.   This is NOT an
> >>>> unusual
> >>>> situation; it is perfectly reasonably to assume that an application
> >>>> might
> >>>> have "k" streams with data that can be dropped, and "n - k" streams
> >>>> of data
> >>>> that must be treated as reliable.  In the case of Brian's example, n
> >>>> is 10,
> >>>> and k is 9, but I think that his example will still produce similar
> >>>> results
> >>>> even for other values of "k" and "n".
> >>>>
> >>>
> >>> Note.. streams themselves are not reliable or partially reliable. With
> >>> the new draft it is the message itself which is sent with the
> >>> reliablilty
> >>> set to "fully reliable". One possilbe implementation would be to make
> >>> things correlate to a particular stream.. but this is just an
> >>> implementation
> >>> option..
> >>>
> >>>> Bottom line:
> >>>> ========
> >>>>
> >>>> I don't think its a corner case at all.   We have all the PR streams
> >>>> potentially being HOL blocked by loss of a single packet.  I think
> >>>> we need
> >>>> to fix this.
> >>>
> >>> I agree..
> >>>
> >>>>
> >>>> The solution of forward SSNs is one viable solution I think.  There
> >>>> may be
> >>>> others, but that's as good as any for right now I think.   There may
> >>>> be
> >>>> another solution as well, but let me do some writing and thinking
> >>>> about it
> >>>> off line before I muddy the waters with it here... just to make sure
> >>>> it is
> >>>> a viable alternative.
> >>>
> >>> agreed
> >>>
> >>>
> >>>>
> >>>> Randy... I'm willing to make the text changes and send a proposed
> >>>> version
> >>>> around to the co-authors for comment if there is consensus to move
> >>>> in that
> >>>> direction.   Let me know.
> >>>>
> >>>
> >>> I will get it out this weekend if I may.. there are some interesting
> >>> things that I have thought about that must be covered... I had
> >>> something
> >>> like this in an old u-sctp draft (during Q and I's debate earlier :>)
> >>>
> >>>
> >>>
> >>>>
> >>>> (Side note: I was worried that the "technical definition" of the word
> >>>> "dropped" would cause confusion, and it already has.   Perhaps
> >>>> should use
> >>>> another word in the draft to describe the operation where the sender
> >>>> decides to no longer transmit and/or retransmit a specific data
> >>>> chunk.  How
> >>>> about the term "abandoned" in place of "dropped"?)
> >>>
> >>>
> >>> Sounds good..
> >>>
> >>> note, Michael R also agrees .. or seems to .. we will talk more..
> >>>
> >>> R
> >>>
> >>>
> >>>>
> >>>> /ptc
> >>>>
> >>>> At 11:44 AM 6/5/2002 -0500, Randall Stewart wrote:
> >>>>> Brian/Qiaobing:
> >>>>>
> >>>>> I have read through this interesting debate. Qiaobing and
> >>>>> I had a major discussion on this way back when it was
> >>>>> considered U-SCTP. Renee Revis first pointed this out to me.. but
> >>>>> from a different persepective i.e. the receiver being able to
> >>>>> efficently process stuck TSN's ...
> >>>>>
> >>>>> After a lengthy debate I agreed to not have the SSN/SID listed
> >>>>> in the FWD TSN... I did NOT think of the HOL blocking issue.. or
> >>>>> I would have used it as a issue (I was the one arguing for the
> >>>>> SSN/SID in the FWD-TSN :->).
> >>>>>
> >>>>> After watching this debate.. I think I have come to the conclusion
> >>>>> that
> >>>>> Brian is correct :)
> >>>>>
> >>>>> Michael R. is here... I will discuss it with him.. I want to get
> >>>>> the other co-author's take on this one... If the others agree with
> >>>>> Brian (and me :>) then I think we will put this in.... also I would
> >>>>> like to hear what the rest of the WG feels on this???
> >>>>>
> >>>>>
> >>>>> I think it is reasonable and not that much overhead to include JUST
> >>>>> the dropped streams NEXT expected SSN... You would be adding 4 bytes
> >>>>> for EVERY stream dropping one or more TSN's... note that this does
> >>>>> NOT mean 4 bytes per TSN but 4 bytes per where one or more TSN's
> >>>>> were
> >>>>> dropped..
> >>>>>
> >>>>>
> >>>>> I think this is reasonable.. it improves processing AND avoids HOL
> >>>>> blocking..
> >>>>>
> >>>>>
> >>>>> Michael T/Phil? what do you think... Michael R.. I will talk to you
> >>>>> at
> >>>>> the
> >>>>> next class break..
> >>>>>
> >>>>>
> >>>>>
> >>>>> R
> >>>>>
> >>>>> --
> >>>>> 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)
> >>
> >> --
> >> Brian F. G. Bidulock    ¦ The reasonable man adapts himself to the ¦
> >> bidulock@openss7.org    ¦ world; the unreasonable one persists in  ¦
> >> http://www.openss7.org/ ¦ trying  to adapt the  world  to himself. ¦
> >>                         ¦ Therefore  all  progress  depends on the ¦
> >>                         ¦ unreasonable man. -- George Bernard Shaw ¦
> >
> > --
> > 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
> >
> >
> Michael.Tuexen@micmac.franken.de


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



From daemon@optimus.ietf.org  Fri Jun  7 20:31:35 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 UAA17954
	for <tsvwg-archive@odin.ietf.org>; Fri, 7 Jun 2002 20:31:35 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id UAA17932
	for tsvwg-archive@odin.ietf.org; Fri, 7 Jun 2002 20:32:05 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id UAA16109;
	Fri, 7 Jun 2002 20:05:42 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id MAA10645
	for <tsvwg@optimus.ietf.org>; Fri, 7 Jun 2002 12:37:53 -0400 (EDT)
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 MAA03600
	for <tsvwg@ietf.org>; Fri, 7 Jun 2002 12:37:20 -0400 (EDT)
Received: from mira-sjc5-2.cisco.com (IDENT:mirapoint@mira-sjc5-2.cisco.com [171.71.163.16])
	by sj-msg-core-4.cisco.com (8.12.2/8.12.2) with ESMTP id g57GbDIZ007338;
	Fri, 7 Jun 2002 09:37:18 -0700 (PDT)
Received: from stewart.chicago.il.us (dhcp-171-69-54-231.cisco.com [171.69.54.231])
	by mira-sjc5-2.cisco.com (Mirapoint)
	with ESMTP id ACP35593;
	Fri, 7 Jun 2002 09:37:12 -0700 (PDT)
Message-ID: <3D00E134.D5981833@stewart.chicago.il.us>
Date: Fri, 07 Jun 2002 11:37:08 -0500
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 <qxie1@email.mot.com>
CC: bidulock@openss7.org, Qiaobing Xie <Qiaobing.Xie@motorola.com>,
        "Armando L. Caro Jr." <acaro@mail.eecis.udel.edu>, tsvwg@ietf.org
Subject: Re: [Tsvwg] More on PRSCTP-00
References: <20020531195333.D11468@openss7.org> <3CFBC11C.1EFA747@email.mot.com> <20020603144342.A30667@openss7.org> <3CFC144D.65697A41@email.mot.com> <20020603215952.E30667@openss7.org> <3CFD57D8.5637E431@email.mot.com> <20020604195020.A16020@openss7.org> <3CFE7403.59FB6F4C@email.mot.com> <20020605163502.A2012@openss7.org> <3CFE8C3F.C935774C@email.mot.com> <20020605171556.C2012@openss7.org> <3CFE9238.930A5B15@email.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

Q:

Qiaobing Xie wrote:
> 
> ...
> > I'm afraid you're on your own on this one.
> 
> Sadly but that seems to be the case :-)

I must agree here :<  

Having gone through this debate before with you in private
I feel rather badly that I have "changed my mind" in public :0
But as you know I always thought it was more efficent processing 
wise but I did agree that if there were no other benefit to the
bits... then this would just be a "cost to pay". But I think
with Brian's additional HOL issue it changes the balance.

I do think this thread has had a very calm tone (compared to
issues in the past)... I don't think anything was said in anything
but concern for the BEST solution that we all can reach "rough"
consensus
on :>

> 
> After all, democratic consensus building is meant to bring what the
> *group as a whole* views as the best. But, make no mistake, we are *not*
> the same as what's called "designed by committed". We are better than
> "them" :-)


Here I could not agree with you more :>


Q: When I get back from my travels I will see if I can't make ammends
   to you on my "change" by offering you a glass of cab at my house :>


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  Fri Jun  7 21:20:38 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 VAA18639
	for <tsvwg-archive@odin.ietf.org>; Fri, 7 Jun 2002 21:20:36 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id VAA19937
	for tsvwg-archive@odin.ietf.org; Fri, 7 Jun 2002 21:21:07 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id UAA18845;
	Fri, 7 Jun 2002 20:52:15 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id UAA18763
	for <tsvwg@optimus.ietf.org>; Fri, 7 Jun 2002 20:52:11 -0400 (EDT)
Received: from bfgbhome.inetint.com (IDENT:root@tnt-dal-42-211.dallas.net [209.44.42.211])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id UAA18336
	for <tsvwg@ietf.org>; Fri, 7 Jun 2002 20:51:36 -0400 (EDT)
Received: (from brian@localhost)
	by bfgbhome.inetint.com (8.9.3/8.9.3) id TAA10042;
	Fri, 7 Jun 2002 19:49:54 -0500
Date: Fri, 7 Jun 2002 19:49:54 -0500
From: "Brian F. G. Bidulock" <bidulock@openss7.org>
To: Qiaobing Xie <Qiaobing.Xie@motorola.com>
Cc: Michael Tuexen <Michael.Tuexen@lurchi.franken.de>,
        Randall Stewart <randall@stewart.chicago.il.us>,
        Phillip Conrad <conrad@acm.org>,
        "Armando L. Caro Jr." <acaro@mail.eecis.udel.edu>, tsvwg@ietf.org,
        "Michael A. Ramalho" <mramalho@cisco.com>
Subject: Re: [Tsvwg] More on PRSCTP-00
Message-ID: <20020607194954.B8970@openss7.org>
Reply-To: bidulock@openss7.org
References: <D4966AFE-7A3E-11D6-9E0D-0030654C1AB6@lurchi.franken.de> <3D010E5E.D51114EF@motorola.com> <20020607153911.D2907@openss7.org> <3D015235.C2767BA1@motorola.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
User-Agent: Mutt/1.2i
In-Reply-To: <3D015235.C2767BA1@motorola.com>; from Qiaobing.Xie@motorola.com on Fri, Jun 07, 2002 at 07:39:17PM -0500
Organization: http://www.openss7.org/
Dsn-Notification-To: <bidulock@openss7.org>
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

Qiaobing,

On Fri, 07 Jun 2002, Qiaobing Xie wrote:

> "Brian F. G. Bidulock" wrote:
> > 
> > Michael, Qiaobing,
> > 
> > Here is the sender side of the example.  It is quite reasonable
> > that this could occur.  The receiver is blocked on streams 1
> > through 9 until retransmission of TSN 119 is successful.  This
> > is the sender side of the same example.
> > 
> >     TSN 101  SID 1   SSN 1   |
> >     TSN 102  SID 2   SSN 1   |
> >     TSN 103  SID 3   SSN 1   |
> >     TSN 104  SID 4   SSN 1   |   SCTP Packet
> >     TSN 105  SID 5   SSN 1   +--------------------->
> >     TSN 106  SID 6   SSN 1   |
> >     TSN 107  SID 7   SSN 1   |
> >     TSN 108  SID 8   SSN 1   |
> >     TSN 109  SID 9   SSN 1   |

        <---- SACK CUM ACK 109

> > 
> >     TSN 110  SID 1   SSN 2   |
> >     TSN 111  SID 2   SSN 2   |
> >     TSN 112  SID 3   SSN 2   |
> >     TSN 113  SID 4   SSN 2   |   SCTP Packet
> >     TSN 114  SID 5   SSN 2   +----------X
> >     TSN 115  SID 6   SSN 2   |         Lost
> >     TSN 116  SID 7   SSN 2   |
> >     TSN 117  SID 8   SSN 2   |
> >     TSN 118  SID 9   SSN 2   |

        <---- SACK CUM ACK 109

> >                                  SCTP Packet
> >     TSN 119  SID 0   SSN X   |----------X
> >                                        Lost
> > 
> >     (Timed reliability expiry on TSNs 110 to 118)

        <---- SACK CUM ACK 109

> > 
> >     FORWARD TSN 118          |--------------------->
> 
> Why TSN119 is not re-sent at this exactly same moment? As Michael and I
> both noted that, in PR-SCTP, Fwd TSN can only be triggered due to either
> a "RTX Timer expiration or fast retransmission." Could you fill in the
> details that lead to TSN119 not being re-sent at this moment?

With the SACKs shown, the FORWARD TSN "MUST" be sent according to
Rule C3.  Retransmission of TSN 119 has nothing to do with it.

--brian

> 
> > 
> >     TSN 120  SID 1   SSN 3   |
> >     TSN 121  SID 2   SSN 3   |
> >     TSN 122  SID 3   SSN 3   |
> >     TSN 123  SID 4   SSN 3   |   SCTP Packet
> >     TSN 124  SID 5   SSN 3   +--------------------->
> >     TSN 125  SID 6   SSN 3   |
> >     TSN 126  SID 7   SSN 3   |
> >     TSN 127  SID 8   SSN 3   |
> >     TSN 128  SID 9   SSN 3   |
> > 
> >     (Receiver is now blocked on streams 1 through 9.)
> > 
> >     (Some time later, retrans timeout on TSN 119)
> > 
> >                                  SCTP Packet
> >     TSN 119  SID 0   SSN X   |--------------------->
> > 
> > Q.E.D.
> > 
> > --brian
> > 
> > On Fri, 07 Jun 2002, Qiaobing Xie wrote:
> > 
> > > Michael,
> > >
> > > Right on! You summarized my previous point beautifully. I start to
> > > realize that lately my arguments tend to become lengthy and off the
> > > point... maybe a sign of getting old :<
> > >
> > > So far I kind of buy Randy's point better that "adding some bits in
> > > exchange for ease of implementation", but still yet to be convinced
> > > about the HOL blocking argument made by Brian.
> > >
> > > I've asked for an "all-considered" example of the HOL blocking, but
> > > haven't seen one yet.
> > >
> > > regards,
> > > -Qiaobing
> > >
> > > Michael Tuexen wrote:
> > > >
> > > > Dear all,
> > > >
> > > > reading the thread I would like to make some remarks:
> > > >
> > > > 1. I agree that using SSN in the FWD chunk avoids HOL (which is good).
> > > > 2. Brians example show how HOL can occur if no SSN info in the FWD chunk
> > > >     is used. He only looks at the receiver.
> > > >
> > > > Brians example are always based on the assumption that there is a GAP
> > > > in the TSN space at the receiver side and the receiver gets a FWD TSN
> > > > which does not fill the whole gap. Then the HOL happens at the receiver.
> > > >
> > > > I'm just wondering how often this will happen. What are the reasons that
> > > > a FWD chunk will be send: RTX Timer expiration or fast retransmission.
> > > > In both cases I would send a FWD chunk and bundle the remaining DATA
> > > > chunks
> > > > (if possible). So there is a good chance that the whole gap is filled by
> > > > the FWD chunk and the retransmitted DATA chunks.
> > > > I'm not sure how likely Brians scenario is taking the sender into
> > > > account.
> > > >
> > > > So I think one has to take the senders behavior into account when
> > > > discussing
> > > > this issue.
> > > >
> > > > Best regards
> > > > Michael
> > > >
> > > > On Thursday, June 6, 2002, at 01:43 AM, Randall Stewart wrote:
> > > >
> > > > > "Brian F. G. Bidulock" wrote:
> > > > >>
> > > > >> Randall,
> > > > >>
> > > > >> On Wed, 05 Jun 2002, Randall Stewart wrote:
> > > > >>
> > > > >>> A few comments..
> > > > >>>
> > > > >>> Phillip Conrad wrote:
> > > > >>>>
> > > > >>>> Dear Colleagues:
> > > > >>>>       I agree with Randy that Brian has raised an important issue
> > > > >>>> for us to
> > > > >>>> consider, and I think has explained it quite well with a very clear,
> > > > >>>> illustrative example.
> > > > >>>>
> > > > >>>> The question is one of a tradeoff between two designs:
> > > > >>>>
> > > > >>>> (A) current design: simpler, fewer bits on the wire, but may
> > > > >>>> (re)-introduce
> > > > >>>> an HOL blocking problem
> > > > >>>
> > > > >>> I must disagree with your "it is simpler" ... it is NOT. The question
> > > > >>> is
> > > > >>> where is the burden. Having implemented pr-sctp in the kernel of
> > > > >>> freebsd
> > > > >>> NOT having the stream number/SSN in the FWD-TSN means I must keep
> > > > >>> an extra set of linked lists and such to track any stream that has
> > > > >>> pending stuck data. When a fwd-tsn arrives I must HUNT all of these
> > > > >>> streams and look at each inbound queue.
> > > > >>>
> > > > >>> The counter to this is the sender (in Brian and my earlier proposal
> > > > >>> that was to Q only) must track this info to send with the FWD-TSN.
> > > > >>> Actually
> > > > >>> less of a burden IMHO...
> > > > >>
> > > > >> I agree.  I discovered this situation while implementing PR-SCTP in the
> > > > >> Linux Kernel.  The normal receiver can rely on the fact that the next
> > > > >> expected SSN is the last fully received SSN plus 1.  PR-SCTP without
> > > > >> a forward SSN causes the receive to have to update this expected value
> > > > >> with the next SSN in the reorder queue whenever there are no TSN gaps.
> > > > >> This even might require that when TSN gaps are filled that the receiver
> > > > >> walk all inbound stream structures to update the next expected SSN.
> > > > >> In an independent implementation, I have arrived at the same
> > > > >> conclusion:
> > > > >> that the reciever has more work to do for FORWARD TSN without a forward
> > > > >> SSN.
> > > > >>
> > > > >
> > > > > Yeah.. I basically had to make a special linked list set that the
> > > > > stream queues could be placed on.. a bit more efficient but a
> > > > > reall hassel to maintian a "wheel" (as I call it) of streams with
> > > > > pending data on it...
> > > > >
> > > > >> In fact the receiver has this work to do if it has ever received a
> > > > >> FORWARD TSN.  This is because it is not possible for the receiver to
> > > > >> know a priori which streams could be affected by this scenairo.
> > > > >> Therefore,
> > > > >> the load on the receiver is increased even when this case does not
> > > > >> occur.
> > > > >> It is the mere possibility of its occurence which introduces the load.
> > > > >> (Each time gaps close, the next expected SSN for all streams need to be
> > > > >> re-examined whether there are SSN gaps or not.)
> > > > >
> > > > > Exactly.. I constantly must place stream q;s on and off the :"wheel" no
> > > > > matter if a fwd-tsn is EVER received...
> > > > >
> > > > >>
> > > > >> On the sender side, the sender must always track the last sent SSN and
> > > > >> assign the next SSN (++SSN).  Indeed the sender when generating a
> > > > >> FORWARD
> > > > >> TSN chunk would need to traverse outbound stream structures with
> > > > >> dropped
> > > > >> TSNs to formulate Forward TSNs, but this is likely to be a far shorter
> > > > >> list than at the receiver and non-PR streams would never need to be
> > > > >> involved.
> > > > >
> > > > > And it must continue to KEEP this information until it is sure the
> > > > > peer has received and moved the cum-ack forward..
> > > > >
> > > > > What I end up doing on the sender side is I hold just the bookkeeping
> > > > > structures of the TSN (stream seq/stream no/tsn ... etc) but I
> > > > > drop the mbuf holding the data so I gain a bit of help in mbuf
> > > > > utilization.
> > > > >
> > > > > But I can't get rid of the chunk structure UNTIL the cum-ack says he
> > > > > got my FWD-TSN and accepted it...
> > > > >
> > > > > So there is NO additional info to save on the sender side and only
> > > > > do I save what I must... on the receiver side I have a LOT of extra
> > > > > bookkeeping....
> > > > >
> > > > >
> > > > > In fact with your confirmation I would claim just the opposite of
> > > > > Phil's statement... that doing it the way it is NOW documented takes
> > > > > MORE overhead.. only by adding the SSN/SID can you reduce this..
> > > > >
> > > > > By the way Renee also pointed out the extra overhead for the IOS
> > > > > implementation as well :>... so this makes three different
> > > > > implementations
> > > > > NOT 2....
> > > > >
> > > > > I plan on adding this to the next pr-sctp unless Michael Tuexen
> > > > > violently
> > > > > agrees with Qiaobing ...
> > > > >
> > > > > I still await Michaels answer????
> > > > >
> > > > >
> > > > >>
> > > > >> --brian
> > > > >>
> > > > >>>>
> > > > >>>> (B) alternative design: more complex, may require larger Forward TSN
> > > > >>>> chunks, but avoids re-introducing an HOL blocking problem, and thus
> > > > >>>> improves performance in some situations.
> > > > >>>
> > > > >>>
> > > > >>> see above it is NOT more complex...
> > > > >>>
> > > > >>>>
> > > > >>>> The question of whether to choose A or B depends on whether you
> > > > >>>> believe
> > > > >>>> that the situation Brian describes is a "corner case" or not.   So
> > > > >>>> I've
> > > > >>>> walked through Brian's example, and have determined that I don't
> > > > >>>> think it
> > > > >>>> is a corner case at all.   So, I'm inclined to agree with Randy and
> > > > >>>> Brian
> > > > >>>> that we should consider a change.
> > > > >>>>
> > > > >>>> But I'd like to invite others to walk through my line of reasoning
> > > > >>>> to test
> > > > >>>> whether I've reasoned this correctly.
> > > > >>>
> > > > >>>
> > > > >>> I agree it is NOT a corner case..
> > > > >>>
> > > > >>>>
> > > > >>>> My rationale that Brian's example is NOT a corner case:
> > > > >>>> ========================================
> > > > >>>>
> > > > >>>> In Brian's example, messages are being sent round-robin on a number
> > > > >>>> of
> > > > >>>> streams.  This seems like a fairly reasonable use of SCTP streams,
> > > > >>>> do that
> > > > >>>> aspect does not strike me as unusual.
> > > > >>>
> > > > >>> I agree :>
> > > > >>>
> > > > >>>>
> > > > >>>> Now for the next part of this explanation, I'm going to have to infer
> > > > >>>> certain things, so, Brian, please correct me if I infer wrongly.
> > > > >>>> *One*
> > > > >>>> possibility for the sequence of events leading to the
> > > > >>>> situation Brian describes is that there is a burst loss of data
> > > > >>>> chunks
> > > > >>>> TSN110 through TSN119 (ten chunks, but possibly less than ten
> > > > >>>> packets..
> > > > >>>> perhaps even only ONE packet if the chunks are small enough to be
> > > > >>>> bundled!)... and then data chunks TSN120 through TSN128 make it
> > > > >>>> through
> > > > >>>> OK.   This does not strike me as particularly unusual; a loss of a
> > > > >>>> single
> > > > >>>> packet followed by a successful packet is sufficient to cause this
> > > > >>>> situation.
> > > > >>>>
> > > > >>>
> > > > >>> Or the other possiblility would be a set of bundled chunks in one
> > > > >>> packet
> > > > >>> being lost.. either way is the same result I would think..
> > > > >>>
> > > > >>>> The next thing that happens is that TSNs 110 through 118 are
> > > > >>>> "dropped" in
> > > > >>>> the technical sense of that word; see the defn in the current
> > > > >>>> draft.  This
> > > > >>>> means that the sender gives up on sending them, and issues a forward
> > > > >>>> TSN.   However, no Forward TSN is issued for TSN119.  This may be
> > > > >>>> because
> > > > >>>> the messages on stream 9 are treated as Reliable.   This is NOT an
> > > > >>>> unusual
> > > > >>>> situation; it is perfectly reasonably to assume that an application
> > > > >>>> might
> > > > >>>> have "k" streams with data that can be dropped, and "n - k" streams
> > > > >>>> of data
> > > > >>>> that must be treated as reliable.  In the case of Brian's example, n
> > > > >>>> is 10,
> > > > >>>> and k is 9, but I think that his example will still produce similar
> > > > >>>> results
> > > > >>>> even for other values of "k" and "n".
> > > > >>>>
> > > > >>>
> > > > >>> Note.. streams themselves are not reliable or partially reliable. With
> > > > >>> the new draft it is the message itself which is sent with the
> > > > >>> reliablilty
> > > > >>> set to "fully reliable". One possilbe implementation would be to make
> > > > >>> things correlate to a particular stream.. but this is just an
> > > > >>> implementation
> > > > >>> option..
> > > > >>>
> > > > >>>> Bottom line:
> > > > >>>> ========
> > > > >>>>
> > > > >>>> I don't think its a corner case at all.   We have all the PR streams
> > > > >>>> potentially being HOL blocked by loss of a single packet.  I think
> > > > >>>> we need
> > > > >>>> to fix this.
> > > > >>>
> > > > >>> I agree..
> > > > >>>
> > > > >>>>
> > > > >>>> The solution of forward SSNs is one viable solution I think.  There
> > > > >>>> may be
> > > > >>>> others, but that's as good as any for right now I think.   There may
> > > > >>>> be
> > > > >>>> another solution as well, but let me do some writing and thinking
> > > > >>>> about it
> > > > >>>> off line before I muddy the waters with it here... just to make sure
> > > > >>>> it is
> > > > >>>> a viable alternative.
> > > > >>>
> > > > >>> agreed
> > > > >>>
> > > > >>>
> > > > >>>>
> > > > >>>> Randy... I'm willing to make the text changes and send a proposed
> > > > >>>> version
> > > > >>>> around to the co-authors for comment if there is consensus to move
> > > > >>>> in that
> > > > >>>> direction.   Let me know.
> > > > >>>>
> > > > >>>
> > > > >>> I will get it out this weekend if I may.. there are some interesting
> > > > >>> things that I have thought about that must be covered... I had
> > > > >>> something
> > > > >>> like this in an old u-sctp draft (during Q and I's debate earlier :>)
> > > > >>>
> > > > >>>
> > > > >>>
> > > > >>>>
> > > > >>>> (Side note: I was worried that the "technical definition" of the word
> > > > >>>> "dropped" would cause confusion, and it already has.   Perhaps
> > > > >>>> should use
> > > > >>>> another word in the draft to describe the operation where the sender
> > > > >>>> decides to no longer transmit and/or retransmit a specific data
> > > > >>>> chunk.  How
> > > > >>>> about the term "abandoned" in place of "dropped"?)
> > > > >>>
> > > > >>>
> > > > >>> Sounds good..
> > > > >>>
> > > > >>> note, Michael R also agrees .. or seems to .. we will talk more..
> > > > >>>
> > > > >>> R
> > > > >>>
> > > > >>>
> > > > >>>>
> > > > >>>> /ptc
> > > > >>>>
> > > > >>>> At 11:44 AM 6/5/2002 -0500, Randall Stewart wrote:
> > > > >>>>> Brian/Qiaobing:
> > > > >>>>>
> > > > >>>>> I have read through this interesting debate. Qiaobing and
> > > > >>>>> I had a major discussion on this way back when it was
> > > > >>>>> considered U-SCTP. Renee Revis first pointed this out to me.. but
> > > > >>>>> from a different persepective i.e. the receiver being able to
> > > > >>>>> efficently process stuck TSN's ...
> > > > >>>>>
> > > > >>>>> After a lengthy debate I agreed to not have the SSN/SID listed
> > > > >>>>> in the FWD TSN... I did NOT think of the HOL blocking issue.. or
> > > > >>>>> I would have used it as a issue (I was the one arguing for the
> > > > >>>>> SSN/SID in the FWD-TSN :->).
> > > > >>>>>
> > > > >>>>> After watching this debate.. I think I have come to the conclusion
> > > > >>>>> that
> > > > >>>>> Brian is correct :)
> > > > >>>>>
> > > > >>>>> Michael R. is here... I will discuss it with him.. I want to get
> > > > >>>>> the other co-author's take on this one... If the others agree with
> > > > >>>>> Brian (and me :>) then I think we will put this in.... also I would
> > > > >>>>> like to hear what the rest of the WG feels on this???
> > > > >>>>>
> > > > >>>>>
> > > > >>>>> I think it is reasonable and not that much overhead to include JUST
> > > > >>>>> the dropped streams NEXT expected SSN... You would be adding 4 bytes
> > > > >>>>> for EVERY stream dropping one or more TSN's... note that this does
> > > > >>>>> NOT mean 4 bytes per TSN but 4 bytes per where one or more TSN's
> > > > >>>>> were
> > > > >>>>> dropped..
> > > > >>>>>
> > > > >>>>>
> > > > >>>>> I think this is reasonable.. it improves processing AND avoids HOL
> > > > >>>>> blocking..
> > > > >>>>>
> > > > >>>>>
> > > > >>>>> Michael T/Phil? what do you think... Michael R.. I will talk to you
> > > > >>>>> at
> > > > >>>>> the
> > > > >>>>> next class break..
> > > > >>>>>
> > > > >>>>>
> > > > >>>>>
> > > > >>>>> R
> > > > >>>>>
> > > > >>>>> --
> > > > >>>>> 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)
> > > > >>
> > > > >> --
> > > > >> Brian F. G. Bidulock    ¦ The reasonable man adapts himself to the ¦
> > > > >> bidulock@openss7.org    ¦ world; the unreasonable one persists in  ¦
> > > > >> http://www.openss7.org/ ¦ trying  to adapt the  world  to himself. ¦
> > > > >>                         ¦ Therefore  all  progress  depends on the ¦
> > > > >>                         ¦ unreasonable man. -- George Bernard Shaw ¦
> > > > >
> > > > > --
> > > > > 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
> > > > >
> > > > >
> > > > Michael.Tuexen@micmac.franken.de
> > 
> > --
> > Brian F. G. Bidulock    ¦ The reasonable man adapts himself to the ¦
> > bidulock@openss7.org    ¦ world; the unreasonable one persists in  ¦
> > http://www.openss7.org/ ¦ trying  to adapt the  world  to himself. ¦
> >                         ¦ Therefore  all  progress  depends on the ¦
> >                         ¦ unreasonable man. -- George Bernard Shaw ¦

-- 
Brian F. G. Bidulock    ¦ The reasonable man adapts himself to the ¦
bidulock@openss7.org    ¦ world; the unreasonable one persists in  ¦
http://www.openss7.org/ ¦ trying  to adapt the  world  to himself. ¦
                        ¦ Therefore  all  progress  depends on the ¦
                        ¦ unreasonable man. -- George Bernard Shaw ¦

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



From daemon@optimus.ietf.org  Fri Jun  7 21:51:27 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 VAA19771
	for <tsvwg-archive@odin.ietf.org>; Fri, 7 Jun 2002 21:51:26 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id VAA21168
	for tsvwg-archive@odin.ietf.org; Fri, 7 Jun 2002 21:51:57 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id VAA20076;
	Fri, 7 Jun 2002 21:24:49 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id UAA18067
	for <tsvwg@optimus.ietf.org>; Fri, 7 Jun 2002 20:36:41 -0400 (EDT)
Received: from motgate.mot.com (motgate.mot.com [129.188.136.100])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id UAA18007
	for <tsvwg@ietf.org>; Fri, 7 Jun 2002 20:36:10 -0400 (EDT)
Received: [from pobox.mot.com (pobox.mot.com [129.188.137.100]) by motgate.mot.com (motgate 2.1) with ESMTP id RAA17190; Fri, 7 Jun 2002 17:36:38 -0700 (MST)]
Received: [from relay1.cig.mot.com (relay1.cig.mot.com [136.182.15.23]) by pobox.mot.com (MOT-pobox 2.0) with ESMTP id RAA21586; Fri, 7 Jun 2002 17:36:38 -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 g580aYb03488;
	Fri, 7 Jun 2002 19:36:34 -0500 (CDT)
Received: from motorola.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 TAA01515;
	Fri, 7 Jun 2002 19:36:56 -0500 (CDT)
Message-ID: <3D015235.C2767BA1@motorola.com>
Date: Fri, 07 Jun 2002 19:39:17 -0500
From: Qiaobing Xie <Qiaobing.Xie@motorola.com>
X-Mailer: Mozilla 4.77 [en] (Win98; U)
X-Accept-Language: en
MIME-Version: 1.0
To: bidulock@openss7.org
CC: Michael Tuexen <Michael.Tuexen@lurchi.franken.de>,
        Randall Stewart <randall@stewart.chicago.il.us>,
        Phillip Conrad <conrad@acm.org>,
        "Armando L. Caro Jr." <acaro@mail.eecis.udel.edu>, tsvwg@ietf.org,
        "Michael A. Ramalho" <mramalho@cisco.com>
Subject: Re: [Tsvwg] More on PRSCTP-00
References: <D4966AFE-7A3E-11D6-9E0D-0030654C1AB6@lurchi.franken.de> <3D010E5E.D51114EF@motorola.com> <20020607153911.D2907@openss7.org>
Content-Type: text/plain; charset=iso-8859-1
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

"Brian F. G. Bidulock" wrote:
> 
> Michael, Qiaobing,
> 
> Here is the sender side of the example.  It is quite reasonable
> that this could occur.  The receiver is blocked on streams 1
> through 9 until retransmission of TSN 119 is successful.  This
> is the sender side of the same example.
> 
>     TSN 101  SID 1   SSN 1   |
>     TSN 102  SID 2   SSN 1   |
>     TSN 103  SID 3   SSN 1   |
>     TSN 104  SID 4   SSN 1   |   SCTP Packet
>     TSN 105  SID 5   SSN 1   +--------------------->
>     TSN 106  SID 6   SSN 1   |
>     TSN 107  SID 7   SSN 1   |
>     TSN 108  SID 8   SSN 1   |
>     TSN 109  SID 9   SSN 1   |
> 
>     TSN 110  SID 1   SSN 2   |
>     TSN 111  SID 2   SSN 2   |
>     TSN 112  SID 3   SSN 2   |
>     TSN 113  SID 4   SSN 2   |   SCTP Packet
>     TSN 114  SID 5   SSN 2   +----------X
>     TSN 115  SID 6   SSN 2   |         Lost
>     TSN 116  SID 7   SSN 2   |
>     TSN 117  SID 8   SSN 2   |
>     TSN 118  SID 9   SSN 2   |
>                                  SCTP Packet
>     TSN 119  SID 0   SSN X   |----------X
>                                        Lost
> 
>     (Timed reliability expiry on TSNs 110 to 118)
> 
>     FORWARD TSN 118          |--------------------->

Why TSN119 is not re-sent at this exactly same moment? As Michael and I
both noted that, in PR-SCTP, Fwd TSN can only be triggered due to either
a "RTX Timer expiration or fast retransmission." Could you fill in the
details that lead to TSN119 not being re-sent at this moment?

> 
>     TSN 120  SID 1   SSN 3   |
>     TSN 121  SID 2   SSN 3   |
>     TSN 122  SID 3   SSN 3   |
>     TSN 123  SID 4   SSN 3   |   SCTP Packet
>     TSN 124  SID 5   SSN 3   +--------------------->
>     TSN 125  SID 6   SSN 3   |
>     TSN 126  SID 7   SSN 3   |
>     TSN 127  SID 8   SSN 3   |
>     TSN 128  SID 9   SSN 3   |
> 
>     (Receiver is now blocked on streams 1 through 9.)
> 
>     (Some time later, retrans timeout on TSN 119)
> 
>                                  SCTP Packet
>     TSN 119  SID 0   SSN X   |--------------------->
> 
> Q.E.D.
> 
> --brian
> 
> On Fri, 07 Jun 2002, Qiaobing Xie wrote:
> 
> > Michael,
> >
> > Right on! You summarized my previous point beautifully. I start to
> > realize that lately my arguments tend to become lengthy and off the
> > point... maybe a sign of getting old :<
> >
> > So far I kind of buy Randy's point better that "adding some bits in
> > exchange for ease of implementation", but still yet to be convinced
> > about the HOL blocking argument made by Brian.
> >
> > I've asked for an "all-considered" example of the HOL blocking, but
> > haven't seen one yet.
> >
> > regards,
> > -Qiaobing
> >
> > Michael Tuexen wrote:
> > >
> > > Dear all,
> > >
> > > reading the thread I would like to make some remarks:
> > >
> > > 1. I agree that using SSN in the FWD chunk avoids HOL (which is good).
> > > 2. Brians example show how HOL can occur if no SSN info in the FWD chunk
> > >     is used. He only looks at the receiver.
> > >
> > > Brians example are always based on the assumption that there is a GAP
> > > in the TSN space at the receiver side and the receiver gets a FWD TSN
> > > which does not fill the whole gap. Then the HOL happens at the receiver.
> > >
> > > I'm just wondering how often this will happen. What are the reasons that
> > > a FWD chunk will be send: RTX Timer expiration or fast retransmission.
> > > In both cases I would send a FWD chunk and bundle the remaining DATA
> > > chunks
> > > (if possible). So there is a good chance that the whole gap is filled by
> > > the FWD chunk and the retransmitted DATA chunks.
> > > I'm not sure how likely Brians scenario is taking the sender into
> > > account.
> > >
> > > So I think one has to take the senders behavior into account when
> > > discussing
> > > this issue.
> > >
> > > Best regards
> > > Michael
> > >
> > > On Thursday, June 6, 2002, at 01:43 AM, Randall Stewart wrote:
> > >
> > > > "Brian F. G. Bidulock" wrote:
> > > >>
> > > >> Randall,
> > > >>
> > > >> On Wed, 05 Jun 2002, Randall Stewart wrote:
> > > >>
> > > >>> A few comments..
> > > >>>
> > > >>> Phillip Conrad wrote:
> > > >>>>
> > > >>>> Dear Colleagues:
> > > >>>>       I agree with Randy that Brian has raised an important issue
> > > >>>> for us to
> > > >>>> consider, and I think has explained it quite well with a very clear,
> > > >>>> illustrative example.
> > > >>>>
> > > >>>> The question is one of a tradeoff between two designs:
> > > >>>>
> > > >>>> (A) current design: simpler, fewer bits on the wire, but may
> > > >>>> (re)-introduce
> > > >>>> an HOL blocking problem
> > > >>>
> > > >>> I must disagree with your "it is simpler" ... it is NOT. The question
> > > >>> is
> > > >>> where is the burden. Having implemented pr-sctp in the kernel of
> > > >>> freebsd
> > > >>> NOT having the stream number/SSN in the FWD-TSN means I must keep
> > > >>> an extra set of linked lists and such to track any stream that has
> > > >>> pending stuck data. When a fwd-tsn arrives I must HUNT all of these
> > > >>> streams and look at each inbound queue.
> > > >>>
> > > >>> The counter to this is the sender (in Brian and my earlier proposal
> > > >>> that was to Q only) must track this info to send with the FWD-TSN.
> > > >>> Actually
> > > >>> less of a burden IMHO...
> > > >>
> > > >> I agree.  I discovered this situation while implementing PR-SCTP in the
> > > >> Linux Kernel.  The normal receiver can rely on the fact that the next
> > > >> expected SSN is the last fully received SSN plus 1.  PR-SCTP without
> > > >> a forward SSN causes the receive to have to update this expected value
> > > >> with the next SSN in the reorder queue whenever there are no TSN gaps.
> > > >> This even might require that when TSN gaps are filled that the receiver
> > > >> walk all inbound stream structures to update the next expected SSN.
> > > >> In an independent implementation, I have arrived at the same
> > > >> conclusion:
> > > >> that the reciever has more work to do for FORWARD TSN without a forward
> > > >> SSN.
> > > >>
> > > >
> > > > Yeah.. I basically had to make a special linked list set that the
> > > > stream queues could be placed on.. a bit more efficient but a
> > > > reall hassel to maintian a "wheel" (as I call it) of streams with
> > > > pending data on it...
> > > >
> > > >> In fact the receiver has this work to do if it has ever received a
> > > >> FORWARD TSN.  This is because it is not possible for the receiver to
> > > >> know a priori which streams could be affected by this scenairo.
> > > >> Therefore,
> > > >> the load on the receiver is increased even when this case does not
> > > >> occur.
> > > >> It is the mere possibility of its occurence which introduces the load.
> > > >> (Each time gaps close, the next expected SSN for all streams need to be
> > > >> re-examined whether there are SSN gaps or not.)
> > > >
> > > > Exactly.. I constantly must place stream q;s on and off the :"wheel" no
> > > > matter if a fwd-tsn is EVER received...
> > > >
> > > >>
> > > >> On the sender side, the sender must always track the last sent SSN and
> > > >> assign the next SSN (++SSN).  Indeed the sender when generating a
> > > >> FORWARD
> > > >> TSN chunk would need to traverse outbound stream structures with
> > > >> dropped
> > > >> TSNs to formulate Forward TSNs, but this is likely to be a far shorter
> > > >> list than at the receiver and non-PR streams would never need to be
> > > >> involved.
> > > >
> > > > And it must continue to KEEP this information until it is sure the
> > > > peer has received and moved the cum-ack forward..
> > > >
> > > > What I end up doing on the sender side is I hold just the bookkeeping
> > > > structures of the TSN (stream seq/stream no/tsn ... etc) but I
> > > > drop the mbuf holding the data so I gain a bit of help in mbuf
> > > > utilization.
> > > >
> > > > But I can't get rid of the chunk structure UNTIL the cum-ack says he
> > > > got my FWD-TSN and accepted it...
> > > >
> > > > So there is NO additional info to save on the sender side and only
> > > > do I save what I must... on the receiver side I have a LOT of extra
> > > > bookkeeping....
> > > >
> > > >
> > > > In fact with your confirmation I would claim just the opposite of
> > > > Phil's statement... that doing it the way it is NOW documented takes
> > > > MORE overhead.. only by adding the SSN/SID can you reduce this..
> > > >
> > > > By the way Renee also pointed out the extra overhead for the IOS
> > > > implementation as well :>... so this makes three different
> > > > implementations
> > > > NOT 2....
> > > >
> > > > I plan on adding this to the next pr-sctp unless Michael Tuexen
> > > > violently
> > > > agrees with Qiaobing ...
> > > >
> > > > I still await Michaels answer????
> > > >
> > > >
> > > >>
> > > >> --brian
> > > >>
> > > >>>>
> > > >>>> (B) alternative design: more complex, may require larger Forward TSN
> > > >>>> chunks, but avoids re-introducing an HOL blocking problem, and thus
> > > >>>> improves performance in some situations.
> > > >>>
> > > >>>
> > > >>> see above it is NOT more complex...
> > > >>>
> > > >>>>
> > > >>>> The question of whether to choose A or B depends on whether you
> > > >>>> believe
> > > >>>> that the situation Brian describes is a "corner case" or not.   So
> > > >>>> I've
> > > >>>> walked through Brian's example, and have determined that I don't
> > > >>>> think it
> > > >>>> is a corner case at all.   So, I'm inclined to agree with Randy and
> > > >>>> Brian
> > > >>>> that we should consider a change.
> > > >>>>
> > > >>>> But I'd like to invite others to walk through my line of reasoning
> > > >>>> to test
> > > >>>> whether I've reasoned this correctly.
> > > >>>
> > > >>>
> > > >>> I agree it is NOT a corner case..
> > > >>>
> > > >>>>
> > > >>>> My rationale that Brian's example is NOT a corner case:
> > > >>>> ========================================
> > > >>>>
> > > >>>> In Brian's example, messages are being sent round-robin on a number
> > > >>>> of
> > > >>>> streams.  This seems like a fairly reasonable use of SCTP streams,
> > > >>>> do that
> > > >>>> aspect does not strike me as unusual.
> > > >>>
> > > >>> I agree :>
> > > >>>
> > > >>>>
> > > >>>> Now for the next part of this explanation, I'm going to have to infer
> > > >>>> certain things, so, Brian, please correct me if I infer wrongly.
> > > >>>> *One*
> > > >>>> possibility for the sequence of events leading to the
> > > >>>> situation Brian describes is that there is a burst loss of data
> > > >>>> chunks
> > > >>>> TSN110 through TSN119 (ten chunks, but possibly less than ten
> > > >>>> packets..
> > > >>>> perhaps even only ONE packet if the chunks are small enough to be
> > > >>>> bundled!)... and then data chunks TSN120 through TSN128 make it
> > > >>>> through
> > > >>>> OK.   This does not strike me as particularly unusual; a loss of a
> > > >>>> single
> > > >>>> packet followed by a successful packet is sufficient to cause this
> > > >>>> situation.
> > > >>>>
> > > >>>
> > > >>> Or the other possiblility would be a set of bundled chunks in one
> > > >>> packet
> > > >>> being lost.. either way is the same result I would think..
> > > >>>
> > > >>>> The next thing that happens is that TSNs 110 through 118 are
> > > >>>> "dropped" in
> > > >>>> the technical sense of that word; see the defn in the current
> > > >>>> draft.  This
> > > >>>> means that the sender gives up on sending them, and issues a forward
> > > >>>> TSN.   However, no Forward TSN is issued for TSN119.  This may be
> > > >>>> because
> > > >>>> the messages on stream 9 are treated as Reliable.   This is NOT an
> > > >>>> unusual
> > > >>>> situation; it is perfectly reasonably to assume that an application
> > > >>>> might
> > > >>>> have "k" streams with data that can be dropped, and "n - k" streams
> > > >>>> of data
> > > >>>> that must be treated as reliable.  In the case of Brian's example, n
> > > >>>> is 10,
> > > >>>> and k is 9, but I think that his example will still produce similar
> > > >>>> results
> > > >>>> even for other values of "k" and "n".
> > > >>>>
> > > >>>
> > > >>> Note.. streams themselves are not reliable or partially reliable. With
> > > >>> the new draft it is the message itself which is sent with the
> > > >>> reliablilty
> > > >>> set to "fully reliable". One possilbe implementation would be to make
> > > >>> things correlate to a particular stream.. but this is just an
> > > >>> implementation
> > > >>> option..
> > > >>>
> > > >>>> Bottom line:
> > > >>>> ========
> > > >>>>
> > > >>>> I don't think its a corner case at all.   We have all the PR streams
> > > >>>> potentially being HOL blocked by loss of a single packet.  I think
> > > >>>> we need
> > > >>>> to fix this.
> > > >>>
> > > >>> I agree..
> > > >>>
> > > >>>>
> > > >>>> The solution of forward SSNs is one viable solution I think.  There
> > > >>>> may be
> > > >>>> others, but that's as good as any for right now I think.   There may
> > > >>>> be
> > > >>>> another solution as well, but let me do some writing and thinking
> > > >>>> about it
> > > >>>> off line before I muddy the waters with it here... just to make sure
> > > >>>> it is
> > > >>>> a viable alternative.
> > > >>>
> > > >>> agreed
> > > >>>
> > > >>>
> > > >>>>
> > > >>>> Randy... I'm willing to make the text changes and send a proposed
> > > >>>> version
> > > >>>> around to the co-authors for comment if there is consensus to move
> > > >>>> in that
> > > >>>> direction.   Let me know.
> > > >>>>
> > > >>>
> > > >>> I will get it out this weekend if I may.. there are some interesting
> > > >>> things that I have thought about that must be covered... I had
> > > >>> something
> > > >>> like this in an old u-sctp draft (during Q and I's debate earlier :>)
> > > >>>
> > > >>>
> > > >>>
> > > >>>>
> > > >>>> (Side note: I was worried that the "technical definition" of the word
> > > >>>> "dropped" would cause confusion, and it already has.   Perhaps
> > > >>>> should use
> > > >>>> another word in the draft to describe the operation where the sender
> > > >>>> decides to no longer transmit and/or retransmit a specific data
> > > >>>> chunk.  How
> > > >>>> about the term "abandoned" in place of "dropped"?)
> > > >>>
> > > >>>
> > > >>> Sounds good..
> > > >>>
> > > >>> note, Michael R also agrees .. or seems to .. we will talk more..
> > > >>>
> > > >>> R
> > > >>>
> > > >>>
> > > >>>>
> > > >>>> /ptc
> > > >>>>
> > > >>>> At 11:44 AM 6/5/2002 -0500, Randall Stewart wrote:
> > > >>>>> Brian/Qiaobing:
> > > >>>>>
> > > >>>>> I have read through this interesting debate. Qiaobing and
> > > >>>>> I had a major discussion on this way back when it was
> > > >>>>> considered U-SCTP. Renee Revis first pointed this out to me.. but
> > > >>>>> from a different persepective i.e. the receiver being able to
> > > >>>>> efficently process stuck TSN's ...
> > > >>>>>
> > > >>>>> After a lengthy debate I agreed to not have the SSN/SID listed
> > > >>>>> in the FWD TSN... I did NOT think of the HOL blocking issue.. or
> > > >>>>> I would have used it as a issue (I was the one arguing for the
> > > >>>>> SSN/SID in the FWD-TSN :->).
> > > >>>>>
> > > >>>>> After watching this debate.. I think I have come to the conclusion
> > > >>>>> that
> > > >>>>> Brian is correct :)
> > > >>>>>
> > > >>>>> Michael R. is here... I will discuss it with him.. I want to get
> > > >>>>> the other co-author's take on this one... If the others agree with
> > > >>>>> Brian (and me :>) then I think we will put this in.... also I would
> > > >>>>> like to hear what the rest of the WG feels on this???
> > > >>>>>
> > > >>>>>
> > > >>>>> I think it is reasonable and not that much overhead to include JUST
> > > >>>>> the dropped streams NEXT expected SSN... You would be adding 4 bytes
> > > >>>>> for EVERY stream dropping one or more TSN's... note that this does
> > > >>>>> NOT mean 4 bytes per TSN but 4 bytes per where one or more TSN's
> > > >>>>> were
> > > >>>>> dropped..
> > > >>>>>
> > > >>>>>
> > > >>>>> I think this is reasonable.. it improves processing AND avoids HOL
> > > >>>>> blocking..
> > > >>>>>
> > > >>>>>
> > > >>>>> Michael T/Phil? what do you think... Michael R.. I will talk to you
> > > >>>>> at
> > > >>>>> the
> > > >>>>> next class break..
> > > >>>>>
> > > >>>>>
> > > >>>>>
> > > >>>>> R
> > > >>>>>
> > > >>>>> --
> > > >>>>> 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)
> > > >>
> > > >> --
> > > >> Brian F. G. Bidulock    ¦ The reasonable man adapts himself to the ¦
> > > >> bidulock@openss7.org    ¦ world; the unreasonable one persists in  ¦
> > > >> http://www.openss7.org/ ¦ trying  to adapt the  world  to himself. ¦
> > > >>                         ¦ Therefore  all  progress  depends on the ¦
> > > >>                         ¦ unreasonable man. -- George Bernard Shaw ¦
> > > >
> > > > --
> > > > 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
> > > >
> > > >
> > > Michael.Tuexen@micmac.franken.de
> 
> --
> Brian F. G. Bidulock    ¦ The reasonable man adapts himself to the ¦
> bidulock@openss7.org    ¦ world; the unreasonable one persists in  ¦
> http://www.openss7.org/ ¦ trying  to adapt the  world  to himself. ¦
>                         ¦ Therefore  all  progress  depends on the ¦
>                         ¦ unreasonable man. -- George Bernard Shaw ¦


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



From daemon@optimus.ietf.org  Fri Jun  7 21:54:47 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 VAA19829
	for <tsvwg-archive@odin.ietf.org>; Fri, 7 Jun 2002 21:54:47 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id VAA21223
	for tsvwg-archive@odin.ietf.org; Fri, 7 Jun 2002 21:55:18 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id VAA20048;
	Fri, 7 Jun 2002 21:24:48 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id UAA18142
	for <tsvwg@optimus.ietf.org>; Fri, 7 Jun 2002 20:38:48 -0400 (EDT)
Received: from motgate4.mot.com (motgate4.mot.com [144.189.100.102])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id UAA18034
	for <tsvwg@ietf.org>; Fri, 7 Jun 2002 20:38:17 -0400 (EDT)
Received: [from mothost.mot.com (mothost.mot.com [129.188.137.101]) by motgate4.mot.com (motgate4 2.1) with ESMTP id RAA10162; Fri, 7 Jun 2002 17:38:47 -0700 (MST)]
Received: [from relay1.cig.mot.com (relay1.cig.mot.com [136.182.15.23]) by mothost.mot.com (MOT-pobox 2.0) with ESMTP id RAA16758; Fri, 7 Jun 2002 17:38:40 -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 g580cib03590;
	Fri, 7 Jun 2002 19:38:44 -0500 (CDT)
Received: from motorola.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 TAA01545;
	Fri, 7 Jun 2002 19:39:06 -0500 (CDT)
Message-ID: <3D0152B8.B88CD76C@motorola.com>
Date: Fri, 07 Jun 2002 19:41:28 -0500
From: Qiaobing Xie <Qiaobing.Xie@motorola.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: bidulock@openss7.org, "Armando L. Caro Jr." <acaro@mail.eecis.udel.edu>,
        tsvwg@ietf.org
Subject: Re: [Tsvwg] More on PRSCTP-00
References: <20020531195333.D11468@openss7.org> <3CFBC11C.1EFA747@email.mot.com> <20020603144342.A30667@openss7.org> <3CFC144D.65697A41@email.mot.com> <20020603215952.E30667@openss7.org> <3CFD57D8.5637E431@email.mot.com> <20020604195020.A16020@openss7.org> <3CFE7403.59FB6F4C@email.mot.com> <20020605163502.A2012@openss7.org> <3CFE8C3F.C935774C@email.mot.com> <20020605171556.C2012@openss7.org> <3CFE9238.930A5B15@email.mot.com> <3D00E134.D5981833@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

...
> Q: When I get back from my travels I will see if I can't make ammends
>    to you on my "change" by offering you a glass of cab at my house :>

This is a nice offer and I will take it :-)

-Q

> 
> 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@optimus.ietf.org  Sat Jun  8 08:12:50 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 IAA05496
	for <tsvwg-archive@odin.ietf.org>; Sat, 8 Jun 2002 08:12:50 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id IAA21445
	for tsvwg-archive@odin.ietf.org; Sat, 8 Jun 2002 08:13:22 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id HAA20132;
	Sat, 8 Jun 2002 07:49:06 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id HAA20060
	for <tsvwg@optimus.ietf.org>; Sat, 8 Jun 2002 07:49:01 -0400 (EDT)
Received: from ilsa.franken.de (postfix@ilsa.franken.de [193.175.24.42])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id HAA05169
	for <tsvwg@ietf.org>; Sat, 8 Jun 2002 07:48:27 -0400 (EDT)
Received: from localhost (pD9E4F7CF.dip.t-dialin.net [217.228.247.207])
	by ilsa.franken.de (Postfix) with ESMTP
	id 7F38B245D4; Sat,  8 Jun 2002 13:48:54 +0200 (CEST)
Date: Sat, 8 Jun 2002 13:48:38 +0200
Subject: Re: [Tsvwg] More on PRSCTP-00
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Mime-Version: 1.0 (Apple Message framework v482)
Cc: Qiaobing Xie <Qiaobing.Xie@motorola.com>,
        Randall Stewart <randall@stewart.chicago.il.us>,
        Phillip Conrad <conrad@acm.org>,
        "Armando L. Caro Jr." <acaro@mail.eecis.udel.edu>, tsvwg@ietf.org,
        "Michael A. Ramalho" <mramalho@cisco.com>
To: bidulock@openss7.org
From: Michael Tuexen <Michael.Tuexen@lurchi.franken.de>
In-Reply-To: <20020607194954.B8970@openss7.org>
Message-Id: <AC64F48B-7AD5-11D6-AD39-0030654C1AB6@lurchi.franken.de>
X-Mailer: Apple Mail (2.482)
Content-Transfer-Encoding: 8bit
X-MIME-Autoconverted: from quoted-printable to 8bit by optimus.ietf.org id HAA20061
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

Brian,

see my questions below.

Best regards
Michael

On Saturday, June 8, 2002, at 02:49 AM, Brian F. G. Bidulock wrote:

> Qiaobing,
>
> On Fri, 07 Jun 2002, Qiaobing Xie wrote:
>
>> "Brian F. G. Bidulock" wrote:
>>>
>>> Michael, Qiaobing,
>>>
>>> Here is the sender side of the example.  It is quite reasonable
>>> that this could occur.  The receiver is blocked on streams 1
>>> through 9 until retransmission of TSN 119 is successful.  This
>>> is the sender side of the same example.
>>>
>>>     TSN 101  SID 1   SSN 1   |
>>>     TSN 102  SID 2   SSN 1   |
>>>     TSN 103  SID 3   SSN 1   |
>>>     TSN 104  SID 4   SSN 1   |   SCTP Packet
>>>     TSN 105  SID 5   SSN 1   +--------------------->
>>>     TSN 106  SID 6   SSN 1   |
>>>     TSN 107  SID 7   SSN 1   |
>>>     TSN 108  SID 8   SSN 1   |
>>>     TSN 109  SID 9   SSN 1   |
>
>         <---- SACK CUM ACK 109
>
>>>
>>>     TSN 110  SID 1   SSN 2   |
>>>     TSN 111  SID 2   SSN 2   |
>>>     TSN 112  SID 3   SSN 2   |
>>>     TSN 113  SID 4   SSN 2   |   SCTP Packet
>>>     TSN 114  SID 5   SSN 2   +----------X
>>>     TSN 115  SID 6   SSN 2   |         Lost
>>>     TSN 116  SID 7   SSN 2   |
>>>     TSN 117  SID 8   SSN 2   |
>>>     TSN 118  SID 9   SSN 2   |
>
>         <---- SACK CUM ACK 109
Why is this SACK be send? There is no event at the receiver since the 
packet is lost on the wire.
>
>>>                                  SCTP Packet
>>>     TSN 119  SID 0   SSN X   |----------X
>>>                                        Lost
>>>
>>>     (Timed reliability expiry on TSNs 110 to 118)
>
>         <---- SACK CUM ACK 109
Why is this SACK be send
>
>>>
>>>     FORWARD TSN 118          |--------------------->
This now means that the retransmission timer runs off. So bundle a 
FWD-TSN and DATA.
So I really do not understand this example.
>>
>> Why TSN119 is not re-sent at this exactly same moment? As Michael and I
>> both noted that, in PR-SCTP, Fwd TSN can only be triggered due to 
>> either
>> a "RTX Timer expiration or fast retransmission." Could you fill in the
>> details that lead to TSN119 not being re-sent at this moment?
>
> With the SACKs shown, the FORWARD TSN "MUST" be sent according to
> Rule C3.  Retransmission of TSN 119 has nothing to do with it.
>
> --brian
>
>>
>>>
>>>     TSN 120  SID 1   SSN 3   |
>>>     TSN 121  SID 2   SSN 3   |
>>>     TSN 122  SID 3   SSN 3   |
>>>     TSN 123  SID 4   SSN 3   |   SCTP Packet
>>>     TSN 124  SID 5   SSN 3   +--------------------->
>>>     TSN 125  SID 6   SSN 3   |
>>>     TSN 126  SID 7   SSN 3   |
>>>     TSN 127  SID 8   SSN 3   |
>>>     TSN 128  SID 9   SSN 3   |
>>>
>>>     (Receiver is now blocked on streams 1 through 9.)
>>>
>>>     (Some time later, retrans timeout on TSN 119)
>>>
>>>                                  SCTP Packet
>>>     TSN 119  SID 0   SSN X   |--------------------->
>>>
>>> Q.E.D.
>>>
>>> --brian
>>>
>>> On Fri, 07 Jun 2002, Qiaobing Xie wrote:
>>>
>>>> Michael,
>>>>
>>>> Right on! You summarized my previous point beautifully. I start to
>>>> realize that lately my arguments tend to become lengthy and off the
>>>> point... maybe a sign of getting old :<
>>>>
>>>> So far I kind of buy Randy's point better that "adding some bits in
>>>> exchange for ease of implementation", but still yet to be convinced
>>>> about the HOL blocking argument made by Brian.
>>>>
>>>> I've asked for an "all-considered" example of the HOL blocking, but
>>>> haven't seen one yet.
>>>>
>>>> regards,
>>>> -Qiaobing
>>>>
>>>> Michael Tuexen wrote:
>>>>>
>>>>> Dear all,
>>>>>
>>>>> reading the thread I would like to make some remarks:
>>>>>
>>>>> 1. I agree that using SSN in the FWD chunk avoids HOL (which is 
>>>>> good).
>>>>> 2. Brians example show how HOL can occur if no SSN info in the FWD 
>>>>> chunk
>>>>>     is used. He only looks at the receiver.
>>>>>
>>>>> Brians example are always based on the assumption that there is a 
>>>>> GAP
>>>>> in the TSN space at the receiver side and the receiver gets a FWD 
>>>>> TSN
>>>>> which does not fill the whole gap. Then the HOL happens at the 
>>>>> receiver.
>>>>>
>>>>> I'm just wondering how often this will happen. What are the reasons 
>>>>> that
>>>>> a FWD chunk will be send: RTX Timer expiration or fast 
>>>>> retransmission.
>>>>> In both cases I would send a FWD chunk and bundle the remaining DATA
>>>>> chunks
>>>>> (if possible). So there is a good chance that the whole gap is 
>>>>> filled by
>>>>> the FWD chunk and the retransmitted DATA chunks.
>>>>> I'm not sure how likely Brians scenario is taking the sender into
>>>>> account.
>>>>>
>>>>> So I think one has to take the senders behavior into account when
>>>>> discussing
>>>>> this issue.
>>>>>
>>>>> Best regards
>>>>> Michael
>>>>>
>>>>> On Thursday, June 6, 2002, at 01:43 AM, Randall Stewart wrote:
>>>>>
>>>>>> "Brian F. G. Bidulock" wrote:
>>>>>>>
>>>>>>> Randall,
>>>>>>>
>>>>>>> On Wed, 05 Jun 2002, Randall Stewart wrote:
>>>>>>>
>>>>>>>> A few comments..
>>>>>>>>
>>>>>>>> Phillip Conrad wrote:
>>>>>>>>>
>>>>>>>>> Dear Colleagues:
>>>>>>>>>       I agree with Randy that Brian has raised an important 
>>>>>>>>> issue
>>>>>>>>> for us to
>>>>>>>>> consider, and I think has explained it quite well with a very 
>>>>>>>>> clear,
>>>>>>>>> illustrative example.
>>>>>>>>>
>>>>>>>>> The question is one of a tradeoff between two designs:
>>>>>>>>>
>>>>>>>>> (A) current design: simpler, fewer bits on the wire, but may
>>>>>>>>> (re)-introduce
>>>>>>>>> an HOL blocking problem
>>>>>>>>
>>>>>>>> I must disagree with your "it is simpler" ... it is NOT. The 
>>>>>>>> question
>>>>>>>> is
>>>>>>>> where is the burden. Having implemented pr-sctp in the kernel of
>>>>>>>> freebsd
>>>>>>>> NOT having the stream number/SSN in the FWD-TSN means I must keep
>>>>>>>> an extra set of linked lists and such to track any stream that 
>>>>>>>> has
>>>>>>>> pending stuck data. When a fwd-tsn arrives I must HUNT all of 
>>>>>>>> these
>>>>>>>> streams and look at each inbound queue.
>>>>>>>>
>>>>>>>> The counter to this is the sender (in Brian and my earlier 
>>>>>>>> proposal
>>>>>>>> that was to Q only) must track this info to send with the 
>>>>>>>> FWD-TSN.
>>>>>>>> Actually
>>>>>>>> less of a burden IMHO...
>>>>>>>
>>>>>>> I agree.  I discovered this situation while implementing PR-SCTP 
>>>>>>> in the
>>>>>>> Linux Kernel.  The normal receiver can rely on the fact that the 
>>>>>>> next
>>>>>>> expected SSN is the last fully received SSN plus 1.  PR-SCTP 
>>>>>>> without
>>>>>>> a forward SSN causes the receive to have to update this expected 
>>>>>>> value
>>>>>>> with the next SSN in the reorder queue whenever there are no TSN 
>>>>>>> gaps.
>>>>>>> This even might require that when TSN gaps are filled that the 
>>>>>>> receiver
>>>>>>> walk all inbound stream structures to update the next expected 
>>>>>>> SSN.
>>>>>>> In an independent implementation, I have arrived at the same
>>>>>>> conclusion:
>>>>>>> that the reciever has more work to do for FORWARD TSN without a 
>>>>>>> forward
>>>>>>> SSN.
>>>>>>>
>>>>>>
>>>>>> Yeah.. I basically had to make a special linked list set that the
>>>>>> stream queues could be placed on.. a bit more efficient but a
>>>>>> reall hassel to maintian a "wheel" (as I call it) of streams with
>>>>>> pending data on it...
>>>>>>
>>>>>>> In fact the receiver has this work to do if it has ever received a
>>>>>>> FORWARD TSN.  This is because it is not possible for the receiver 
>>>>>>> to
>>>>>>> know a priori which streams could be affected by this scenairo.
>>>>>>> Therefore,
>>>>>>> the load on the receiver is increased even when this case does not
>>>>>>> occur.
>>>>>>> It is the mere possibility of its occurence which introduces the 
>>>>>>> load.
>>>>>>> (Each time gaps close, the next expected SSN for all streams need 
>>>>>>> to be
>>>>>>> re-examined whether there are SSN gaps or not.)
>>>>>>
>>>>>> Exactly.. I constantly must place stream q;s on and off the 
>>>>>> :"wheel" no
>>>>>> matter if a fwd-tsn is EVER received...
>>>>>>
>>>>>>>
>>>>>>> On the sender side, the sender must always track the last sent 
>>>>>>> SSN and
>>>>>>> assign the next SSN (++SSN).  Indeed the sender when generating a
>>>>>>> FORWARD
>>>>>>> TSN chunk would need to traverse outbound stream structures with
>>>>>>> dropped
>>>>>>> TSNs to formulate Forward TSNs, but this is likely to be a far 
>>>>>>> shorter
>>>>>>> list than at the receiver and non-PR streams would never need to 
>>>>>>> be
>>>>>>> involved.
>>>>>>
>>>>>> And it must continue to KEEP this information until it is sure the
>>>>>> peer has received and moved the cum-ack forward..
>>>>>>
>>>>>> What I end up doing on the sender side is I hold just the 
>>>>>> bookkeeping
>>>>>> structures of the TSN (stream seq/stream no/tsn ... etc) but I
>>>>>> drop the mbuf holding the data so I gain a bit of help in mbuf
>>>>>> utilization.
>>>>>>
>>>>>> But I can't get rid of the chunk structure UNTIL the cum-ack says 
>>>>>> he
>>>>>> got my FWD-TSN and accepted it...
>>>>>>
>>>>>> So there is NO additional info to save on the sender side and only
>>>>>> do I save what I must... on the receiver side I have a LOT of extra
>>>>>> bookkeeping....
>>>>>>
>>>>>>
>>>>>> In fact with your confirmation I would claim just the opposite of
>>>>>> Phil's statement... that doing it the way it is NOW documented 
>>>>>> takes
>>>>>> MORE overhead.. only by adding the SSN/SID can you reduce this..
>>>>>>
>>>>>> By the way Renee also pointed out the extra overhead for the IOS
>>>>>> implementation as well :>... so this makes three different
>>>>>> implementations
>>>>>> NOT 2....
>>>>>>
>>>>>> I plan on adding this to the next pr-sctp unless Michael Tuexen
>>>>>> violently
>>>>>> agrees with Qiaobing ...
>>>>>>
>>>>>> I still await Michaels answer????
>>>>>>
>>>>>>
>>>>>>>
>>>>>>> --brian
>>>>>>>
>>>>>>>>>
>>>>>>>>> (B) alternative design: more complex, may require larger 
>>>>>>>>> Forward TSN
>>>>>>>>> chunks, but avoids re-introducing an HOL blocking problem, and 
>>>>>>>>> thus
>>>>>>>>> improves performance in some situations.
>>>>>>>>
>>>>>>>>
>>>>>>>> see above it is NOT more complex...
>>>>>>>>
>>>>>>>>>
>>>>>>>>> The question of whether to choose A or B depends on whether you
>>>>>>>>> believe
>>>>>>>>> that the situation Brian describes is a "corner case" or not.   
>>>>>>>>> So
>>>>>>>>> I've
>>>>>>>>> walked through Brian's example, and have determined that I don't
>>>>>>>>> think it
>>>>>>>>> is a corner case at all.   So, I'm inclined to agree with Randy 
>>>>>>>>> and
>>>>>>>>> Brian
>>>>>>>>> that we should consider a change.
>>>>>>>>>
>>>>>>>>> But I'd like to invite others to walk through my line of 
>>>>>>>>> reasoning
>>>>>>>>> to test
>>>>>>>>> whether I've reasoned this correctly.
>>>>>>>>
>>>>>>>>
>>>>>>>> I agree it is NOT a corner case..
>>>>>>>>
>>>>>>>>>
>>>>>>>>> My rationale that Brian's example is NOT a corner case:
>>>>>>>>> ========================================
>>>>>>>>>
>>>>>>>>> In Brian's example, messages are being sent round-robin on a 
>>>>>>>>> number
>>>>>>>>> of
>>>>>>>>> streams.  This seems like a fairly reasonable use of SCTP 
>>>>>>>>> streams,
>>>>>>>>> do that
>>>>>>>>> aspect does not strike me as unusual.
>>>>>>>>
>>>>>>>> I agree :>
>>>>>>>>
>>>>>>>>>
>>>>>>>>> Now for the next part of this explanation, I'm going to have to 
>>>>>>>>> infer
>>>>>>>>> certain things, so, Brian, please correct me if I infer wrongly.
>>>>>>>>> *One*
>>>>>>>>> possibility for the sequence of events leading to the
>>>>>>>>> situation Brian describes is that there is a burst loss of data
>>>>>>>>> chunks
>>>>>>>>> TSN110 through TSN119 (ten chunks, but possibly less than ten
>>>>>>>>> packets..
>>>>>>>>> perhaps even only ONE packet if the chunks are small enough to 
>>>>>>>>> be
>>>>>>>>> bundled!)... and then data chunks TSN120 through TSN128 make it
>>>>>>>>> through
>>>>>>>>> OK.   This does not strike me as particularly unusual; a loss 
>>>>>>>>> of a
>>>>>>>>> single
>>>>>>>>> packet followed by a successful packet is sufficient to cause 
>>>>>>>>> this
>>>>>>>>> situation.
>>>>>>>>>
>>>>>>>>
>>>>>>>> Or the other possiblility would be a set of bundled chunks in one
>>>>>>>> packet
>>>>>>>> being lost.. either way is the same result I would think..
>>>>>>>>
>>>>>>>>> The next thing that happens is that TSNs 110 through 118 are
>>>>>>>>> "dropped" in
>>>>>>>>> the technical sense of that word; see the defn in the current
>>>>>>>>> draft.  This
>>>>>>>>> means that the sender gives up on sending them, and issues a 
>>>>>>>>> forward
>>>>>>>>> TSN.   However, no Forward TSN is issued for TSN119.  This may 
>>>>>>>>> be
>>>>>>>>> because
>>>>>>>>> the messages on stream 9 are treated as Reliable.   This is NOT 
>>>>>>>>> an
>>>>>>>>> unusual
>>>>>>>>> situation; it is perfectly reasonably to assume that an 
>>>>>>>>> application
>>>>>>>>> might
>>>>>>>>> have "k" streams with data that can be dropped, and "n - k" 
>>>>>>>>> streams
>>>>>>>>> of data
>>>>>>>>> that must be treated as reliable.  In the case of Brian's 
>>>>>>>>> example, n
>>>>>>>>> is 10,
>>>>>>>>> and k is 9, but I think that his example will still produce 
>>>>>>>>> similar
>>>>>>>>> results
>>>>>>>>> even for other values of "k" and "n".
>>>>>>>>>
>>>>>>>>
>>>>>>>> Note.. streams themselves are not reliable or partially 
>>>>>>>> reliable. With
>>>>>>>> the new draft it is the message itself which is sent with the
>>>>>>>> reliablilty
>>>>>>>> set to "fully reliable". One possilbe implementation would be to 
>>>>>>>> make
>>>>>>>> things correlate to a particular stream.. but this is just an
>>>>>>>> implementation
>>>>>>>> option..
>>>>>>>>
>>>>>>>>> Bottom line:
>>>>>>>>> ========
>>>>>>>>>
>>>>>>>>> I don't think its a corner case at all.   We have all the PR 
>>>>>>>>> streams
>>>>>>>>> potentially being HOL blocked by loss of a single packet.  I 
>>>>>>>>> think
>>>>>>>>> we need
>>>>>>>>> to fix this.
>>>>>>>>
>>>>>>>> I agree..
>>>>>>>>
>>>>>>>>>
>>>>>>>>> The solution of forward SSNs is one viable solution I think.  
>>>>>>>>> There
>>>>>>>>> may be
>>>>>>>>> others, but that's as good as any for right now I think.   
>>>>>>>>> There may
>>>>>>>>> be
>>>>>>>>> another solution as well, but let me do some writing and 
>>>>>>>>> thinking
>>>>>>>>> about it
>>>>>>>>> off line before I muddy the waters with it here... just to make 
>>>>>>>>> sure
>>>>>>>>> it is
>>>>>>>>> a viable alternative.
>>>>>>>>
>>>>>>>> agreed
>>>>>>>>
>>>>>>>>
>>>>>>>>>
>>>>>>>>> Randy... I'm willing to make the text changes and send a 
>>>>>>>>> proposed
>>>>>>>>> version
>>>>>>>>> around to the co-authors for comment if there is consensus to 
>>>>>>>>> move
>>>>>>>>> in that
>>>>>>>>> direction.   Let me know.
>>>>>>>>>
>>>>>>>>
>>>>>>>> I will get it out this weekend if I may.. there are some 
>>>>>>>> interesting
>>>>>>>> things that I have thought about that must be covered... I had
>>>>>>>> something
>>>>>>>> like this in an old u-sctp draft (during Q and I's debate 
>>>>>>>> earlier :>)
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>>
>>>>>>>>> (Side note: I was worried that the "technical definition" of 
>>>>>>>>> the word
>>>>>>>>> "dropped" would cause confusion, and it already has.   Perhaps
>>>>>>>>> should use
>>>>>>>>> another word in the draft to describe the operation where the 
>>>>>>>>> sender
>>>>>>>>> decides to no longer transmit and/or retransmit a specific data
>>>>>>>>> chunk.  How
>>>>>>>>> about the term "abandoned" in place of "dropped"?)
>>>>>>>>
>>>>>>>>
>>>>>>>> Sounds good..
>>>>>>>>
>>>>>>>> note, Michael R also agrees .. or seems to .. we will talk more..
>>>>>>>>
>>>>>>>> R
>>>>>>>>
>>>>>>>>
>>>>>>>>>
>>>>>>>>> /ptc
>>>>>>>>>
>>>>>>>>> At 11:44 AM 6/5/2002 -0500, Randall Stewart wrote:
>>>>>>>>>> Brian/Qiaobing:
>>>>>>>>>>
>>>>>>>>>> I have read through this interesting debate. Qiaobing and
>>>>>>>>>> I had a major discussion on this way back when it was
>>>>>>>>>> considered U-SCTP. Renee Revis first pointed this out to me.. 
>>>>>>>>>> but
>>>>>>>>>> from a different persepective i.e. the receiver being able to
>>>>>>>>>> efficently process stuck TSN's ...
>>>>>>>>>>
>>>>>>>>>> After a lengthy debate I agreed to not have the SSN/SID listed
>>>>>>>>>> in the FWD TSN... I did NOT think of the HOL blocking issue.. 
>>>>>>>>>> or
>>>>>>>>>> I would have used it as a issue (I was the one arguing for the
>>>>>>>>>> SSN/SID in the FWD-TSN :->).
>>>>>>>>>>
>>>>>>>>>> After watching this debate.. I think I have come to the 
>>>>>>>>>> conclusion
>>>>>>>>>> that
>>>>>>>>>> Brian is correct :)
>>>>>>>>>>
>>>>>>>>>> Michael R. is here... I will discuss it with him.. I want to 
>>>>>>>>>> get
>>>>>>>>>> the other co-author's take on this one... If the others agree 
>>>>>>>>>> with
>>>>>>>>>> Brian (and me :>) then I think we will put this in.... also I 
>>>>>>>>>> would
>>>>>>>>>> like to hear what the rest of the WG feels on this???
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> I think it is reasonable and not that much overhead to include 
>>>>>>>>>> JUST
>>>>>>>>>> the dropped streams NEXT expected SSN... You would be adding 4 
>>>>>>>>>> bytes
>>>>>>>>>> for EVERY stream dropping one or more TSN's... note that this 
>>>>>>>>>> does
>>>>>>>>>> NOT mean 4 bytes per TSN but 4 bytes per where one or more 
>>>>>>>>>> TSN's
>>>>>>>>>> were
>>>>>>>>>> dropped..
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> I think this is reasonable.. it improves processing AND avoids 
>>>>>>>>>> HOL
>>>>>>>>>> blocking..
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> Michael T/Phil? what do you think... Michael R.. I will talk 
>>>>>>>>>> to you
>>>>>>>>>> at
>>>>>>>>>> the
>>>>>>>>>> next class break..
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> R
>>>>>>>>>>
>>>>>>>>>> --
>>>>>>>>>> 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)
>>>>>>>
>>>>>>> --
>>>>>>> Brian F. G. Bidulock    ¦ The reasonable man adapts himself to 
>>>>>>> the ¦
>>>>>>> bidulock@openss7.org    ¦ world; the unreasonable one persists 
>>>>>>> in  ¦
>>>>>>> http://www.openss7.org/ ¦ trying  to adapt the  world  to 
>>>>>>> himself. ¦
>>>>>>>                         ¦ Therefore  all  progress  depends on 
>>>>>>> the ¦
>>>>>>>                         ¦ unreasonable man. -- George Bernard 
>>>>>>> Shaw ¦
>>>>>>
>>>>>> --
>>>>>> 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
>>>>>>
>>>>>>
Michael.Tuexen@micmac.franken.de
>>>
>>> --
>>> Brian F. G. Bidulock    ¦ The reasonable man adapts himself to the ¦
>>> bidulock@openss7.org    ¦ world; the unreasonable one persists in  ¦
>>> http://www.openss7.org/ ¦ trying  to adapt the  world  to himself. ¦
>>>                         ¦ Therefore  all  progress  depends on the ¦
>>>                         ¦ unreasonable man. -- George Bernard Shaw ¦
>
> --
> Brian F. G. Bidulock    ¦ The reasonable man adapts himself to the ¦
> bidulock@openss7.org    ¦ world; the unreasonable one persists in  ¦
> http://www.openss7.org/ ¦ trying  to adapt the  world  to himself. ¦
>                         ¦ Therefore  all  progress  depends on the ¦
>                         ¦ unreasonable man. -- George Bernard Shaw ¦
>


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



From daemon@optimus.ietf.org  Sat Jun  8 14:02:21 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 OAA10017
	for <tsvwg-archive@odin.ietf.org>; Sat, 8 Jun 2002 14:02:20 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id OAA05797
	for tsvwg-archive@odin.ietf.org; Sat, 8 Jun 2002 14:02:52 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id NAA03680;
	Sat, 8 Jun 2002 13:18:13 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id NAA03648
	for <tsvwg@optimus.ietf.org>; Sat, 8 Jun 2002 13:18:11 -0400 (EDT)
Received: from bfgbhome.inetint.com (IDENT:root@tnt-dal-42-211.dallas.net [209.44.42.211])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id NAA09518
	for <tsvwg@ietf.org>; Sat, 8 Jun 2002 13:17:33 -0400 (EDT)
Received: (from brian@localhost)
	by bfgbhome.inetint.com (8.9.3/8.9.3) id MAA20066;
	Sat, 8 Jun 2002 12:16:37 -0500
Date: Sat, 8 Jun 2002 12:16:37 -0500
From: "Brian F. G. Bidulock" <bidulock@openss7.org>
To: Michael Tuexen <Michael.Tuexen@lurchi.franken.de>
Cc: Qiaobing Xie <Qiaobing.Xie@motorola.com>,
        Randall Stewart <randall@stewart.chicago.il.us>,
        Phillip Conrad <conrad@acm.org>,
        "Armando L. Caro Jr." <acaro@mail.eecis.udel.edu>, tsvwg@ietf.org,
        "Michael A. Ramalho" <mramalho@cisco.com>
Subject: Re: [Tsvwg] More on PRSCTP-00
Message-ID: <20020608121637.F8970@openss7.org>
Reply-To: bidulock@openss7.org
References: <20020607194954.B8970@openss7.org> <AC64F48B-7AD5-11D6-AD39-0030654C1AB6@lurchi.franken.de>
Mime-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
User-Agent: Mutt/1.2i
In-Reply-To: <AC64F48B-7AD5-11D6-AD39-0030654C1AB6@lurchi.franken.de>; from Michael.Tuexen@lurchi.franken.de on Sat, Jun 08, 2002 at 01:48:38PM +0200
Organization: http://www.openss7.org/
Dsn-Notification-To: <bidulock@openss7.org>
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

Michael,

On Sat, 08 Jun 2002, Michael Tuexen wrote:

> Brian,
> 
> see my questions below.
> 
> Best regards
> Michael
> 
> On Saturday, June 8, 2002, at 02:49 AM, Brian F. G. Bidulock wrote:
> 
> > Qiaobing,
> >
> > On Fri, 07 Jun 2002, Qiaobing Xie wrote:
> >
> >> "Brian F. G. Bidulock" wrote:
> >>>
> >>> Michael, Qiaobing,
> >>>
> >>> Here is the sender side of the example.  It is quite reasonable
> >>> that this could occur.  The receiver is blocked on streams 1
> >>> through 9 until retransmission of TSN 119 is successful.  This
> >>> is the sender side of the same example.
> >>>
> >>>     TSN 101  SID 1   SSN 1   |
> >>>     TSN 102  SID 2   SSN 1   |
> >>>     TSN 103  SID 3   SSN 1   |
> >>>     TSN 104  SID 4   SSN 1   |   SCTP Packet
> >>>     TSN 105  SID 5   SSN 1   +--------------------->
> >>>     TSN 106  SID 6   SSN 1   |
> >>>     TSN 107  SID 7   SSN 1   |
> >>>     TSN 108  SID 8   SSN 1   |
> >>>     TSN 109  SID 9   SSN 1   |
> >>>
> >>>     TSN 110  SID 1   SSN 2   |
> >>>     TSN 111  SID 2   SSN 2   |
> >>>     TSN 112  SID 3   SSN 2   |
> >>>     TSN 113  SID 4   SSN 2   |   SCTP Packet
> >>>     TSN 114  SID 5   SSN 2   +----------X
> >>>     TSN 115  SID 6   SSN 2   |         Lost
> >>>     TSN 116  SID 7   SSN 2   |
> >>>     TSN 117  SID 8   SSN 2   |
> >>>     TSN 118  SID 9   SSN 2   |
> >>>                                  SCTP Packet
> >>>     TSN 119  SID 0   SSN X   |----------X
> >>>                                        Lost
> >>>
> >>>     (Timed reliability expiry on TSNs 110 to 118)
> >
> >         <---- SACK CUM ACK 109
> Why is this SACK be send
To acknowledge the first SCTP Packet

> >
> >>>
> >>>     FORWARD TSN 118          |--------------------->
> This now means that the retransmission timer runs off. So bundle a 
> FWD-TSN and DATA.

No it does not: according to rule C3 a SACK has been received.  The
retrans timer has not expired. We are forced to send FORWARD TSN
just because a sack has arrived.

> So I really do not understand this example.

You have no imagination: remove the first two SACKs.
Then it is completely possible.

--brian


> >>
> >> Why TSN119 is not re-sent at this exactly same moment? As Michael and I
> >> both noted that, in PR-SCTP, Fwd TSN can only be triggered due to 
> >> either
> >> a "RTX Timer expiration or fast retransmission." Could you fill in the
> >> details that lead to TSN119 not being re-sent at this moment?
> >
> > With the SACKs shown, the FORWARD TSN "MUST" be sent according to
> > Rule C3.  Retransmission of TSN 119 has nothing to do with it.
> >
> > --brian
> >
> >>
> >>>
> >>>     TSN 120  SID 1   SSN 3   |
> >>>     TSN 121  SID 2   SSN 3   |
> >>>     TSN 122  SID 3   SSN 3   |
> >>>     TSN 123  SID 4   SSN 3   |   SCTP Packet
> >>>     TSN 124  SID 5   SSN 3   +--------------------->
> >>>     TSN 125  SID 6   SSN 3   |
> >>>     TSN 126  SID 7   SSN 3   |
> >>>     TSN 127  SID 8   SSN 3   |
> >>>     TSN 128  SID 9   SSN 3   |
> >>>
> >>>     (Receiver is now blocked on streams 1 through 9.)
> >>>
> >>>     (Some time later, retrans timeout on TSN 119)
> >>>
> >>>                                  SCTP Packet
> >>>     TSN 119  SID 0   SSN X   |--------------------->
> >>>
> >>> Q.E.D.
> >>>
> >>> --brian
> >>>
> >>> On Fri, 07 Jun 2002, Qiaobing Xie wrote:
> >>>
> >>>> Michael,
> >>>>
> >>>> Right on! You summarized my previous point beautifully. I start to
> >>>> realize that lately my arguments tend to become lengthy and off the
> >>>> point... maybe a sign of getting old :<
> >>>>
> >>>> So far I kind of buy Randy's point better that "adding some bits in
> >>>> exchange for ease of implementation", but still yet to be convinced
> >>>> about the HOL blocking argument made by Brian.
> >>>>
> >>>> I've asked for an "all-considered" example of the HOL blocking, but
> >>>> haven't seen one yet.
> >>>>
> >>>> regards,
> >>>> -Qiaobing
> >>>>
> >>>> Michael Tuexen wrote:
> >>>>>
> >>>>> Dear all,
> >>>>>
> >>>>> reading the thread I would like to make some remarks:
> >>>>>
> >>>>> 1. I agree that using SSN in the FWD chunk avoids HOL (which is 
> >>>>> good).
> >>>>> 2. Brians example show how HOL can occur if no SSN info in the FWD 
> >>>>> chunk
> >>>>>     is used. He only looks at the receiver.
> >>>>>
> >>>>> Brians example are always based on the assumption that there is a 
> >>>>> GAP
> >>>>> in the TSN space at the receiver side and the receiver gets a FWD 
> >>>>> TSN
> >>>>> which does not fill the whole gap. Then the HOL happens at the 
> >>>>> receiver.
> >>>>>
> >>>>> I'm just wondering how often this will happen. What are the reasons 
> >>>>> that
> >>>>> a FWD chunk will be send: RTX Timer expiration or fast 
> >>>>> retransmission.
> >>>>> In both cases I would send a FWD chunk and bundle the remaining DATA
> >>>>> chunks
> >>>>> (if possible). So there is a good chance that the whole gap is 
> >>>>> filled by
> >>>>> the FWD chunk and the retransmitted DATA chunks.
> >>>>> I'm not sure how likely Brians scenario is taking the sender into
> >>>>> account.
> >>>>>
> >>>>> So I think one has to take the senders behavior into account when
> >>>>> discussing
> >>>>> this issue.
> >>>>>
> >>>>> Best regards
> >>>>> Michael
> >>>>>
> >>>>> On Thursday, June 6, 2002, at 01:43 AM, Randall Stewart wrote:
> >>>>>
> >>>>>> "Brian F. G. Bidulock" wrote:
> >>>>>>>
> >>>>>>> Randall,
> >>>>>>>
> >>>>>>> On Wed, 05 Jun 2002, Randall Stewart wrote:
> >>>>>>>
> >>>>>>>> A few comments..
> >>>>>>>>
> >>>>>>>> Phillip Conrad wrote:
> >>>>>>>>>
> >>>>>>>>> Dear Colleagues:
> >>>>>>>>>       I agree with Randy that Brian has raised an important 
> >>>>>>>>> issue
> >>>>>>>>> for us to
> >>>>>>>>> consider, and I think has explained it quite well with a very 
> >>>>>>>>> clear,
> >>>>>>>>> illustrative example.
> >>>>>>>>>
> >>>>>>>>> The question is one of a tradeoff between two designs:
> >>>>>>>>>
> >>>>>>>>> (A) current design: simpler, fewer bits on the wire, but may
> >>>>>>>>> (re)-introduce
> >>>>>>>>> an HOL blocking problem
> >>>>>>>>
> >>>>>>>> I must disagree with your "it is simpler" ... it is NOT. The 
> >>>>>>>> question
> >>>>>>>> is
> >>>>>>>> where is the burden. Having implemented pr-sctp in the kernel of
> >>>>>>>> freebsd
> >>>>>>>> NOT having the stream number/SSN in the FWD-TSN means I must keep
> >>>>>>>> an extra set of linked lists and such to track any stream that 
> >>>>>>>> has
> >>>>>>>> pending stuck data. When a fwd-tsn arrives I must HUNT all of 
> >>>>>>>> these
> >>>>>>>> streams and look at each inbound queue.
> >>>>>>>>
> >>>>>>>> The counter to this is the sender (in Brian and my earlier 
> >>>>>>>> proposal
> >>>>>>>> that was to Q only) must track this info to send with the 
> >>>>>>>> FWD-TSN.
> >>>>>>>> Actually
> >>>>>>>> less of a burden IMHO...
> >>>>>>>
> >>>>>>> I agree.  I discovered this situation while implementing PR-SCTP 
> >>>>>>> in the
> >>>>>>> Linux Kernel.  The normal receiver can rely on the fact that the 
> >>>>>>> next
> >>>>>>> expected SSN is the last fully received SSN plus 1.  PR-SCTP 
> >>>>>>> without
> >>>>>>> a forward SSN causes the receive to have to update this expected 
> >>>>>>> value
> >>>>>>> with the next SSN in the reorder queue whenever there are no TSN 
> >>>>>>> gaps.
> >>>>>>> This even might require that when TSN gaps are filled that the 
> >>>>>>> receiver
> >>>>>>> walk all inbound stream structures to update the next expected 
> >>>>>>> SSN.
> >>>>>>> In an independent implementation, I have arrived at the same
> >>>>>>> conclusion:
> >>>>>>> that the reciever has more work to do for FORWARD TSN without a 
> >>>>>>> forward
> >>>>>>> SSN.
> >>>>>>>
> >>>>>>
> >>>>>> Yeah.. I basically had to make a special linked list set that the
> >>>>>> stream queues could be placed on.. a bit more efficient but a
> >>>>>> reall hassel to maintian a "wheel" (as I call it) of streams with
> >>>>>> pending data on it...
> >>>>>>
> >>>>>>> In fact the receiver has this work to do if it has ever received a
> >>>>>>> FORWARD TSN.  This is because it is not possible for the receiver 
> >>>>>>> to
> >>>>>>> know a priori which streams could be affected by this scenairo.
> >>>>>>> Therefore,
> >>>>>>> the load on the receiver is increased even when this case does not
> >>>>>>> occur.
> >>>>>>> It is the mere possibility of its occurence which introduces the 
> >>>>>>> load.
> >>>>>>> (Each time gaps close, the next expected SSN for all streams need 
> >>>>>>> to be
> >>>>>>> re-examined whether there are SSN gaps or not.)
> >>>>>>
> >>>>>> Exactly.. I constantly must place stream q;s on and off the 
> >>>>>>:"wheel" no
> >>>>>> matter if a fwd-tsn is EVER received...
> >>>>>>
> >>>>>>>
> >>>>>>> On the sender side, the sender must always track the last sent 
> >>>>>>> SSN and
> >>>>>>> assign the next SSN (++SSN).  Indeed the sender when generating a
> >>>>>>> FORWARD
> >>>>>>> TSN chunk would need to traverse outbound stream structures with
> >>>>>>> dropped
> >>>>>>> TSNs to formulate Forward TSNs, but this is likely to be a far 
> >>>>>>> shorter
> >>>>>>> list than at the receiver and non-PR streams would never need to 
> >>>>>>> be
> >>>>>>> involved.
> >>>>>>
> >>>>>> And it must continue to KEEP this information until it is sure the
> >>>>>> peer has received and moved the cum-ack forward..
> >>>>>>
> >>>>>> What I end up doing on the sender side is I hold just the 
> >>>>>> bookkeeping
> >>>>>> structures of the TSN (stream seq/stream no/tsn ... etc) but I
> >>>>>> drop the mbuf holding the data so I gain a bit of help in mbuf
> >>>>>> utilization.
> >>>>>>
> >>>>>> But I can't get rid of the chunk structure UNTIL the cum-ack says 
> >>>>>> he
> >>>>>> got my FWD-TSN and accepted it...
> >>>>>>
> >>>>>> So there is NO additional info to save on the sender side and only
> >>>>>> do I save what I must... on the receiver side I have a LOT of extra
> >>>>>> bookkeeping....
> >>>>>>
> >>>>>>
> >>>>>> In fact with your confirmation I would claim just the opposite of
> >>>>>> Phil's statement... that doing it the way it is NOW documented 
> >>>>>> takes
> >>>>>> MORE overhead.. only by adding the SSN/SID can you reduce this..
> >>>>>>
> >>>>>> By the way Renee also pointed out the extra overhead for the IOS
> >>>>>> implementation as well :>... so this makes three different
> >>>>>> implementations
> >>>>>> NOT 2....
> >>>>>>
> >>>>>> I plan on adding this to the next pr-sctp unless Michael Tuexen
> >>>>>> violently
> >>>>>> agrees with Qiaobing ...
> >>>>>>
> >>>>>> I still await Michaels answer????
> >>>>>>
> >>>>>>
> >>>>>>>
> >>>>>>> --brian
> >>>>>>>
> >>>>>>>>>
> >>>>>>>>> (B) alternative design: more complex, may require larger 
> >>>>>>>>> Forward TSN
> >>>>>>>>> chunks, but avoids re-introducing an HOL blocking problem, and 
> >>>>>>>>> thus
> >>>>>>>>> improves performance in some situations.
> >>>>>>>>
> >>>>>>>>
> >>>>>>>> see above it is NOT more complex...
> >>>>>>>>
> >>>>>>>>>
> >>>>>>>>> The question of whether to choose A or B depends on whether you
> >>>>>>>>> believe
> >>>>>>>>> that the situation Brian describes is a "corner case" or not.   
> >>>>>>>>> So
> >>>>>>>>> I've
> >>>>>>>>> walked through Brian's example, and have determined that I don't
> >>>>>>>>> think it
> >>>>>>>>> is a corner case at all.   So, I'm inclined to agree with Randy 
> >>>>>>>>> and
> >>>>>>>>> Brian
> >>>>>>>>> that we should consider a change.
> >>>>>>>>>
> >>>>>>>>> But I'd like to invite others to walk through my line of 
> >>>>>>>>> reasoning
> >>>>>>>>> to test
> >>>>>>>>> whether I've reasoned this correctly.
> >>>>>>>>
> >>>>>>>>
> >>>>>>>> I agree it is NOT a corner case..
> >>>>>>>>
> >>>>>>>>>
> >>>>>>>>> My rationale that Brian's example is NOT a corner case:
> >>>>>>>>> ========================================
> >>>>>>>>>
> >>>>>>>>> In Brian's example, messages are being sent round-robin on a 
> >>>>>>>>> number
> >>>>>>>>> of
> >>>>>>>>> streams.  This seems like a fairly reasonable use of SCTP 
> >>>>>>>>> streams,
> >>>>>>>>> do that
> >>>>>>>>> aspect does not strike me as unusual.
> >>>>>>>>
> >>>>>>>> I agree :>
> >>>>>>>>
> >>>>>>>>>
> >>>>>>>>> Now for the next part of this explanation, I'm going to have to 
> >>>>>>>>> infer
> >>>>>>>>> certain things, so, Brian, please correct me if I infer wrongly.
> >>>>>>>>> *One*
> >>>>>>>>> possibility for the sequence of events leading to the
> >>>>>>>>> situation Brian describes is that there is a burst loss of data
> >>>>>>>>> chunks
> >>>>>>>>> TSN110 through TSN119 (ten chunks, but possibly less than ten
> >>>>>>>>> packets..
> >>>>>>>>> perhaps even only ONE packet if the chunks are small enough to 
> >>>>>>>>> be
> >>>>>>>>> bundled!)... and then data chunks TSN120 through TSN128 make it
> >>>>>>>>> through
> >>>>>>>>> OK.   This does not strike me as particularly unusual; a loss 
> >>>>>>>>> of a
> >>>>>>>>> single
> >>>>>>>>> packet followed by a successful packet is sufficient to cause 
> >>>>>>>>> this
> >>>>>>>>> situation.
> >>>>>>>>>
> >>>>>>>>
> >>>>>>>> Or the other possiblility would be a set of bundled chunks in one
> >>>>>>>> packet
> >>>>>>>> being lost.. either way is the same result I would think..
> >>>>>>>>
> >>>>>>>>> The next thing that happens is that TSNs 110 through 118 are
> >>>>>>>>> "dropped" in
> >>>>>>>>> the technical sense of that word; see the defn in the current
> >>>>>>>>> draft.  This
> >>>>>>>>> means that the sender gives up on sending them, and issues a 
> >>>>>>>>> forward
> >>>>>>>>> TSN.   However, no Forward TSN is issued for TSN119.  This may 
> >>>>>>>>> be
> >>>>>>>>> because
> >>>>>>>>> the messages on stream 9 are treated as Reliable.   This is NOT 
> >>>>>>>>> an
> >>>>>>>>> unusual
> >>>>>>>>> situation; it is perfectly reasonably to assume that an 
> >>>>>>>>> application
> >>>>>>>>> might
> >>>>>>>>> have "k" streams with data that can be dropped, and "n - k" 
> >>>>>>>>> streams
> >>>>>>>>> of data
> >>>>>>>>> that must be treated as reliable.  In the case of Brian's 
> >>>>>>>>> example, n
> >>>>>>>>> is 10,
> >>>>>>>>> and k is 9, but I think that his example will still produce 
> >>>>>>>>> similar
> >>>>>>>>> results
> >>>>>>>>> even for other values of "k" and "n".
> >>>>>>>>>
> >>>>>>>>
> >>>>>>>> Note.. streams themselves are not reliable or partially 
> >>>>>>>> reliable. With
> >>>>>>>> the new draft it is the message itself which is sent with the
> >>>>>>>> reliablilty
> >>>>>>>> set to "fully reliable". One possilbe implementation would be to 
> >>>>>>>> make
> >>>>>>>> things correlate to a particular stream.. but this is just an
> >>>>>>>> implementation
> >>>>>>>> option..
> >>>>>>>>
> >>>>>>>>> Bottom line:
> >>>>>>>>> ========
> >>>>>>>>>
> >>>>>>>>> I don't think its a corner case at all.   We have all the PR 
> >>>>>>>>> streams
> >>>>>>>>> potentially being HOL blocked by loss of a single packet.  I 
> >>>>>>>>> think
> >>>>>>>>> we need
> >>>>>>>>> to fix this.
> >>>>>>>>
> >>>>>>>> I agree..
> >>>>>>>>
> >>>>>>>>>
> >>>>>>>>> The solution of forward SSNs is one viable solution I think.  
> >>>>>>>>> There
> >>>>>>>>> may be
> >>>>>>>>> others, but that's as good as any for right now I think.   
> >>>>>>>>> There may
> >>>>>>>>> be
> >>>>>>>>> another solution as well, but let me do some writing and 
> >>>>>>>>> thinking
> >>>>>>>>> about it
> >>>>>>>>> off line before I muddy the waters with it here... just to make 
> >>>>>>>>> sure
> >>>>>>>>> it is
> >>>>>>>>> a viable alternative.
> >>>>>>>>
> >>>>>>>> agreed
> >>>>>>>>
> >>>>>>>>
> >>>>>>>>>
> >>>>>>>>> Randy... I'm willing to make the text changes and send a 
> >>>>>>>>> proposed
> >>>>>>>>> version
> >>>>>>>>> around to the co-authors for comment if there is consensus to 
> >>>>>>>>> move
> >>>>>>>>> in that
> >>>>>>>>> direction.   Let me know.
> >>>>>>>>>
> >>>>>>>>
> >>>>>>>> I will get it out this weekend if I may.. there are some 
> >>>>>>>> interesting
> >>>>>>>> things that I have thought about that must be covered... I had
> >>>>>>>> something
> >>>>>>>> like this in an old u-sctp draft (during Q and I's debate 
> >>>>>>>> earlier :>)
> >>>>>>>>
> >>>>>>>>
> >>>>>>>>
> >>>>>>>>>
> >>>>>>>>> (Side note: I was worried that the "technical definition" of 
> >>>>>>>>> the word
> >>>>>>>>> "dropped" would cause confusion, and it already has.   Perhaps
> >>>>>>>>> should use
> >>>>>>>>> another word in the draft to describe the operation where the 
> >>>>>>>>> sender
> >>>>>>>>> decides to no longer transmit and/or retransmit a specific data
> >>>>>>>>> chunk.  How
> >>>>>>>>> about the term "abandoned" in place of "dropped"?)
> >>>>>>>>
> >>>>>>>>
> >>>>>>>> Sounds good..
> >>>>>>>>
> >>>>>>>> note, Michael R also agrees .. or seems to .. we will talk more..
> >>>>>>>>
> >>>>>>>> R
> >>>>>>>>
> >>>>>>>>
> >>>>>>>>>
> >>>>>>>>> /ptc
> >>>>>>>>>
> >>>>>>>>> At 11:44 AM 6/5/2002 -0500, Randall Stewart wrote:
> >>>>>>>>>> Brian/Qiaobing:
> >>>>>>>>>>
> >>>>>>>>>> I have read through this interesting debate. Qiaobing and
> >>>>>>>>>> I had a major discussion on this way back when it was
> >>>>>>>>>> considered U-SCTP. Renee Revis first pointed this out to me.. 
> >>>>>>>>>> but
> >>>>>>>>>> from a different persepective i.e. the receiver being able to
> >>>>>>>>>> efficently process stuck TSN's ...
> >>>>>>>>>>
> >>>>>>>>>> After a lengthy debate I agreed to not have the SSN/SID listed
> >>>>>>>>>> in the FWD TSN... I did NOT think of the HOL blocking issue.. 
> >>>>>>>>>> or
> >>>>>>>>>> I would have used it as a issue (I was the one arguing for the
> >>>>>>>>>> SSN/SID in the FWD-TSN :->).
> >>>>>>>>>>
> >>>>>>>>>> After watching this debate.. I think I have come to the 
> >>>>>>>>>> conclusion
> >>>>>>>>>> that
> >>>>>>>>>> Brian is correct :)
> >>>>>>>>>>
> >>>>>>>>>> Michael R. is here... I will discuss it with him.. I want to 
> >>>>>>>>>> get
> >>>>>>>>>> the other co-author's take on this one... If the others agree 
> >>>>>>>>>> with
> >>>>>>>>>> Brian (and me :>) then I think we will put this in.... also I 
> >>>>>>>>>> would
> >>>>>>>>>> like to hear what the rest of the WG feels on this???
> >>>>>>>>>>
> >>>>>>>>>>
> >>>>>>>>>> I think it is reasonable and not that much overhead to include 
> >>>>>>>>>> JUST
> >>>>>>>>>> the dropped streams NEXT expected SSN... You would be adding 4 
> >>>>>>>>>> bytes
> >>>>>>>>>> for EVERY stream dropping one or more TSN's... note that this 
> >>>>>>>>>> does
> >>>>>>>>>> NOT mean 4 bytes per TSN but 4 bytes per where one or more 
> >>>>>>>>>> TSN's
> >>>>>>>>>> were
> >>>>>>>>>> dropped..
> >>>>>>>>>>
> >>>>>>>>>>
> >>>>>>>>>> I think this is reasonable.. it improves processing AND avoids 
> >>>>>>>>>> HOL
> >>>>>>>>>> blocking..
> >>>>>>>>>>
> >>>>>>>>>>
> >>>>>>>>>> Michael T/Phil? what do you think... Michael R.. I will talk 
> >>>>>>>>>> to you
> >>>>>>>>>> at
> >>>>>>>>>> the
> >>>>>>>>>> next class break..
> >>>>>>>>>>
> >>>>>>>>>>
> >>>>>>>>>>
> >>>>>>>>>> R
> >>>>>>>>>>
> >>>>>>>>>> --
> >>>>>>>>>> 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)
> >>>>>>>
> >>>>>>> --
> >>>>>>> Brian F. G. Bidulock    ¦ The reasonable man adapts himself to 
> >>>>>>> the ¦
> >>>>>>> bidulock@openss7.org    ¦ world; the unreasonable one persists 
> >>>>>>> in  ¦
> >>>>>>> http://www.openss7.org/ ¦ trying  to adapt the  world  to 
> >>>>>>> himself. ¦
> >>>>>>>                         ¦ Therefore  all  progress  depends on 
> >>>>>>> the ¦
> >>>>>>>                         ¦ unreasonable man. -- George Bernard 
> >>>>>>> Shaw ¦
> >>>>>>
> >>>>>> --
> >>>>>> 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
> >>>>>>
> >>>>>>
> Michael.Tuexen@micmac.franken.de
> >>>
> >>> --
> >>> Brian F. G. Bidulock    ¦ The reasonable man adapts himself to the ¦
> >>> bidulock@openss7.org    ¦ world; the unreasonable one persists in  ¦
> >>> http://www.openss7.org/ ¦ trying  to adapt the  world  to himself. ¦
> >>>                         ¦ Therefore  all  progress  depends on the ¦
> >>>                         ¦ unreasonable man. -- George Bernard Shaw ¦
> >
> > --
> > Brian F. G. Bidulock    ¦ The reasonable man adapts himself to the ¦
> > bidulock@openss7.org    ¦ world; the unreasonable one persists in  ¦
> > http://www.openss7.org/ ¦ trying  to adapt the  world  to himself. ¦
> >                         ¦ Therefore  all  progress  depends on the ¦
> >                         ¦ unreasonable man. -- George Bernard Shaw ¦
> >

-- 
Brian F. G. Bidulock    ¦ The reasonable man adapts himself to the ¦
bidulock@openss7.org    ¦ world; the unreasonable one persists in  ¦
http://www.openss7.org/ ¦ trying  to adapt the  world  to himself. ¦
                        ¦ Therefore  all  progress  depends on the ¦
                        ¦ unreasonable man. -- George Bernard Shaw ¦

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



From daemon@optimus.ietf.org  Sat Jun  8 14:51:54 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 OAA10488
	for <tsvwg-archive@odin.ietf.org>; Sat, 8 Jun 2002 14:51:54 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id OAA07245
	for tsvwg-archive@odin.ietf.org; Sat, 8 Jun 2002 14:52:26 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id NAA05133;
	Sat, 8 Jun 2002 13:59:27 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id NAA05047
	for <tsvwg@optimus.ietf.org>; Sat, 8 Jun 2002 13:59:23 -0400 (EDT)
Received: from ilsa.franken.de (postfix@ilsa.franken.de [193.175.24.42])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id NAA09961
	for <tsvwg@ietf.org>; Sat, 8 Jun 2002 13:58:50 -0400 (EDT)
Received: from localhost (pD9E4F7CF.dip.t-dialin.net [217.228.247.207])
	by ilsa.franken.de (Postfix) with ESMTP
	id 449DB245D4; Sat,  8 Jun 2002 19:59:17 +0200 (CEST)
Date: Sat, 8 Jun 2002 19:59:00 +0200
Subject: Re: [Tsvwg] More on PRSCTP-00
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Mime-Version: 1.0 (Apple Message framework v482)
Cc: Qiaobing Xie <Qiaobing.Xie@motorola.com>,
        Randall Stewart <randall@stewart.chicago.il.us>,
        Phillip Conrad <conrad@acm.org>,
        "Armando L. Caro Jr." <acaro@mail.eecis.udel.edu>, tsvwg@ietf.org,
        "Michael A. Ramalho" <mramalho@cisco.com>
To: bidulock@openss7.org
From: Michael Tuexen <Michael.Tuexen@lurchi.franken.de>
In-Reply-To: <20020608121637.F8970@openss7.org>
Message-Id: <69718456-7B09-11D6-9BA0-0030654C1AB6@lurchi.franken.de>
X-Mailer: Apple Mail (2.482)
Content-Transfer-Encoding: 8bit
X-MIME-Autoconverted: from quoted-printable to 8bit by optimus.ietf.org id NAA05048
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

Brian, Qiaobing,

I'm a bit confused by the answer, because there are some SACKs less than
in the previous email.

I will repeat the scenario:

Consider
packet1 = TSN 101-109; SID=1-9; SSN=1;
packet2 = TSN 110-119; SID=1-9; SSN=2;
packet3 = TSN 119;     SID=0;
packet4 = TSN 120-128; SID=1-9; SSN=3;

It seems that you are now considering the szenario:
1. packet1 is transmitted and received
2. packet2 is transmitted and lost on the wire
3. packet3 is transmitted and lost on the wire
4. timed reliability runs off for TSN 110-119
5. packet4 is transmitted and received. The user data can not be 
delivered.
6. A SACK for packet 1 is received.
7. A FWD TSN chunk is sent

This happens only if the timed reliability is shorter than the round 
trip time.
But we can assume this.

So what will happen:
8. The retransmission timer will run off for packet3.
9. packet3 gets retransmitted, and the user data of packet3 and packet4 
can be delivered.

I agree with you that there is HOL, but it is the same as there would be 
if you
would not use PR-SCTP. So PR-SCTP, as it is currently being defined does 
not perform
better than basic SCTP in terms of user-to-user message delay. It does 
perform better in terms
of buffer and bandwidth usage.

So I think using SSN info in the FWD TSN chunk would optimize PR-SCTP 
also in terms of
user-to-user delay. Qiaobing, could you agree on this?

Best regards
Michael
On Saturday, June 8, 2002, at 07:16 PM, Brian F. G. Bidulock wrote:

> Michael,
>
> On Sat, 08 Jun 2002, Michael Tuexen wrote:
>
>> Brian,
>>
>> see my questions below.
>>
>> Best regards
>> Michael
>>
>> On Saturday, June 8, 2002, at 02:49 AM, Brian F. G. Bidulock wrote:
>>
>>> Qiaobing,
>>>
>>> On Fri, 07 Jun 2002, Qiaobing Xie wrote:
>>>
>>>> "Brian F. G. Bidulock" wrote:
>>>>>
>>>>> Michael, Qiaobing,
>>>>>
>>>>> Here is the sender side of the example.  It is quite reasonable
>>>>> that this could occur.  The receiver is blocked on streams 1
>>>>> through 9 until retransmission of TSN 119 is successful.  This
>>>>> is the sender side of the same example.
>>>>>
>>>>>     TSN 101  SID 1   SSN 1   |
>>>>>     TSN 102  SID 2   SSN 1   |
>>>>>     TSN 103  SID 3   SSN 1   |
>>>>>     TSN 104  SID 4   SSN 1   |   SCTP Packet
>>>>>     TSN 105  SID 5   SSN 1   +--------------------->
>>>>>     TSN 106  SID 6   SSN 1   |
>>>>>     TSN 107  SID 7   SSN 1   |
>>>>>     TSN 108  SID 8   SSN 1   |
>>>>>     TSN 109  SID 9   SSN 1   |
>>>>>
>>>>>     TSN 110  SID 1   SSN 2   |
>>>>>     TSN 111  SID 2   SSN 2   |
>>>>>     TSN 112  SID 3   SSN 2   |
>>>>>     TSN 113  SID 4   SSN 2   |   SCTP Packet
>>>>>     TSN 114  SID 5   SSN 2   +----------X
>>>>>     TSN 115  SID 6   SSN 2   |         Lost
>>>>>     TSN 116  SID 7   SSN 2   |
>>>>>     TSN 117  SID 8   SSN 2   |
>>>>>     TSN 118  SID 9   SSN 2   |
>>>>>                                  SCTP Packet
>>>>>     TSN 119  SID 0   SSN X   |----------X
>>>>>                                        Lost
>>>>>
>>>>>     (Timed reliability expiry on TSNs 110 to 118)
>>>
>>>         <---- SACK CUM ACK 109
>> Why is this SACK be send
> To acknowledge the first SCTP Packet
>
>>>
>>>>>
>>>>>     FORWARD TSN 118          |--------------------->
>> This now means that the retransmission timer runs off. So bundle a
>> FWD-TSN and DATA.
>
> No it does not: according to rule C3 a SACK has been received.  The
> retrans timer has not expired. We are forced to send FORWARD TSN
> just because a sack has arrived.
>
>> So I really do not understand this example.
>
> You have no imagination: remove the first two SACKs.
> Then it is completely possible.
>
> --brian
>
>
>>>>
>>>> Why TSN119 is not re-sent at this exactly same moment? As Michael 
>>>> and I
>>>> both noted that, in PR-SCTP, Fwd TSN can only be triggered due to
>>>> either
>>>> a "RTX Timer expiration or fast retransmission." Could you fill in 
>>>> the
>>>> details that lead to TSN119 not being re-sent at this moment?
>>>
>>> With the SACKs shown, the FORWARD TSN "MUST" be sent according to
>>> Rule C3.  Retransmission of TSN 119 has nothing to do with it.
>>>
>>> --brian
>>>
>>>>
>>>>>
>>>>>     TSN 120  SID 1   SSN 3   |
>>>>>     TSN 121  SID 2   SSN 3   |
>>>>>     TSN 122  SID 3   SSN 3   |
>>>>>     TSN 123  SID 4   SSN 3   |   SCTP Packet
>>>>>     TSN 124  SID 5   SSN 3   +--------------------->
>>>>>     TSN 125  SID 6   SSN 3   |
>>>>>     TSN 126  SID 7   SSN 3   |
>>>>>     TSN 127  SID 8   SSN 3   |
>>>>>     TSN 128  SID 9   SSN 3   |
>>>>>
>>>>>     (Receiver is now blocked on streams 1 through 9.)
>>>>>
>>>>>     (Some time later, retrans timeout on TSN 119)
>>>>>
>>>>>                                  SCTP Packet
>>>>>     TSN 119  SID 0   SSN X   |--------------------->
>>>>>
>>>>> Q.E.D.
>>>>>
>>>>> --brian
>>>>>
>>>>> On Fri, 07 Jun 2002, Qiaobing Xie wrote:
>>>>>
>>>>>> Michael,
>>>>>>
>>>>>> Right on! You summarized my previous point beautifully. I start to
>>>>>> realize that lately my arguments tend to become lengthy and off the
>>>>>> point... maybe a sign of getting old :<
>>>>>>
>>>>>> So far I kind of buy Randy's point better that "adding some bits in
>>>>>> exchange for ease of implementation", but still yet to be convinced
>>>>>> about the HOL blocking argument made by Brian.
>>>>>>
>>>>>> I've asked for an "all-considered" example of the HOL blocking, but
>>>>>> haven't seen one yet.
>>>>>>
>>>>>> regards,
>>>>>> -Qiaobing
>>>>>>
>>>>>> Michael Tuexen wrote:
>>>>>>>
>>>>>>> Dear all,
>>>>>>>
>>>>>>> reading the thread I would like to make some remarks:
>>>>>>>
>>>>>>> 1. I agree that using SSN in the FWD chunk avoids HOL (which is
>>>>>>> good).
>>>>>>> 2. Brians example show how HOL can occur if no SSN info in the FWD
>>>>>>> chunk
>>>>>>>     is used. He only looks at the receiver.
>>>>>>>
>>>>>>> Brians example are always based on the assumption that there is a
>>>>>>> GAP
>>>>>>> in the TSN space at the receiver side and the receiver gets a FWD
>>>>>>> TSN
>>>>>>> which does not fill the whole gap. Then the HOL happens at the
>>>>>>> receiver.
>>>>>>>
>>>>>>> I'm just wondering how often this will happen. What are the 
>>>>>>> reasons
>>>>>>> that
>>>>>>> a FWD chunk will be send: RTX Timer expiration or fast
>>>>>>> retransmission.
>>>>>>> In both cases I would send a FWD chunk and bundle the remaining 
>>>>>>> DATA
>>>>>>> chunks
>>>>>>> (if possible). So there is a good chance that the whole gap is
>>>>>>> filled by
>>>>>>> the FWD chunk and the retransmitted DATA chunks.
>>>>>>> I'm not sure how likely Brians scenario is taking the sender into
>>>>>>> account.
>>>>>>>
>>>>>>> So I think one has to take the senders behavior into account when
>>>>>>> discussing
>>>>>>> this issue.
>>>>>>>
>>>>>>> Best regards
>>>>>>> Michael
>>>>>>>
>>>>>>> On Thursday, June 6, 2002, at 01:43 AM, Randall Stewart wrote:
>>>>>>>
>>>>>>>> "Brian F. G. Bidulock" wrote:
>>>>>>>>>
>>>>>>>>> Randall,
>>>>>>>>>
>>>>>>>>> On Wed, 05 Jun 2002, Randall Stewart wrote:
>>>>>>>>>
>>>>>>>>>> A few comments..
>>>>>>>>>>
>>>>>>>>>> Phillip Conrad wrote:
>>>>>>>>>>>
>>>>>>>>>>> Dear Colleagues:
>>>>>>>>>>>       I agree with Randy that Brian has raised an important
>>>>>>>>>>> issue
>>>>>>>>>>> for us to
>>>>>>>>>>> consider, and I think has explained it quite well with a very
>>>>>>>>>>> clear,
>>>>>>>>>>> illustrative example.
>>>>>>>>>>>
>>>>>>>>>>> The question is one of a tradeoff between two designs:
>>>>>>>>>>>
>>>>>>>>>>> (A) current design: simpler, fewer bits on the wire, but may
>>>>>>>>>>> (re)-introduce
>>>>>>>>>>> an HOL blocking problem
>>>>>>>>>>
>>>>>>>>>> I must disagree with your "it is simpler" ... it is NOT. The
>>>>>>>>>> question
>>>>>>>>>> is
>>>>>>>>>> where is the burden. Having implemented pr-sctp in the kernel 
>>>>>>>>>> of
>>>>>>>>>> freebsd
>>>>>>>>>> NOT having the stream number/SSN in the FWD-TSN means I must 
>>>>>>>>>> keep
>>>>>>>>>> an extra set of linked lists and such to track any stream that
>>>>>>>>>> has
>>>>>>>>>> pending stuck data. When a fwd-tsn arrives I must HUNT all of
>>>>>>>>>> these
>>>>>>>>>> streams and look at each inbound queue.
>>>>>>>>>>
>>>>>>>>>> The counter to this is the sender (in Brian and my earlier
>>>>>>>>>> proposal
>>>>>>>>>> that was to Q only) must track this info to send with the
>>>>>>>>>> FWD-TSN.
>>>>>>>>>> Actually
>>>>>>>>>> less of a burden IMHO...
>>>>>>>>>
>>>>>>>>> I agree.  I discovered this situation while implementing PR-SCTP
>>>>>>>>> in the
>>>>>>>>> Linux Kernel.  The normal receiver can rely on the fact that the
>>>>>>>>> next
>>>>>>>>> expected SSN is the last fully received SSN plus 1.  PR-SCTP
>>>>>>>>> without
>>>>>>>>> a forward SSN causes the receive to have to update this expected
>>>>>>>>> value
>>>>>>>>> with the next SSN in the reorder queue whenever there are no TSN
>>>>>>>>> gaps.
>>>>>>>>> This even might require that when TSN gaps are filled that the
>>>>>>>>> receiver
>>>>>>>>> walk all inbound stream structures to update the next expected
>>>>>>>>> SSN.
>>>>>>>>> In an independent implementation, I have arrived at the same
>>>>>>>>> conclusion:
>>>>>>>>> that the reciever has more work to do for FORWARD TSN without a
>>>>>>>>> forward
>>>>>>>>> SSN.
>>>>>>>>>
>>>>>>>>
>>>>>>>> Yeah.. I basically had to make a special linked list set that the
>>>>>>>> stream queues could be placed on.. a bit more efficient but a
>>>>>>>> reall hassel to maintian a "wheel" (as I call it) of streams with
>>>>>>>> pending data on it...
>>>>>>>>
>>>>>>>>> In fact the receiver has this work to do if it has ever 
>>>>>>>>> received a
>>>>>>>>> FORWARD TSN.  This is because it is not possible for the 
>>>>>>>>> receiver
>>>>>>>>> to
>>>>>>>>> know a priori which streams could be affected by this scenairo.
>>>>>>>>> Therefore,
>>>>>>>>> the load on the receiver is increased even when this case does 
>>>>>>>>> not
>>>>>>>>> occur.
>>>>>>>>> It is the mere possibility of its occurence which introduces the
>>>>>>>>> load.
>>>>>>>>> (Each time gaps close, the next expected SSN for all streams 
>>>>>>>>> need
>>>>>>>>> to be
>>>>>>>>> re-examined whether there are SSN gaps or not.)
>>>>>>>>
>>>>>>>> Exactly.. I constantly must place stream q;s on and off the
>>>>>>>> :"wheel" no
>>>>>>>> matter if a fwd-tsn is EVER received...
>>>>>>>>
>>>>>>>>>
>>>>>>>>> On the sender side, the sender must always track the last sent
>>>>>>>>> SSN and
>>>>>>>>> assign the next SSN (++SSN).  Indeed the sender when 
>>>>>>>>> generating a
>>>>>>>>> FORWARD
>>>>>>>>> TSN chunk would need to traverse outbound stream structures with
>>>>>>>>> dropped
>>>>>>>>> TSNs to formulate Forward TSNs, but this is likely to be a far
>>>>>>>>> shorter
>>>>>>>>> list than at the receiver and non-PR streams would never need to
>>>>>>>>> be
>>>>>>>>> involved.
>>>>>>>>
>>>>>>>> And it must continue to KEEP this information until it is sure 
>>>>>>>> the
>>>>>>>> peer has received and moved the cum-ack forward..
>>>>>>>>
>>>>>>>> What I end up doing on the sender side is I hold just the
>>>>>>>> bookkeeping
>>>>>>>> structures of the TSN (stream seq/stream no/tsn ... etc) but I
>>>>>>>> drop the mbuf holding the data so I gain a bit of help in mbuf
>>>>>>>> utilization.
>>>>>>>>
>>>>>>>> But I can't get rid of the chunk structure UNTIL the cum-ack says
>>>>>>>> he
>>>>>>>> got my FWD-TSN and accepted it...
>>>>>>>>
>>>>>>>> So there is NO additional info to save on the sender side and 
>>>>>>>> only
>>>>>>>> do I save what I must... on the receiver side I have a LOT of 
>>>>>>>> extra
>>>>>>>> bookkeeping....
>>>>>>>>
>>>>>>>>
>>>>>>>> In fact with your confirmation I would claim just the opposite of
>>>>>>>> Phil's statement... that doing it the way it is NOW documented
>>>>>>>> takes
>>>>>>>> MORE overhead.. only by adding the SSN/SID can you reduce this..
>>>>>>>>
>>>>>>>> By the way Renee also pointed out the extra overhead for the IOS
>>>>>>>> implementation as well :>... so this makes three different
>>>>>>>> implementations
>>>>>>>> NOT 2....
>>>>>>>>
>>>>>>>> I plan on adding this to the next pr-sctp unless Michael Tuexen
>>>>>>>> violently
>>>>>>>> agrees with Qiaobing ...
>>>>>>>>
>>>>>>>> I still await Michaels answer????
>>>>>>>>
>>>>>>>>
>>>>>>>>>
>>>>>>>>> --brian
>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>> (B) alternative design: more complex, may require larger
>>>>>>>>>>> Forward TSN
>>>>>>>>>>> chunks, but avoids re-introducing an HOL blocking problem, and
>>>>>>>>>>> thus
>>>>>>>>>>> improves performance in some situations.
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> see above it is NOT more complex...
>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>> The question of whether to choose A or B depends on whether 
>>>>>>>>>>> you
>>>>>>>>>>> believe
>>>>>>>>>>> that the situation Brian describes is a "corner case" or not.
>>>>>>>>>>> So
>>>>>>>>>>> I've
>>>>>>>>>>> walked through Brian's example, and have determined that I 
>>>>>>>>>>> don't
>>>>>>>>>>> think it
>>>>>>>>>>> is a corner case at all.   So, I'm inclined to agree with 
>>>>>>>>>>> Randy
>>>>>>>>>>> and
>>>>>>>>>>> Brian
>>>>>>>>>>> that we should consider a change.
>>>>>>>>>>>
>>>>>>>>>>> But I'd like to invite others to walk through my line of
>>>>>>>>>>> reasoning
>>>>>>>>>>> to test
>>>>>>>>>>> whether I've reasoned this correctly.
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> I agree it is NOT a corner case..
>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>> My rationale that Brian's example is NOT a corner case:
>>>>>>>>>>> ========================================
>>>>>>>>>>>
>>>>>>>>>>> In Brian's example, messages are being sent round-robin on a
>>>>>>>>>>> number
>>>>>>>>>>> of
>>>>>>>>>>> streams.  This seems like a fairly reasonable use of SCTP
>>>>>>>>>>> streams,
>>>>>>>>>>> do that
>>>>>>>>>>> aspect does not strike me as unusual.
>>>>>>>>>>
>>>>>>>>>> I agree :>
>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>> Now for the next part of this explanation, I'm going to have 
>>>>>>>>>>> to
>>>>>>>>>>> infer
>>>>>>>>>>> certain things, so, Brian, please correct me if I infer 
>>>>>>>>>>> wrongly.
>>>>>>>>>>> *One*
>>>>>>>>>>> possibility for the sequence of events leading to the
>>>>>>>>>>> situation Brian describes is that there is a burst loss of 
>>>>>>>>>>> data
>>>>>>>>>>> chunks
>>>>>>>>>>> TSN110 through TSN119 (ten chunks, but possibly less than ten
>>>>>>>>>>> packets..
>>>>>>>>>>> perhaps even only ONE packet if the chunks are small enough to
>>>>>>>>>>> be
>>>>>>>>>>> bundled!)... and then data chunks TSN120 through TSN128 make 
>>>>>>>>>>> it
>>>>>>>>>>> through
>>>>>>>>>>> OK.   This does not strike me as particularly unusual; a loss
>>>>>>>>>>> of a
>>>>>>>>>>> single
>>>>>>>>>>> packet followed by a successful packet is sufficient to cause
>>>>>>>>>>> this
>>>>>>>>>>> situation.
>>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> Or the other possiblility would be a set of bundled chunks in 
>>>>>>>>>> one
>>>>>>>>>> packet
>>>>>>>>>> being lost.. either way is the same result I would think..
>>>>>>>>>>
>>>>>>>>>>> The next thing that happens is that TSNs 110 through 118 are
>>>>>>>>>>> "dropped" in
>>>>>>>>>>> the technical sense of that word; see the defn in the current
>>>>>>>>>>> draft.  This
>>>>>>>>>>> means that the sender gives up on sending them, and issues a
>>>>>>>>>>> forward
>>>>>>>>>>> TSN.   However, no Forward TSN is issued for TSN119.  This may
>>>>>>>>>>> be
>>>>>>>>>>> because
>>>>>>>>>>> the messages on stream 9 are treated as Reliable.   This is 
>>>>>>>>>>> NOT
>>>>>>>>>>> an
>>>>>>>>>>> unusual
>>>>>>>>>>> situation; it is perfectly reasonably to assume that an
>>>>>>>>>>> application
>>>>>>>>>>> might
>>>>>>>>>>> have "k" streams with data that can be dropped, and "n - k"
>>>>>>>>>>> streams
>>>>>>>>>>> of data
>>>>>>>>>>> that must be treated as reliable.  In the case of Brian's
>>>>>>>>>>> example, n
>>>>>>>>>>> is 10,
>>>>>>>>>>> and k is 9, but I think that his example will still produce
>>>>>>>>>>> similar
>>>>>>>>>>> results
>>>>>>>>>>> even for other values of "k" and "n".
>>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> Note.. streams themselves are not reliable or partially
>>>>>>>>>> reliable. With
>>>>>>>>>> the new draft it is the message itself which is sent with the
>>>>>>>>>> reliablilty
>>>>>>>>>> set to "fully reliable". One possilbe implementation would be 
>>>>>>>>>> to
>>>>>>>>>> make
>>>>>>>>>> things correlate to a particular stream.. but this is just an
>>>>>>>>>> implementation
>>>>>>>>>> option..
>>>>>>>>>>
>>>>>>>>>>> Bottom line:
>>>>>>>>>>> ========
>>>>>>>>>>>
>>>>>>>>>>> I don't think its a corner case at all.   We have all the PR
>>>>>>>>>>> streams
>>>>>>>>>>> potentially being HOL blocked by loss of a single packet.  I
>>>>>>>>>>> think
>>>>>>>>>>> we need
>>>>>>>>>>> to fix this.
>>>>>>>>>>
>>>>>>>>>> I agree..
>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>> The solution of forward SSNs is one viable solution I think.
>>>>>>>>>>> There
>>>>>>>>>>> may be
>>>>>>>>>>> others, but that's as good as any for right now I think.
>>>>>>>>>>> There may
>>>>>>>>>>> be
>>>>>>>>>>> another solution as well, but let me do some writing and
>>>>>>>>>>> thinking
>>>>>>>>>>> about it
>>>>>>>>>>> off line before I muddy the waters with it here... just to 
>>>>>>>>>>> make
>>>>>>>>>>> sure
>>>>>>>>>>> it is
>>>>>>>>>>> a viable alternative.
>>>>>>>>>>
>>>>>>>>>> agreed
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>> Randy... I'm willing to make the text changes and send a
>>>>>>>>>>> proposed
>>>>>>>>>>> version
>>>>>>>>>>> around to the co-authors for comment if there is consensus to
>>>>>>>>>>> move
>>>>>>>>>>> in that
>>>>>>>>>>> direction.   Let me know.
>>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> I will get it out this weekend if I may.. there are some
>>>>>>>>>> interesting
>>>>>>>>>> things that I have thought about that must be covered... I had
>>>>>>>>>> something
>>>>>>>>>> like this in an old u-sctp draft (during Q and I's debate
>>>>>>>>>> earlier :>)
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>> (Side note: I was worried that the "technical definition" of
>>>>>>>>>>> the word
>>>>>>>>>>> "dropped" would cause confusion, and it already has.   Perhaps
>>>>>>>>>>> should use
>>>>>>>>>>> another word in the draft to describe the operation where the
>>>>>>>>>>> sender
>>>>>>>>>>> decides to no longer transmit and/or retransmit a specific 
>>>>>>>>>>> data
>>>>>>>>>>> chunk.  How
>>>>>>>>>>> about the term "abandoned" in place of "dropped"?)
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> Sounds good..
>>>>>>>>>>
>>>>>>>>>> note, Michael R also agrees .. or seems to .. we will talk 
>>>>>>>>>> more..
>>>>>>>>>>
>>>>>>>>>> R
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>> /ptc
>>>>>>>>>>>
>>>>>>>>>>> At 11:44 AM 6/5/2002 -0500, Randall Stewart wrote:
>>>>>>>>>>>> Brian/Qiaobing:
>>>>>>>>>>>>
>>>>>>>>>>>> I have read through this interesting debate. Qiaobing and
>>>>>>>>>>>> I had a major discussion on this way back when it was
>>>>>>>>>>>> considered U-SCTP. Renee Revis first pointed this out to me..
>>>>>>>>>>>> but
>>>>>>>>>>>> from a different persepective i.e. the receiver being able to
>>>>>>>>>>>> efficently process stuck TSN's ...
>>>>>>>>>>>>
>>>>>>>>>>>> After a lengthy debate I agreed to not have the SSN/SID 
>>>>>>>>>>>> listed
>>>>>>>>>>>> in the FWD TSN... I did NOT think of the HOL blocking issue..
>>>>>>>>>>>> or
>>>>>>>>>>>> I would have used it as a issue (I was the one arguing for 
>>>>>>>>>>>> the
>>>>>>>>>>>> SSN/SID in the FWD-TSN :->).
>>>>>>>>>>>>
>>>>>>>>>>>> After watching this debate.. I think I have come to the
>>>>>>>>>>>> conclusion
>>>>>>>>>>>> that
>>>>>>>>>>>> Brian is correct :)
>>>>>>>>>>>>
>>>>>>>>>>>> Michael R. is here... I will discuss it with him.. I want to
>>>>>>>>>>>> get
>>>>>>>>>>>> the other co-author's take on this one... If the others agree
>>>>>>>>>>>> with
>>>>>>>>>>>> Brian (and me :>) then I think we will put this in.... also I
>>>>>>>>>>>> would
>>>>>>>>>>>> like to hear what the rest of the WG feels on this???
>>>>>>>>>>>>
>>>>>>>>>>>>
>>>>>>>>>>>> I think it is reasonable and not that much overhead to 
>>>>>>>>>>>> include
>>>>>>>>>>>> JUST
>>>>>>>>>>>> the dropped streams NEXT expected SSN... You would be 
>>>>>>>>>>>> adding 4
>>>>>>>>>>>> bytes
>>>>>>>>>>>> for EVERY stream dropping one or more TSN's... note that this
>>>>>>>>>>>> does
>>>>>>>>>>>> NOT mean 4 bytes per TSN but 4 bytes per where one or more
>>>>>>>>>>>> TSN's
>>>>>>>>>>>> were
>>>>>>>>>>>> dropped..
>>>>>>>>>>>>
>>>>>>>>>>>>
>>>>>>>>>>>> I think this is reasonable.. it improves processing AND 
>>>>>>>>>>>> avoids
>>>>>>>>>>>> HOL
>>>>>>>>>>>> blocking..
>>>>>>>>>>>>
>>>>>>>>>>>>
>>>>>>>>>>>> Michael T/Phil? what do you think... Michael R.. I will talk
>>>>>>>>>>>> to you
>>>>>>>>>>>> at
>>>>>>>>>>>> the
>>>>>>>>>>>> next class break..
>>>>>>>>>>>>
>>>>>>>>>>>>
>>>>>>>>>>>>
>>>>>>>>>>>> R
>>>>>>>>>>>>
>>>>>>>>>>>> --
>>>>>>>>>>>> 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)
>>>>>>>>>
>>>>>>>>> --
>>>>>>>>> Brian F. G. Bidulock    ¦ The reasonable man adapts himself to
>>>>>>>>> the ¦
>>>>>>>>> bidulock@openss7.org    ¦ world; the unreasonable one persists
>>>>>>>>> in  ¦
>>>>>>>>> http://www.openss7.org/ ¦ trying  to adapt the  world  to
>>>>>>>>> himself. ¦
>>>>>>>>>                         ¦ Therefore  all  progress  depends on
>>>>>>>>> the ¦
>>>>>>>>>                         ¦ unreasonable man. -- George Bernard
>>>>>>>>> Shaw ¦
>>>>>>>>
>>>>>>>> --
>>>>>>>> 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
>>>>>>>>
>>>>>>>>
Michael.Tuexen@micmac.franken.de
>>>>>
>>>>> --
>>>>> Brian F. G. Bidulock    ¦ The reasonable man adapts himself to the ¦
>>>>> bidulock@openss7.org    ¦ world; the unreasonable one persists in  ¦
>>>>> http://www.openss7.org/ ¦ trying  to adapt the  world  to himself. ¦
>>>>>                         ¦ Therefore  all  progress  depends on the ¦
>>>>>                         ¦ unreasonable man. -- George Bernard Shaw ¦
>>>
>>> --
>>> Brian F. G. Bidulock    ¦ The reasonable man adapts himself to the ¦
>>> bidulock@openss7.org    ¦ world; the unreasonable one persists in  ¦
>>> http://www.openss7.org/ ¦ trying  to adapt the  world  to himself. ¦
>>>                         ¦ Therefore  all  progress  depends on the ¦
>>>                         ¦ unreasonable man. -- George Bernard Shaw ¦
>>>
>
> --
> Brian F. G. Bidulock    ¦ The reasonable man adapts himself to the ¦
> bidulock@openss7.org    ¦ world; the unreasonable one persists in  ¦
> http://www.openss7.org/ ¦ trying  to adapt the  world  to himself. ¦
>                         ¦ Therefore  all  progress  depends on the ¦
>                         ¦ unreasonable man. -- George Bernard Shaw ¦
>


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



From daemon@optimus.ietf.org  Sat Jun  8 15:59:52 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 PAA11050
	for <tsvwg-archive@odin.ietf.org>; Sat, 8 Jun 2002 15:59:52 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id QAA09399
	for tsvwg-archive@odin.ietf.org; Sat, 8 Jun 2002 16:00:24 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id PAA08190;
	Sat, 8 Jun 2002 15:28:30 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id PAA08162
	for <tsvwg@optimus.ietf.org>; Sat, 8 Jun 2002 15:28:28 -0400 (EDT)
Received: from bfgbhome.inetint.com (IDENT:root@tnt-dal-42-211.dallas.net [209.44.42.211])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id PAA10741
	for <tsvwg@ietf.org>; Sat, 8 Jun 2002 15:27:50 -0400 (EDT)
Received: (from brian@localhost)
	by bfgbhome.inetint.com (8.9.3/8.9.3) id OAA21409;
	Sat, 8 Jun 2002 14:26:12 -0500
Date: Sat, 8 Jun 2002 14:26:12 -0500
From: "Brian F. G. Bidulock" <bidulock@openss7.org>
To: Michael Tuexen <Michael.Tuexen@lurchi.franken.de>
Cc: Qiaobing Xie <Qiaobing.Xie@motorola.com>,
        Randall Stewart <randall@stewart.chicago.il.us>,
        Phillip Conrad <conrad@acm.org>,
        "Armando L. Caro Jr." <acaro@mail.eecis.udel.edu>, tsvwg@ietf.org,
        "Michael A. Ramalho" <mramalho@cisco.com>
Subject: Re: [Tsvwg] More on PRSCTP-00
Message-ID: <20020608142612.D20649@openss7.org>
Reply-To: bidulock@openss7.org
References: <20020608121637.F8970@openss7.org> <69718456-7B09-11D6-9BA0-0030654C1AB6@lurchi.franken.de>
Mime-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
User-Agent: Mutt/1.2i
In-Reply-To: <69718456-7B09-11D6-9BA0-0030654C1AB6@lurchi.franken.de>; from Michael.Tuexen@lurchi.franken.de on Sat, Jun 08, 2002 at 07:59:00PM +0200
Organization: http://www.openss7.org/
Dsn-Notification-To: <bidulock@openss7.org>
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

Michael,

On Sat, 08 Jun 2002, Michael Tuexen wrote:

> Brian, Qiaobing,
> 
> I'm a bit confused by the answer, because there are some SACKs less than
> in the previous email.
> 
> I will repeat the scenario:
> 
> Consider
> packet1 = TSN 101-109; SID=1-9; SSN=1;
> packet2 = TSN 110-119; SID=1-9; SSN=2;

I think you mean TSN 110-118 here.

> packet3 = TSN 119;     SID=0;
> packet4 = TSN 120-128; SID=1-9; SSN=3;
> 
> It seems that you are now considering the szenario:
> 1. packet1 is transmitted and received
> 2. packet2 is transmitted and lost on the wire
> 3. packet3 is transmitted and lost on the wire
> 4. timed reliability runs off for TSN 110-119
> 5. packet4 is transmitted and received. The user data can not be 
> delivered.
> 6. A SACK for packet 1 is received.
> 7. A FWD TSN chunk is sent
> 
> This happens only if the timed reliability is shorter than the round 
> trip time.

No that is not necessary.  The SACK (containing no gaps) may be
a "delayed" sack delayed by 500ms so timed reliability only needs to
be shorter than the current (experienced by the SACK) RTT + 500ms.
Nevertheless, it is easy for a fixed "timed reliability" to be less
than the current RTT+500ms at any given point, particularly when
packets are being lost in the network due to congestion.

> But we can assume this.

The retransmission timer may be as high as 60 seconds at the moment,
so it is quite conceivable that "timed reliability" is less than the
sender's RTO.  In fact, for "timed reliability" to work as it is
expected, the timed reliability interval MUST be less than the product
of Assoc.Max.Retrans and Rto.Max, otherwise, packets will never expire.
(The association will fail before they expire.)

> 
> So what will happen:
> 8. The retransmission timer will run off for packet3.
> 9. packet3 gets retransmitted, and the user data of packet3 and packet4 
> can be delivered.
> 
> I agree with you that there is HOL, but it is the same as there would be 
> if you
> would not use PR-SCTP. So PR-SCTP, as it is currently being defined does 
> not perform
> better than basic SCTP in terms of user-to-user message delay. It does 
> perform better in terms
> of buffer and bandwidth usage.

I disagree with that analysis.  Normal SCTP only has to wait for
retransmission of the SSN 2 packet, whereas PR-SCTP in this scenario
has to wait for retransmission of all the TSN gaps subsequent to
the FORWARD TSN.  These (e.g. TSN 119) might have 16k worth of data
in it making it subject to many retransmissions (the network is already
marginally in congestion).  The SSN 2 packet might contain 256 bytes
meaning that retransmission of TSNs 110-118 could be successful when
TSN 119 is going to be unsuccessful for some time.

It is NOT the situation in regular SCTP that a stream must block
awaiting the retransmission of a TSN from a different stream.  For
PR-SCTP, streams 1 through 9 are blocked awaiting the retransmission
of a data chunk on stream 0.  This is the unacceptable HOL blocking.
It is inter-stream, not intra-stream.

--brian

> 
> So I think using SSN info in the FWD TSN chunk would optimize PR-SCTP 
> also in terms of
> user-to-user delay. Qiaobing, could you agree on this?
> 
> Best regards
> Michael
> On Saturday, June 8, 2002, at 07:16 PM, Brian F. G. Bidulock wrote:
> 
> > Michael,
> >
> > On Sat, 08 Jun 2002, Michael Tuexen wrote:
> >
> >> Brian,
> >>
> >> see my questions below.
> >>
> >> Best regards
> >> Michael
> >>
> >> On Saturday, June 8, 2002, at 02:49 AM, Brian F. G. Bidulock wrote:
> >>
> >>> Qiaobing,
> >>>
> >>> On Fri, 07 Jun 2002, Qiaobing Xie wrote:
> >>>
> >>>> "Brian F. G. Bidulock" wrote:
> >>>>>
> >>>>> Michael, Qiaobing,
> >>>>>
> >>>>> Here is the sender side of the example.  It is quite reasonable
> >>>>> that this could occur.  The receiver is blocked on streams 1
> >>>>> through 9 until retransmission of TSN 119 is successful.  This
> >>>>> is the sender side of the same example.
> >>>>>
> >>>>>     TSN 101  SID 1   SSN 1   |
> >>>>>     TSN 102  SID 2   SSN 1   |
> >>>>>     TSN 103  SID 3   SSN 1   |
> >>>>>     TSN 104  SID 4   SSN 1   |   SCTP Packet
> >>>>>     TSN 105  SID 5   SSN 1   +--------------------->
> >>>>>     TSN 106  SID 6   SSN 1   |
> >>>>>     TSN 107  SID 7   SSN 1   |
> >>>>>     TSN 108  SID 8   SSN 1   |
> >>>>>     TSN 109  SID 9   SSN 1   |
> >>>>>
> >>>>>     TSN 110  SID 1   SSN 2   |
> >>>>>     TSN 111  SID 2   SSN 2   |
> >>>>>     TSN 112  SID 3   SSN 2   |
> >>>>>     TSN 113  SID 4   SSN 2   |   SCTP Packet
> >>>>>     TSN 114  SID 5   SSN 2   +----------X
> >>>>>     TSN 115  SID 6   SSN 2   |         Lost
> >>>>>     TSN 116  SID 7   SSN 2   |
> >>>>>     TSN 117  SID 8   SSN 2   |
> >>>>>     TSN 118  SID 9   SSN 2   |
> >>>>>                                  SCTP Packet
> >>>>>     TSN 119  SID 0   SSN X   |----------X
> >>>>>                                        Lost
> >>>>>
> >>>>>     (Timed reliability expiry on TSNs 110 to 118)
> >>>
> >>>         <---- SACK CUM ACK 109
> >> Why is this SACK be send
> > To acknowledge the first SCTP Packet
> >
> >>>
> >>>>>
> >>>>>     FORWARD TSN 118          |--------------------->
> >> This now means that the retransmission timer runs off. So bundle a
> >> FWD-TSN and DATA.
> >
> > No it does not: according to rule C3 a SACK has been received.  The
> > retrans timer has not expired. We are forced to send FORWARD TSN
> > just because a sack has arrived.
> >
> >> So I really do not understand this example.
> >
> > You have no imagination: remove the first two SACKs.
> > Then it is completely possible.
> >
> > --brian
> >
> >
> >>>>
> >>>> Why TSN119 is not re-sent at this exactly same moment? As Michael 
> >>>> and I
> >>>> both noted that, in PR-SCTP, Fwd TSN can only be triggered due to
> >>>> either
> >>>> a "RTX Timer expiration or fast retransmission." Could you fill in 
> >>>> the
> >>>> details that lead to TSN119 not being re-sent at this moment?
> >>>
> >>> With the SACKs shown, the FORWARD TSN "MUST" be sent according to
> >>> Rule C3.  Retransmission of TSN 119 has nothing to do with it.
> >>>
> >>> --brian
> >>>
> >>>>
> >>>>>
> >>>>>     TSN 120  SID 1   SSN 3   |
> >>>>>     TSN 121  SID 2   SSN 3   |
> >>>>>     TSN 122  SID 3   SSN 3   |
> >>>>>     TSN 123  SID 4   SSN 3   |   SCTP Packet
> >>>>>     TSN 124  SID 5   SSN 3   +--------------------->
> >>>>>     TSN 125  SID 6   SSN 3   |
> >>>>>     TSN 126  SID 7   SSN 3   |
> >>>>>     TSN 127  SID 8   SSN 3   |
> >>>>>     TSN 128  SID 9   SSN 3   |
> >>>>>
> >>>>>     (Receiver is now blocked on streams 1 through 9.)
> >>>>>
> >>>>>     (Some time later, retrans timeout on TSN 119)
> >>>>>
> >>>>>                                  SCTP Packet
> >>>>>     TSN 119  SID 0   SSN X   |--------------------->
> >>>>>
> >>>>> Q.E.D.
> >>>>>
> >>>>> --brian
> >>>>>
> >>>>> On Fri, 07 Jun 2002, Qiaobing Xie wrote:
> >>>>>
> >>>>>> Michael,
> >>>>>>
> >>>>>> Right on! You summarized my previous point beautifully. I start to
> >>>>>> realize that lately my arguments tend to become lengthy and off the
> >>>>>> point... maybe a sign of getting old :<
> >>>>>>
> >>>>>> So far I kind of buy Randy's point better that "adding some bits in
> >>>>>> exchange for ease of implementation", but still yet to be convinced
> >>>>>> about the HOL blocking argument made by Brian.
> >>>>>>
> >>>>>> I've asked for an "all-considered" example of the HOL blocking, but
> >>>>>> haven't seen one yet.
> >>>>>>
> >>>>>> regards,
> >>>>>> -Qiaobing
> >>>>>>
> >>>>>> Michael Tuexen wrote:
> >>>>>>>
> >>>>>>> Dear all,
> >>>>>>>
> >>>>>>> reading the thread I would like to make some remarks:
> >>>>>>>
> >>>>>>> 1. I agree that using SSN in the FWD chunk avoids HOL (which is
> >>>>>>> good).
> >>>>>>> 2. Brians example show how HOL can occur if no SSN info in the FWD
> >>>>>>> chunk
> >>>>>>>     is used. He only looks at the receiver.
> >>>>>>>
> >>>>>>> Brians example are always based on the assumption that there is a
> >>>>>>> GAP
> >>>>>>> in the TSN space at the receiver side and the receiver gets a FWD
> >>>>>>> TSN
> >>>>>>> which does not fill the whole gap. Then the HOL happens at the
> >>>>>>> receiver.
> >>>>>>>
> >>>>>>> I'm just wondering how often this will happen. What are the 
> >>>>>>> reasons
> >>>>>>> that
> >>>>>>> a FWD chunk will be send: RTX Timer expiration or fast
> >>>>>>> retransmission.
> >>>>>>> In both cases I would send a FWD chunk and bundle the remaining 
> >>>>>>> DATA
> >>>>>>> chunks
> >>>>>>> (if possible). So there is a good chance that the whole gap is
> >>>>>>> filled by
> >>>>>>> the FWD chunk and the retransmitted DATA chunks.
> >>>>>>> I'm not sure how likely Brians scenario is taking the sender into
> >>>>>>> account.
> >>>>>>>
> >>>>>>> So I think one has to take the senders behavior into account when
> >>>>>>> discussing
> >>>>>>> this issue.
> >>>>>>>
> >>>>>>> Best regards
> >>>>>>> Michael
> >>>>>>>
> >>>>>>> On Thursday, June 6, 2002, at 01:43 AM, Randall Stewart wrote:
> >>>>>>>
> >>>>>>>> "Brian F. G. Bidulock" wrote:
> >>>>>>>>>
> >>>>>>>>> Randall,
> >>>>>>>>>
> >>>>>>>>> On Wed, 05 Jun 2002, Randall Stewart wrote:
> >>>>>>>>>
> >>>>>>>>>> A few comments..
> >>>>>>>>>>
> >>>>>>>>>> Phillip Conrad wrote:
> >>>>>>>>>>>
> >>>>>>>>>>> Dear Colleagues:
> >>>>>>>>>>>       I agree with Randy that Brian has raised an important
> >>>>>>>>>>> issue
> >>>>>>>>>>> for us to
> >>>>>>>>>>> consider, and I think has explained it quite well with a very
> >>>>>>>>>>> clear,
> >>>>>>>>>>> illustrative example.
> >>>>>>>>>>>
> >>>>>>>>>>> The question is one of a tradeoff between two designs:
> >>>>>>>>>>>
> >>>>>>>>>>> (A) current design: simpler, fewer bits on the wire, but may
> >>>>>>>>>>> (re)-introduce
> >>>>>>>>>>> an HOL blocking problem
> >>>>>>>>>>
> >>>>>>>>>> I must disagree with your "it is simpler" ... it is NOT. The
> >>>>>>>>>> question
> >>>>>>>>>> is
> >>>>>>>>>> where is the burden. Having implemented pr-sctp in the kernel 
> >>>>>>>>>> of
> >>>>>>>>>> freebsd
> >>>>>>>>>> NOT having the stream number/SSN in the FWD-TSN means I must 
> >>>>>>>>>> keep
> >>>>>>>>>> an extra set of linked lists and such to track any stream that
> >>>>>>>>>> has
> >>>>>>>>>> pending stuck data. When a fwd-tsn arrives I must HUNT all of
> >>>>>>>>>> these
> >>>>>>>>>> streams and look at each inbound queue.
> >>>>>>>>>>
> >>>>>>>>>> The counter to this is the sender (in Brian and my earlier
> >>>>>>>>>> proposal
> >>>>>>>>>> that was to Q only) must track this info to send with the
> >>>>>>>>>> FWD-TSN.
> >>>>>>>>>> Actually
> >>>>>>>>>> less of a burden IMHO...
> >>>>>>>>>
> >>>>>>>>> I agree.  I discovered this situation while implementing PR-SCTP
> >>>>>>>>> in the
> >>>>>>>>> Linux Kernel.  The normal receiver can rely on the fact that the
> >>>>>>>>> next
> >>>>>>>>> expected SSN is the last fully received SSN plus 1.  PR-SCTP
> >>>>>>>>> without
> >>>>>>>>> a forward SSN causes the receive to have to update this expected
> >>>>>>>>> value
> >>>>>>>>> with the next SSN in the reorder queue whenever there are no TSN
> >>>>>>>>> gaps.
> >>>>>>>>> This even might require that when TSN gaps are filled that the
> >>>>>>>>> receiver
> >>>>>>>>> walk all inbound stream structures to update the next expected
> >>>>>>>>> SSN.
> >>>>>>>>> In an independent implementation, I have arrived at the same
> >>>>>>>>> conclusion:
> >>>>>>>>> that the reciever has more work to do for FORWARD TSN without a
> >>>>>>>>> forward
> >>>>>>>>> SSN.
> >>>>>>>>>
> >>>>>>>>
> >>>>>>>> Yeah.. I basically had to make a special linked list set that the
> >>>>>>>> stream queues could be placed on.. a bit more efficient but a
> >>>>>>>> reall hassel to maintian a "wheel" (as I call it) of streams with
> >>>>>>>> pending data on it...
> >>>>>>>>
> >>>>>>>>> In fact the receiver has this work to do if it has ever 
> >>>>>>>>> received a
> >>>>>>>>> FORWARD TSN.  This is because it is not possible for the 
> >>>>>>>>> receiver
> >>>>>>>>> to
> >>>>>>>>> know a priori which streams could be affected by this scenairo.
> >>>>>>>>> Therefore,
> >>>>>>>>> the load on the receiver is increased even when this case does 
> >>>>>>>>> not
> >>>>>>>>> occur.
> >>>>>>>>> It is the mere possibility of its occurence which introduces the
> >>>>>>>>> load.
> >>>>>>>>> (Each time gaps close, the next expected SSN for all streams 
> >>>>>>>>> need
> >>>>>>>>> to be
> >>>>>>>>> re-examined whether there are SSN gaps or not.)
> >>>>>>>>
> >>>>>>>> Exactly.. I constantly must place stream q;s on and off the
> >>>>>>>>:"wheel" no
> >>>>>>>> matter if a fwd-tsn is EVER received...
> >>>>>>>>
> >>>>>>>>>
> >>>>>>>>> On the sender side, the sender must always track the last sent
> >>>>>>>>> SSN and
> >>>>>>>>> assign the next SSN (++SSN).  Indeed the sender when 
> >>>>>>>>> generating a
> >>>>>>>>> FORWARD
> >>>>>>>>> TSN chunk would need to traverse outbound stream structures with
> >>>>>>>>> dropped
> >>>>>>>>> TSNs to formulate Forward TSNs, but this is likely to be a far
> >>>>>>>>> shorter
> >>>>>>>>> list than at the receiver and non-PR streams would never need to
> >>>>>>>>> be
> >>>>>>>>> involved.
> >>>>>>>>
> >>>>>>>> And it must continue to KEEP this information until it is sure 
> >>>>>>>> the
> >>>>>>>> peer has received and moved the cum-ack forward..
> >>>>>>>>
> >>>>>>>> What I end up doing on the sender side is I hold just the
> >>>>>>>> bookkeeping
> >>>>>>>> structures of the TSN (stream seq/stream no/tsn ... etc) but I
> >>>>>>>> drop the mbuf holding the data so I gain a bit of help in mbuf
> >>>>>>>> utilization.
> >>>>>>>>
> >>>>>>>> But I can't get rid of the chunk structure UNTIL the cum-ack says
> >>>>>>>> he
> >>>>>>>> got my FWD-TSN and accepted it...
> >>>>>>>>
> >>>>>>>> So there is NO additional info to save on the sender side and 
> >>>>>>>> only
> >>>>>>>> do I save what I must... on the receiver side I have a LOT of 
> >>>>>>>> extra
> >>>>>>>> bookkeeping....
> >>>>>>>>
> >>>>>>>>
> >>>>>>>> In fact with your confirmation I would claim just the opposite of
> >>>>>>>> Phil's statement... that doing it the way it is NOW documented
> >>>>>>>> takes
> >>>>>>>> MORE overhead.. only by adding the SSN/SID can you reduce this..
> >>>>>>>>
> >>>>>>>> By the way Renee also pointed out the extra overhead for the IOS
> >>>>>>>> implementation as well :>... so this makes three different
> >>>>>>>> implementations
> >>>>>>>> NOT 2....
> >>>>>>>>
> >>>>>>>> I plan on adding this to the next pr-sctp unless Michael Tuexen
> >>>>>>>> violently
> >>>>>>>> agrees with Qiaobing ...
> >>>>>>>>
> >>>>>>>> I still await Michaels answer????
> >>>>>>>>
> >>>>>>>>
> >>>>>>>>>
> >>>>>>>>> --brian
> >>>>>>>>>
> >>>>>>>>>>>
> >>>>>>>>>>> (B) alternative design: more complex, may require larger
> >>>>>>>>>>> Forward TSN
> >>>>>>>>>>> chunks, but avoids re-introducing an HOL blocking problem, and
> >>>>>>>>>>> thus
> >>>>>>>>>>> improves performance in some situations.
> >>>>>>>>>>
> >>>>>>>>>>
> >>>>>>>>>> see above it is NOT more complex...
> >>>>>>>>>>
> >>>>>>>>>>>
> >>>>>>>>>>> The question of whether to choose A or B depends on whether 
> >>>>>>>>>>> you
> >>>>>>>>>>> believe
> >>>>>>>>>>> that the situation Brian describes is a "corner case" or not.
> >>>>>>>>>>> So
> >>>>>>>>>>> I've
> >>>>>>>>>>> walked through Brian's example, and have determined that I 
> >>>>>>>>>>> don't
> >>>>>>>>>>> think it
> >>>>>>>>>>> is a corner case at all.   So, I'm inclined to agree with 
> >>>>>>>>>>> Randy
> >>>>>>>>>>> and
> >>>>>>>>>>> Brian
> >>>>>>>>>>> that we should consider a change.
> >>>>>>>>>>>
> >>>>>>>>>>> But I'd like to invite others to walk through my line of
> >>>>>>>>>>> reasoning
> >>>>>>>>>>> to test
> >>>>>>>>>>> whether I've reasoned this correctly.
> >>>>>>>>>>
> >>>>>>>>>>
> >>>>>>>>>> I agree it is NOT a corner case..
> >>>>>>>>>>
> >>>>>>>>>>>
> >>>>>>>>>>> My rationale that Brian's example is NOT a corner case:
> >>>>>>>>>>> ========================================
> >>>>>>>>>>>
> >>>>>>>>>>> In Brian's example, messages are being sent round-robin on a
> >>>>>>>>>>> number
> >>>>>>>>>>> of
> >>>>>>>>>>> streams.  This seems like a fairly reasonable use of SCTP
> >>>>>>>>>>> streams,
> >>>>>>>>>>> do that
> >>>>>>>>>>> aspect does not strike me as unusual.
> >>>>>>>>>>
> >>>>>>>>>> I agree :>
> >>>>>>>>>>
> >>>>>>>>>>>
> >>>>>>>>>>> Now for the next part of this explanation, I'm going to have 
> >>>>>>>>>>> to
> >>>>>>>>>>> infer
> >>>>>>>>>>> certain things, so, Brian, please correct me if I infer 
> >>>>>>>>>>> wrongly.
> >>>>>>>>>>> *One*
> >>>>>>>>>>> possibility for the sequence of events leading to the
> >>>>>>>>>>> situation Brian describes is that there is a burst loss of 
> >>>>>>>>>>> data
> >>>>>>>>>>> chunks
> >>>>>>>>>>> TSN110 through TSN119 (ten chunks, but possibly less than ten
> >>>>>>>>>>> packets..
> >>>>>>>>>>> perhaps even only ONE packet if the chunks are small enough to
> >>>>>>>>>>> be
> >>>>>>>>>>> bundled!)... and then data chunks TSN120 through TSN128 make 
> >>>>>>>>>>> it
> >>>>>>>>>>> through
> >>>>>>>>>>> OK.   This does not strike me as particularly unusual; a loss
> >>>>>>>>>>> of a
> >>>>>>>>>>> single
> >>>>>>>>>>> packet followed by a successful packet is sufficient to cause
> >>>>>>>>>>> this
> >>>>>>>>>>> situation.
> >>>>>>>>>>>
> >>>>>>>>>>
> >>>>>>>>>> Or the other possiblility would be a set of bundled chunks in 
> >>>>>>>>>> one
> >>>>>>>>>> packet
> >>>>>>>>>> being lost.. either way is the same result I would think..
> >>>>>>>>>>
> >>>>>>>>>>> The next thing that happens is that TSNs 110 through 118 are
> >>>>>>>>>>> "dropped" in
> >>>>>>>>>>> the technical sense of that word; see the defn in the current
> >>>>>>>>>>> draft.  This
> >>>>>>>>>>> means that the sender gives up on sending them, and issues a
> >>>>>>>>>>> forward
> >>>>>>>>>>> TSN.   However, no Forward TSN is issued for TSN119.  This may
> >>>>>>>>>>> be
> >>>>>>>>>>> because
> >>>>>>>>>>> the messages on stream 9 are treated as Reliable.   This is 
> >>>>>>>>>>> NOT
> >>>>>>>>>>> an
> >>>>>>>>>>> unusual
> >>>>>>>>>>> situation; it is perfectly reasonably to assume that an
> >>>>>>>>>>> application
> >>>>>>>>>>> might
> >>>>>>>>>>> have "k" streams with data that can be dropped, and "n - k"
> >>>>>>>>>>> streams
> >>>>>>>>>>> of data
> >>>>>>>>>>> that must be treated as reliable.  In the case of Brian's
> >>>>>>>>>>> example, n
> >>>>>>>>>>> is 10,
> >>>>>>>>>>> and k is 9, but I think that his example will still produce
> >>>>>>>>>>> similar
> >>>>>>>>>>> results
> >>>>>>>>>>> even for other values of "k" and "n".
> >>>>>>>>>>>
> >>>>>>>>>>
> >>>>>>>>>> Note.. streams themselves are not reliable or partially
> >>>>>>>>>> reliable. With
> >>>>>>>>>> the new draft it is the message itself which is sent with the
> >>>>>>>>>> reliablilty
> >>>>>>>>>> set to "fully reliable". One possilbe implementation would be 
> >>>>>>>>>> to
> >>>>>>>>>> make
> >>>>>>>>>> things correlate to a particular stream.. but this is just an
> >>>>>>>>>> implementation
> >>>>>>>>>> option..
> >>>>>>>>>>
> >>>>>>>>>>> Bottom line:
> >>>>>>>>>>> ========
> >>>>>>>>>>>
> >>>>>>>>>>> I don't think its a corner case at all.   We have all the PR
> >>>>>>>>>>> streams
> >>>>>>>>>>> potentially being HOL blocked by loss of a single packet.  I
> >>>>>>>>>>> think
> >>>>>>>>>>> we need
> >>>>>>>>>>> to fix this.
> >>>>>>>>>>
> >>>>>>>>>> I agree..
> >>>>>>>>>>
> >>>>>>>>>>>
> >>>>>>>>>>> The solution of forward SSNs is one viable solution I think.
> >>>>>>>>>>> There
> >>>>>>>>>>> may be
> >>>>>>>>>>> others, but that's as good as any for right now I think.
> >>>>>>>>>>> There may
> >>>>>>>>>>> be
> >>>>>>>>>>> another solution as well, but let me do some writing and
> >>>>>>>>>>> thinking
> >>>>>>>>>>> about it
> >>>>>>>>>>> off line before I muddy the waters with it here... just to 
> >>>>>>>>>>> make
> >>>>>>>>>>> sure
> >>>>>>>>>>> it is
> >>>>>>>>>>> a viable alternative.
> >>>>>>>>>>
> >>>>>>>>>> agreed
> >>>>>>>>>>
> >>>>>>>>>>
> >>>>>>>>>>>
> >>>>>>>>>>> Randy... I'm willing to make the text changes and send a
> >>>>>>>>>>> proposed
> >>>>>>>>>>> version
> >>>>>>>>>>> around to the co-authors for comment if there is consensus to
> >>>>>>>>>>> move
> >>>>>>>>>>> in that
> >>>>>>>>>>> direction.   Let me know.
> >>>>>>>>>>>
> >>>>>>>>>>
> >>>>>>>>>> I will get it out this weekend if I may.. there are some
> >>>>>>>>>> interesting
> >>>>>>>>>> things that I have thought about that must be covered... I had
> >>>>>>>>>> something
> >>>>>>>>>> like this in an old u-sctp draft (during Q and I's debate
> >>>>>>>>>> earlier :>)
> >>>>>>>>>>
> >>>>>>>>>>
> >>>>>>>>>>
> >>>>>>>>>>>
> >>>>>>>>>>> (Side note: I was worried that the "technical definition" of
> >>>>>>>>>>> the word
> >>>>>>>>>>> "dropped" would cause confusion, and it already has.   Perhaps
> >>>>>>>>>>> should use
> >>>>>>>>>>> another word in the draft to describe the operation where the
> >>>>>>>>>>> sender
> >>>>>>>>>>> decides to no longer transmit and/or retransmit a specific 
> >>>>>>>>>>> data
> >>>>>>>>>>> chunk.  How
> >>>>>>>>>>> about the term "abandoned" in place of "dropped"?)
> >>>>>>>>>>
> >>>>>>>>>>
> >>>>>>>>>> Sounds good..
> >>>>>>>>>>
> >>>>>>>>>> note, Michael R also agrees .. or seems to .. we will talk 
> >>>>>>>>>> more..
> >>>>>>>>>>
> >>>>>>>>>> R
> >>>>>>>>>>
> >>>>>>>>>>
> >>>>>>>>>>>
> >>>>>>>>>>> /ptc
> >>>>>>>>>>>
> >>>>>>>>>>> At 11:44 AM 6/5/2002 -0500, Randall Stewart wrote:
> >>>>>>>>>>>> Brian/Qiaobing:
> >>>>>>>>>>>>
> >>>>>>>>>>>> I have read through this interesting debate. Qiaobing and
> >>>>>>>>>>>> I had a major discussion on this way back when it was
> >>>>>>>>>>>> considered U-SCTP. Renee Revis first pointed this out to me..
> >>>>>>>>>>>> but
> >>>>>>>>>>>> from a different persepective i.e. the receiver being able to
> >>>>>>>>>>>> efficently process stuck TSN's ...
> >>>>>>>>>>>>
> >>>>>>>>>>>> After a lengthy debate I agreed to not have the SSN/SID 
> >>>>>>>>>>>> listed
> >>>>>>>>>>>> in the FWD TSN... I did NOT think of the HOL blocking issue..
> >>>>>>>>>>>> or
> >>>>>>>>>>>> I would have used it as a issue (I was the one arguing for 
> >>>>>>>>>>>> the
> >>>>>>>>>>>> SSN/SID in the FWD-TSN :->).
> >>>>>>>>>>>>
> >>>>>>>>>>>> After watching this debate.. I think I have come to the
> >>>>>>>>>>>> conclusion
> >>>>>>>>>>>> that
> >>>>>>>>>>>> Brian is correct :)
> >>>>>>>>>>>>
> >>>>>>>>>>>> Michael R. is here... I will discuss it with him.. I want to
> >>>>>>>>>>>> get
> >>>>>>>>>>>> the other co-author's take on this one... If the others agree
> >>>>>>>>>>>> with
> >>>>>>>>>>>> Brian (and me :>) then I think we will put this in.... also I
> >>>>>>>>>>>> would
> >>>>>>>>>>>> like to hear what the rest of the WG feels on this???
> >>>>>>>>>>>>
> >>>>>>>>>>>>
> >>>>>>>>>>>> I think it is reasonable and not that much overhead to 
> >>>>>>>>>>>> include
> >>>>>>>>>>>> JUST
> >>>>>>>>>>>> the dropped streams NEXT expected SSN... You would be 
> >>>>>>>>>>>> adding 4
> >>>>>>>>>>>> bytes
> >>>>>>>>>>>> for EVERY stream dropping one or more TSN's... note that this
> >>>>>>>>>>>> does
> >>>>>>>>>>>> NOT mean 4 bytes per TSN but 4 bytes per where one or more
> >>>>>>>>>>>> TSN's
> >>>>>>>>>>>> were
> >>>>>>>>>>>> dropped..
> >>>>>>>>>>>>
> >>>>>>>>>>>>
> >>>>>>>>>>>> I think this is reasonable.. it improves processing AND 
> >>>>>>>>>>>> avoids
> >>>>>>>>>>>> HOL
> >>>>>>>>>>>> blocking..
> >>>>>>>>>>>>
> >>>>>>>>>>>>
> >>>>>>>>>>>> Michael T/Phil? what do you think... Michael R.. I will talk
> >>>>>>>>>>>> to you
> >>>>>>>>>>>> at
> >>>>>>>>>>>> the
> >>>>>>>>>>>> next class break..
> >>>>>>>>>>>>
> >>>>>>>>>>>>
> >>>>>>>>>>>>
> >>>>>>>>>>>> R
> >>>>>>>>>>>>
> >>>>>>>>>>>> --
> >>>>>>>>>>>> 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)
> >>>>>>>>>
> >>>>>>>>> --
> >>>>>>>>> Brian F. G. Bidulock    ¦ The reasonable man adapts himself to
> >>>>>>>>> the ¦
> >>>>>>>>> bidulock@openss7.org    ¦ world; the unreasonable one persists
> >>>>>>>>> in  ¦
> >>>>>>>>> http://www.openss7.org/ ¦ trying  to adapt the  world  to
> >>>>>>>>> himself. ¦
> >>>>>>>>>                         ¦ Therefore  all  progress  depends on
> >>>>>>>>> the ¦
> >>>>>>>>>                         ¦ unreasonable man. -- George Bernard
> >>>>>>>>> Shaw ¦
> >>>>>>>>
> >>>>>>>> --
> >>>>>>>> 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
> >>>>>>>>
> >>>>>>>>
> Michael.Tuexen@micmac.franken.de
> >>>>>
> >>>>> --
> >>>>> Brian F. G. Bidulock    ¦ The reasonable man adapts himself to the ¦
> >>>>> bidulock@openss7.org    ¦ world; the unreasonable one persists in  ¦
> >>>>> http://www.openss7.org/ ¦ trying  to adapt the  world  to himself. ¦
> >>>>>                         ¦ Therefore  all  progress  depends on the ¦
> >>>>>                         ¦ unreasonable man. -- George Bernard Shaw ¦
> >>>
> >>> --
> >>> Brian F. G. Bidulock    ¦ The reasonable man adapts himself to the ¦
> >>> bidulock@openss7.org    ¦ world; the unreasonable one persists in  ¦
> >>> http://www.openss7.org/ ¦ trying  to adapt the  world  to himself. ¦
> >>>                         ¦ Therefore  all  progress  depends on the ¦
> >>>                         ¦ unreasonable man. -- George Bernard Shaw ¦
> >>>
> >
> > --
> > Brian F. G. Bidulock    ¦ The reasonable man adapts himself to the ¦
> > bidulock@openss7.org    ¦ world; the unreasonable one persists in  ¦
> > http://www.openss7.org/ ¦ trying  to adapt the  world  to himself. ¦
> >                         ¦ Therefore  all  progress  depends on the ¦
> >                         ¦ unreasonable man. -- George Bernard Shaw ¦
> >

-- 
Brian F. G. Bidulock    ¦ The reasonable man adapts himself to the ¦
bidulock@openss7.org    ¦ world; the unreasonable one persists in  ¦
http://www.openss7.org/ ¦ trying  to adapt the  world  to himself. ¦
                        ¦ Therefore  all  progress  depends on the ¦
                        ¦ unreasonable man. -- George Bernard Shaw ¦

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



From daemon@optimus.ietf.org  Sat Jun  8 17:10:05 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 RAA11884
	for <tsvwg-archive@odin.ietf.org>; Sat, 8 Jun 2002 17:10:05 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id RAA12580
	for tsvwg-archive@odin.ietf.org; Sat, 8 Jun 2002 17:10:38 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id QAA10902;
	Sat, 8 Jun 2002 16:32:16 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id QAA10817
	for <tsvwg@optimus.ietf.org>; Sat, 8 Jun 2002 16:32:12 -0400 (EDT)
Received: from ilsa.franken.de (postfix@ilsa.franken.de [193.175.24.42])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id QAA11397
	for <tsvwg@ietf.org>; Sat, 8 Jun 2002 16:31:37 -0400 (EDT)
Received: from localhost (pD9E4F7CF.dip.t-dialin.net [217.228.247.207])
	by ilsa.franken.de (Postfix) with ESMTP
	id 06165245CF; Sat,  8 Jun 2002 22:31:50 +0200 (CEST)
Date: Sat, 8 Jun 2002 22:31:33 +0200
Subject: Re: [Tsvwg] More on PRSCTP-00
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Mime-Version: 1.0 (Apple Message framework v482)
Cc: Qiaobing Xie <Qiaobing.Xie@motorola.com>,
        Randall Stewart <randall@stewart.chicago.il.us>,
        Phillip Conrad <conrad@acm.org>,
        "Armando L. Caro Jr." <acaro@mail.eecis.udel.edu>, tsvwg@ietf.org,
        "Michael A. Ramalho" <mramalho@cisco.com>
To: bidulock@openss7.org
From: Michael Tuexen <Michael.Tuexen@lurchi.franken.de>
In-Reply-To: <20020608142612.D20649@openss7.org>
Message-Id: <B8DDC5CE-7B1E-11D6-9BA0-0030654C1AB6@lurchi.franken.de>
X-Mailer: Apple Mail (2.482)
Content-Transfer-Encoding: 8bit
X-MIME-Autoconverted: from quoted-printable to 8bit by optimus.ietf.org id QAA10818
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

Brian,

see my comments below.

Best regards
Michael

On Saturday, June 8, 2002, at 09:26 PM, Brian F. G. Bidulock wrote:

> Michael,
>
> On Sat, 08 Jun 2002, Michael Tuexen wrote:
>
>> Brian, Qiaobing,
>>
>> I'm a bit confused by the answer, because there are some SACKs less 
>> than
>> in the previous email.
>>
>> I will repeat the scenario:
>>
>> Consider
>> packet1 = TSN 101-109; SID=1-9; SSN=1;
>> packet2 = TSN 110-119; SID=1-9; SSN=2;
>
> I think you mean TSN 110-118 here.
You are correct, it is a typo.
>
>> packet3 = TSN 119;     SID=0;
>> packet4 = TSN 120-128; SID=1-9; SSN=3;
>>
>> It seems that you are now considering the szenario:
>> 1. packet1 is transmitted and received
>> 2. packet2 is transmitted and lost on the wire
>> 3. packet3 is transmitted and lost on the wire
>> 4. timed reliability runs off for TSN 110-119
>> 5. packet4 is transmitted and received. The user data can not be
>> delivered.
>> 6. A SACK for packet 1 is received.
>> 7. A FWD TSN chunk is sent
>>
>> This happens only if the timed reliability is shorter than the round
>> trip time.
>
> No that is not necessary.  The SACK (containing no gaps) may be
> a "delayed" sack delayed by 500ms so timed reliability only needs to
> be shorter than the current (experienced by the SACK) RTT + 500ms.
> Nevertheless, it is easy for a fixed "timed reliability" to be less
> than the current RTT+500ms at any given point, particularly when
> packets are being lost in the network due to congestion.
Yes, the SACK can be delayed.
>
>> But we can assume this.
>
> The retransmission timer may be as high as 60 seconds at the moment,
> so it is quite conceivable that "timed reliability" is less than the
> sender's RTO.  In fact, for "timed reliability" to work as it is
> expected, the timed reliability interval MUST be less than the product
> of Assoc.Max.Retrans and Rto.Max, otherwise, packets will never expire.
> (The association will fail before they expire.)
>
>>
>> So what will happen:
>> 8. The retransmission timer will run off for packet3.
>> 9. packet3 gets retransmitted, and the user data of packet3 and packet4
>> can be delivered.
>>
>> I agree with you that there is HOL, but it is the same as there would 
>> be
>> if you
>> would not use PR-SCTP. So PR-SCTP, as it is currently being defined 
>> does
>> not perform
>> better than basic SCTP in terms of user-to-user message delay. It does
>> perform better in terms
>> of buffer and bandwidth usage.
>
> I disagree with that analysis.  Normal SCTP only has to wait for
> retransmission of the SSN 2 packet, whereas PR-SCTP in this scenario
> has to wait for retransmission of all the TSN gaps subsequent to
> the FORWARD TSN.  These (e.g. TSN 119) might have 16k worth of data
> in it making it subject to many retransmissions (the network is already
> marginally in congestion).  The SSN 2 packet might contain 256 bytes
> meaning that retransmission of TSNs 110-118 could be successful when
> TSN 119 is going to be unsuccessful for some time.
>
> It is NOT the situation in regular SCTP that a stream must block
> awaiting the retransmission of a TSN from a different stream.  For
> PR-SCTP, streams 1 through 9 are blocked awaiting the retransmission
> of a data chunk on stream 0.  This is the unacceptable HOL blocking.
> It is inter-stream, not intra-stream.
Hmmm. I see. The user data of packet4 can be delivered to the upper layer
after the FWD-TSN (something like a retransmission of packet2) and 
packet3
have been received. In normal SCTP only packet2 needs to be received.
Since the user-to-user delay of PR-SCTP should not be greater than for
SCTP, I think we should put the SSN/SID info in the FWD-TSN chunk.
But only for the affected streams. Qiaobing, what do you think?
Best regards
Michael
>
> --brian
>
>>
>> So I think using SSN info in the FWD TSN chunk would optimize PR-SCTP
>> also in terms of
>> user-to-user delay. Qiaobing, could you agree on this?
>>
>> Best regards
>> Michael
>> On Saturday, June 8, 2002, at 07:16 PM, Brian F. G. Bidulock wrote:
>>
>>> Michael,
>>>
>>> On Sat, 08 Jun 2002, Michael Tuexen wrote:
>>>
>>>> Brian,
>>>>
>>>> see my questions below.
>>>>
>>>> Best regards
>>>> Michael
>>>>
>>>> On Saturday, June 8, 2002, at 02:49 AM, Brian F. G. Bidulock wrote:
>>>>
>>>>> Qiaobing,
>>>>>
>>>>> On Fri, 07 Jun 2002, Qiaobing Xie wrote:
>>>>>
>>>>>> "Brian F. G. Bidulock" wrote:
>>>>>>>
>>>>>>> Michael, Qiaobing,
>>>>>>>
>>>>>>> Here is the sender side of the example.  It is quite reasonable
>>>>>>> that this could occur.  The receiver is blocked on streams 1
>>>>>>> through 9 until retransmission of TSN 119 is successful.  This
>>>>>>> is the sender side of the same example.
>>>>>>>
>>>>>>>     TSN 101  SID 1   SSN 1   |
>>>>>>>     TSN 102  SID 2   SSN 1   |
>>>>>>>     TSN 103  SID 3   SSN 1   |
>>>>>>>     TSN 104  SID 4   SSN 1   |   SCTP Packet
>>>>>>>     TSN 105  SID 5   SSN 1   +--------------------->
>>>>>>>     TSN 106  SID 6   SSN 1   |
>>>>>>>     TSN 107  SID 7   SSN 1   |
>>>>>>>     TSN 108  SID 8   SSN 1   |
>>>>>>>     TSN 109  SID 9   SSN 1   |
>>>>>>>
>>>>>>>     TSN 110  SID 1   SSN 2   |
>>>>>>>     TSN 111  SID 2   SSN 2   |
>>>>>>>     TSN 112  SID 3   SSN 2   |
>>>>>>>     TSN 113  SID 4   SSN 2   |   SCTP Packet
>>>>>>>     TSN 114  SID 5   SSN 2   +----------X
>>>>>>>     TSN 115  SID 6   SSN 2   |         Lost
>>>>>>>     TSN 116  SID 7   SSN 2   |
>>>>>>>     TSN 117  SID 8   SSN 2   |
>>>>>>>     TSN 118  SID 9   SSN 2   |
>>>>>>>                                  SCTP Packet
>>>>>>>     TSN 119  SID 0   SSN X   |----------X
>>>>>>>                                        Lost
>>>>>>>
>>>>>>>     (Timed reliability expiry on TSNs 110 to 118)
>>>>>
>>>>>         <---- SACK CUM ACK 109
>>>> Why is this SACK be send
>>> To acknowledge the first SCTP Packet
>>>
>>>>>
>>>>>>>
>>>>>>>     FORWARD TSN 118          |--------------------->
>>>> This now means that the retransmission timer runs off. So bundle a
>>>> FWD-TSN and DATA.
>>>
>>> No it does not: according to rule C3 a SACK has been received.  The
>>> retrans timer has not expired. We are forced to send FORWARD TSN
>>> just because a sack has arrived.
>>>
>>>> So I really do not understand this example.
>>>
>>> You have no imagination: remove the first two SACKs.
>>> Then it is completely possible.
>>>
>>> --brian
>>>
>>>
>>>>>>
>>>>>> Why TSN119 is not re-sent at this exactly same moment? As Michael
>>>>>> and I
>>>>>> both noted that, in PR-SCTP, Fwd TSN can only be triggered due to
>>>>>> either
>>>>>> a "RTX Timer expiration or fast retransmission." Could you fill in
>>>>>> the
>>>>>> details that lead to TSN119 not being re-sent at this moment?
>>>>>
>>>>> With the SACKs shown, the FORWARD TSN "MUST" be sent according to
>>>>> Rule C3.  Retransmission of TSN 119 has nothing to do with it.
>>>>>
>>>>> --brian
>>>>>
>>>>>>
>>>>>>>
>>>>>>>     TSN 120  SID 1   SSN 3   |
>>>>>>>     TSN 121  SID 2   SSN 3   |
>>>>>>>     TSN 122  SID 3   SSN 3   |
>>>>>>>     TSN 123  SID 4   SSN 3   |   SCTP Packet
>>>>>>>     TSN 124  SID 5   SSN 3   +--------------------->
>>>>>>>     TSN 125  SID 6   SSN 3   |
>>>>>>>     TSN 126  SID 7   SSN 3   |
>>>>>>>     TSN 127  SID 8   SSN 3   |
>>>>>>>     TSN 128  SID 9   SSN 3   |
>>>>>>>
>>>>>>>     (Receiver is now blocked on streams 1 through 9.)
>>>>>>>
>>>>>>>     (Some time later, retrans timeout on TSN 119)
>>>>>>>
>>>>>>>                                  SCTP Packet
>>>>>>>     TSN 119  SID 0   SSN X   |--------------------->
>>>>>>>
>>>>>>> Q.E.D.
>>>>>>>
>>>>>>> --brian
>>>>>>>
>>>>>>> On Fri, 07 Jun 2002, Qiaobing Xie wrote:
>>>>>>>
>>>>>>>> Michael,
>>>>>>>>
>>>>>>>> Right on! You summarized my previous point beautifully. I start 
>>>>>>>> to
>>>>>>>> realize that lately my arguments tend to become lengthy and off 
>>>>>>>> the
>>>>>>>> point... maybe a sign of getting old :<
>>>>>>>>
>>>>>>>> So far I kind of buy Randy's point better that "adding some bits 
>>>>>>>> in
>>>>>>>> exchange for ease of implementation", but still yet to be 
>>>>>>>> convinced
>>>>>>>> about the HOL blocking argument made by Brian.
>>>>>>>>
>>>>>>>> I've asked for an "all-considered" example of the HOL blocking, 
>>>>>>>> but
>>>>>>>> haven't seen one yet.
>>>>>>>>
>>>>>>>> regards,
>>>>>>>> -Qiaobing
>>>>>>>>
>>>>>>>> Michael Tuexen wrote:
>>>>>>>>>
>>>>>>>>> Dear all,
>>>>>>>>>
>>>>>>>>> reading the thread I would like to make some remarks:
>>>>>>>>>
>>>>>>>>> 1. I agree that using SSN in the FWD chunk avoids HOL (which is
>>>>>>>>> good).
>>>>>>>>> 2. Brians example show how HOL can occur if no SSN info in the 
>>>>>>>>> FWD
>>>>>>>>> chunk
>>>>>>>>>     is used. He only looks at the receiver.
>>>>>>>>>
>>>>>>>>> Brians example are always based on the assumption that there 
>>>>>>>>> is a
>>>>>>>>> GAP
>>>>>>>>> in the TSN space at the receiver side and the receiver gets a 
>>>>>>>>> FWD
>>>>>>>>> TSN
>>>>>>>>> which does not fill the whole gap. Then the HOL happens at the
>>>>>>>>> receiver.
>>>>>>>>>
>>>>>>>>> I'm just wondering how often this will happen. What are the
>>>>>>>>> reasons
>>>>>>>>> that
>>>>>>>>> a FWD chunk will be send: RTX Timer expiration or fast
>>>>>>>>> retransmission.
>>>>>>>>> In both cases I would send a FWD chunk and bundle the remaining
>>>>>>>>> DATA
>>>>>>>>> chunks
>>>>>>>>> (if possible). So there is a good chance that the whole gap is
>>>>>>>>> filled by
>>>>>>>>> the FWD chunk and the retransmitted DATA chunks.
>>>>>>>>> I'm not sure how likely Brians scenario is taking the sender 
>>>>>>>>> into
>>>>>>>>> account.
>>>>>>>>>
>>>>>>>>> So I think one has to take the senders behavior into account 
>>>>>>>>> when
>>>>>>>>> discussing
>>>>>>>>> this issue.
>>>>>>>>>
>>>>>>>>> Best regards
>>>>>>>>> Michael
>>>>>>>>>
>>>>>>>>> On Thursday, June 6, 2002, at 01:43 AM, Randall Stewart wrote:
>>>>>>>>>
>>>>>>>>>> "Brian F. G. Bidulock" wrote:
>>>>>>>>>>>
>>>>>>>>>>> Randall,
>>>>>>>>>>>
>>>>>>>>>>> On Wed, 05 Jun 2002, Randall Stewart wrote:
>>>>>>>>>>>
>>>>>>>>>>>> A few comments..
>>>>>>>>>>>>
>>>>>>>>>>>> Phillip Conrad wrote:
>>>>>>>>>>>>>
>>>>>>>>>>>>> Dear Colleagues:
>>>>>>>>>>>>>       I agree with Randy that Brian has raised an important
>>>>>>>>>>>>> issue
>>>>>>>>>>>>> for us to
>>>>>>>>>>>>> consider, and I think has explained it quite well with a 
>>>>>>>>>>>>> very
>>>>>>>>>>>>> clear,
>>>>>>>>>>>>> illustrative example.
>>>>>>>>>>>>>
>>>>>>>>>>>>> The question is one of a tradeoff between two designs:
>>>>>>>>>>>>>
>>>>>>>>>>>>> (A) current design: simpler, fewer bits on the wire, but may
>>>>>>>>>>>>> (re)-introduce
>>>>>>>>>>>>> an HOL blocking problem
>>>>>>>>>>>>
>>>>>>>>>>>> I must disagree with your "it is simpler" ... it is NOT. The
>>>>>>>>>>>> question
>>>>>>>>>>>> is
>>>>>>>>>>>> where is the burden. Having implemented pr-sctp in the kernel
>>>>>>>>>>>> of
>>>>>>>>>>>> freebsd
>>>>>>>>>>>> NOT having the stream number/SSN in the FWD-TSN means I must
>>>>>>>>>>>> keep
>>>>>>>>>>>> an extra set of linked lists and such to track any stream 
>>>>>>>>>>>> that
>>>>>>>>>>>> has
>>>>>>>>>>>> pending stuck data. When a fwd-tsn arrives I must HUNT all of
>>>>>>>>>>>> these
>>>>>>>>>>>> streams and look at each inbound queue.
>>>>>>>>>>>>
>>>>>>>>>>>> The counter to this is the sender (in Brian and my earlier
>>>>>>>>>>>> proposal
>>>>>>>>>>>> that was to Q only) must track this info to send with the
>>>>>>>>>>>> FWD-TSN.
>>>>>>>>>>>> Actually
>>>>>>>>>>>> less of a burden IMHO...
>>>>>>>>>>>
>>>>>>>>>>> I agree.  I discovered this situation while implementing 
>>>>>>>>>>> PR-SCTP
>>>>>>>>>>> in the
>>>>>>>>>>> Linux Kernel.  The normal receiver can rely on the fact that 
>>>>>>>>>>> the
>>>>>>>>>>> next
>>>>>>>>>>> expected SSN is the last fully received SSN plus 1.  PR-SCTP
>>>>>>>>>>> without
>>>>>>>>>>> a forward SSN causes the receive to have to update this 
>>>>>>>>>>> expected
>>>>>>>>>>> value
>>>>>>>>>>> with the next SSN in the reorder queue whenever there are no 
>>>>>>>>>>> TSN
>>>>>>>>>>> gaps.
>>>>>>>>>>> This even might require that when TSN gaps are filled that the
>>>>>>>>>>> receiver
>>>>>>>>>>> walk all inbound stream structures to update the next expected
>>>>>>>>>>> SSN.
>>>>>>>>>>> In an independent implementation, I have arrived at the same
>>>>>>>>>>> conclusion:
>>>>>>>>>>> that the reciever has more work to do for FORWARD TSN 
>>>>>>>>>>> without a
>>>>>>>>>>> forward
>>>>>>>>>>> SSN.
>>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> Yeah.. I basically had to make a special linked list set that 
>>>>>>>>>> the
>>>>>>>>>> stream queues could be placed on.. a bit more efficient but a
>>>>>>>>>> reall hassel to maintian a "wheel" (as I call it) of streams 
>>>>>>>>>> with
>>>>>>>>>> pending data on it...
>>>>>>>>>>
>>>>>>>>>>> In fact the receiver has this work to do if it has ever
>>>>>>>>>>> received a
>>>>>>>>>>> FORWARD TSN.  This is because it is not possible for the
>>>>>>>>>>> receiver
>>>>>>>>>>> to
>>>>>>>>>>> know a priori which streams could be affected by this 
>>>>>>>>>>> scenairo.
>>>>>>>>>>> Therefore,
>>>>>>>>>>> the load on the receiver is increased even when this case does
>>>>>>>>>>> not
>>>>>>>>>>> occur.
>>>>>>>>>>> It is the mere possibility of its occurence which introduces 
>>>>>>>>>>> the
>>>>>>>>>>> load.
>>>>>>>>>>> (Each time gaps close, the next expected SSN for all streams
>>>>>>>>>>> need
>>>>>>>>>>> to be
>>>>>>>>>>> re-examined whether there are SSN gaps or not.)
>>>>>>>>>>
>>>>>>>>>> Exactly.. I constantly must place stream q;s on and off the
>>>>>>>>>> :"wheel" no
>>>>>>>>>> matter if a fwd-tsn is EVER received...
>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>> On the sender side, the sender must always track the last sent
>>>>>>>>>>> SSN and
>>>>>>>>>>> assign the next SSN (++SSN).  Indeed the sender when
>>>>>>>>>>> generating a
>>>>>>>>>>> FORWARD
>>>>>>>>>>> TSN chunk would need to traverse outbound stream structures 
>>>>>>>>>>> with
>>>>>>>>>>> dropped
>>>>>>>>>>> TSNs to formulate Forward TSNs, but this is likely to be a far
>>>>>>>>>>> shorter
>>>>>>>>>>> list than at the receiver and non-PR streams would never need 
>>>>>>>>>>> to
>>>>>>>>>>> be
>>>>>>>>>>> involved.
>>>>>>>>>>
>>>>>>>>>> And it must continue to KEEP this information until it is sure
>>>>>>>>>> the
>>>>>>>>>> peer has received and moved the cum-ack forward..
>>>>>>>>>>
>>>>>>>>>> What I end up doing on the sender side is I hold just the
>>>>>>>>>> bookkeeping
>>>>>>>>>> structures of the TSN (stream seq/stream no/tsn ... etc) but I
>>>>>>>>>> drop the mbuf holding the data so I gain a bit of help in mbuf
>>>>>>>>>> utilization.
>>>>>>>>>>
>>>>>>>>>> But I can't get rid of the chunk structure UNTIL the cum-ack 
>>>>>>>>>> says
>>>>>>>>>> he
>>>>>>>>>> got my FWD-TSN and accepted it...
>>>>>>>>>>
>>>>>>>>>> So there is NO additional info to save on the sender side and
>>>>>>>>>> only
>>>>>>>>>> do I save what I must... on the receiver side I have a LOT of
>>>>>>>>>> extra
>>>>>>>>>> bookkeeping....
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> In fact with your confirmation I would claim just the opposite 
>>>>>>>>>> of
>>>>>>>>>> Phil's statement... that doing it the way it is NOW documented
>>>>>>>>>> takes
>>>>>>>>>> MORE overhead.. only by adding the SSN/SID can you reduce 
>>>>>>>>>> this..
>>>>>>>>>>
>>>>>>>>>> By the way Renee also pointed out the extra overhead for the 
>>>>>>>>>> IOS
>>>>>>>>>> implementation as well :>... so this makes three different
>>>>>>>>>> implementations
>>>>>>>>>> NOT 2....
>>>>>>>>>>
>>>>>>>>>> I plan on adding this to the next pr-sctp unless Michael Tuexen
>>>>>>>>>> violently
>>>>>>>>>> agrees with Qiaobing ...
>>>>>>>>>>
>>>>>>>>>> I still await Michaels answer????
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>> --brian
>>>>>>>>>>>
>>>>>>>>>>>>>
>>>>>>>>>>>>> (B) alternative design: more complex, may require larger
>>>>>>>>>>>>> Forward TSN
>>>>>>>>>>>>> chunks, but avoids re-introducing an HOL blocking problem, 
>>>>>>>>>>>>> and
>>>>>>>>>>>>> thus
>>>>>>>>>>>>> improves performance in some situations.
>>>>>>>>>>>>
>>>>>>>>>>>>
>>>>>>>>>>>> see above it is NOT more complex...
>>>>>>>>>>>>
>>>>>>>>>>>>>
>>>>>>>>>>>>> The question of whether to choose A or B depends on whether
>>>>>>>>>>>>> you
>>>>>>>>>>>>> believe
>>>>>>>>>>>>> that the situation Brian describes is a "corner case" or 
>>>>>>>>>>>>> not.
>>>>>>>>>>>>> So
>>>>>>>>>>>>> I've
>>>>>>>>>>>>> walked through Brian's example, and have determined that I
>>>>>>>>>>>>> don't
>>>>>>>>>>>>> think it
>>>>>>>>>>>>> is a corner case at all.   So, I'm inclined to agree with
>>>>>>>>>>>>> Randy
>>>>>>>>>>>>> and
>>>>>>>>>>>>> Brian
>>>>>>>>>>>>> that we should consider a change.
>>>>>>>>>>>>>
>>>>>>>>>>>>> But I'd like to invite others to walk through my line of
>>>>>>>>>>>>> reasoning
>>>>>>>>>>>>> to test
>>>>>>>>>>>>> whether I've reasoned this correctly.
>>>>>>>>>>>>
>>>>>>>>>>>>
>>>>>>>>>>>> I agree it is NOT a corner case..
>>>>>>>>>>>>
>>>>>>>>>>>>>
>>>>>>>>>>>>> My rationale that Brian's example is NOT a corner case:
>>>>>>>>>>>>> ========================================
>>>>>>>>>>>>>
>>>>>>>>>>>>> In Brian's example, messages are being sent round-robin on a
>>>>>>>>>>>>> number
>>>>>>>>>>>>> of
>>>>>>>>>>>>> streams.  This seems like a fairly reasonable use of SCTP
>>>>>>>>>>>>> streams,
>>>>>>>>>>>>> do that
>>>>>>>>>>>>> aspect does not strike me as unusual.
>>>>>>>>>>>>
>>>>>>>>>>>> I agree :>
>>>>>>>>>>>>
>>>>>>>>>>>>>
>>>>>>>>>>>>> Now for the next part of this explanation, I'm going to have
>>>>>>>>>>>>> to
>>>>>>>>>>>>> infer
>>>>>>>>>>>>> certain things, so, Brian, please correct me if I infer
>>>>>>>>>>>>> wrongly.
>>>>>>>>>>>>> *One*
>>>>>>>>>>>>> possibility for the sequence of events leading to the
>>>>>>>>>>>>> situation Brian describes is that there is a burst loss of
>>>>>>>>>>>>> data
>>>>>>>>>>>>> chunks
>>>>>>>>>>>>> TSN110 through TSN119 (ten chunks, but possibly less than 
>>>>>>>>>>>>> ten
>>>>>>>>>>>>> packets..
>>>>>>>>>>>>> perhaps even only ONE packet if the chunks are small enough 
>>>>>>>>>>>>> to
>>>>>>>>>>>>> be
>>>>>>>>>>>>> bundled!)... and then data chunks TSN120 through TSN128 make
>>>>>>>>>>>>> it
>>>>>>>>>>>>> through
>>>>>>>>>>>>> OK.   This does not strike me as particularly unusual; a 
>>>>>>>>>>>>> loss
>>>>>>>>>>>>> of a
>>>>>>>>>>>>> single
>>>>>>>>>>>>> packet followed by a successful packet is sufficient to 
>>>>>>>>>>>>> cause
>>>>>>>>>>>>> this
>>>>>>>>>>>>> situation.
>>>>>>>>>>>>>
>>>>>>>>>>>>
>>>>>>>>>>>> Or the other possiblility would be a set of bundled chunks in
>>>>>>>>>>>> one
>>>>>>>>>>>> packet
>>>>>>>>>>>> being lost.. either way is the same result I would think..
>>>>>>>>>>>>
>>>>>>>>>>>>> The next thing that happens is that TSNs 110 through 118 are
>>>>>>>>>>>>> "dropped" in
>>>>>>>>>>>>> the technical sense of that word; see the defn in the 
>>>>>>>>>>>>> current
>>>>>>>>>>>>> draft.  This
>>>>>>>>>>>>> means that the sender gives up on sending them, and issues a
>>>>>>>>>>>>> forward
>>>>>>>>>>>>> TSN.   However, no Forward TSN is issued for TSN119.  This 
>>>>>>>>>>>>> may
>>>>>>>>>>>>> be
>>>>>>>>>>>>> because
>>>>>>>>>>>>> the messages on stream 9 are treated as Reliable.   This is
>>>>>>>>>>>>> NOT
>>>>>>>>>>>>> an
>>>>>>>>>>>>> unusual
>>>>>>>>>>>>> situation; it is perfectly reasonably to assume that an
>>>>>>>>>>>>> application
>>>>>>>>>>>>> might
>>>>>>>>>>>>> have "k" streams with data that can be dropped, and "n - k"
>>>>>>>>>>>>> streams
>>>>>>>>>>>>> of data
>>>>>>>>>>>>> that must be treated as reliable.  In the case of Brian's
>>>>>>>>>>>>> example, n
>>>>>>>>>>>>> is 10,
>>>>>>>>>>>>> and k is 9, but I think that his example will still produce
>>>>>>>>>>>>> similar
>>>>>>>>>>>>> results
>>>>>>>>>>>>> even for other values of "k" and "n".
>>>>>>>>>>>>>
>>>>>>>>>>>>
>>>>>>>>>>>> Note.. streams themselves are not reliable or partially
>>>>>>>>>>>> reliable. With
>>>>>>>>>>>> the new draft it is the message itself which is sent with the
>>>>>>>>>>>> reliablilty
>>>>>>>>>>>> set to "fully reliable". One possilbe implementation would be
>>>>>>>>>>>> to
>>>>>>>>>>>> make
>>>>>>>>>>>> things correlate to a particular stream.. but this is just an
>>>>>>>>>>>> implementation
>>>>>>>>>>>> option..
>>>>>>>>>>>>
>>>>>>>>>>>>> Bottom line:
>>>>>>>>>>>>> ========
>>>>>>>>>>>>>
>>>>>>>>>>>>> I don't think its a corner case at all.   We have all the PR
>>>>>>>>>>>>> streams
>>>>>>>>>>>>> potentially being HOL blocked by loss of a single packet.  I
>>>>>>>>>>>>> think
>>>>>>>>>>>>> we need
>>>>>>>>>>>>> to fix this.
>>>>>>>>>>>>
>>>>>>>>>>>> I agree..
>>>>>>>>>>>>
>>>>>>>>>>>>>
>>>>>>>>>>>>> The solution of forward SSNs is one viable solution I think.
>>>>>>>>>>>>> There
>>>>>>>>>>>>> may be
>>>>>>>>>>>>> others, but that's as good as any for right now I think.
>>>>>>>>>>>>> There may
>>>>>>>>>>>>> be
>>>>>>>>>>>>> another solution as well, but let me do some writing and
>>>>>>>>>>>>> thinking
>>>>>>>>>>>>> about it
>>>>>>>>>>>>> off line before I muddy the waters with it here... just to
>>>>>>>>>>>>> make
>>>>>>>>>>>>> sure
>>>>>>>>>>>>> it is
>>>>>>>>>>>>> a viable alternative.
>>>>>>>>>>>>
>>>>>>>>>>>> agreed
>>>>>>>>>>>>
>>>>>>>>>>>>
>>>>>>>>>>>>>
>>>>>>>>>>>>> Randy... I'm willing to make the text changes and send a
>>>>>>>>>>>>> proposed
>>>>>>>>>>>>> version
>>>>>>>>>>>>> around to the co-authors for comment if there is consensus 
>>>>>>>>>>>>> to
>>>>>>>>>>>>> move
>>>>>>>>>>>>> in that
>>>>>>>>>>>>> direction.   Let me know.
>>>>>>>>>>>>>
>>>>>>>>>>>>
>>>>>>>>>>>> I will get it out this weekend if I may.. there are some
>>>>>>>>>>>> interesting
>>>>>>>>>>>> things that I have thought about that must be covered... I 
>>>>>>>>>>>> had
>>>>>>>>>>>> something
>>>>>>>>>>>> like this in an old u-sctp draft (during Q and I's debate
>>>>>>>>>>>> earlier :>)
>>>>>>>>>>>>
>>>>>>>>>>>>
>>>>>>>>>>>>
>>>>>>>>>>>>>
>>>>>>>>>>>>> (Side note: I was worried that the "technical definition" of
>>>>>>>>>>>>> the word
>>>>>>>>>>>>> "dropped" would cause confusion, and it already has.   
>>>>>>>>>>>>> Perhaps
>>>>>>>>>>>>> should use
>>>>>>>>>>>>> another word in the draft to describe the operation where 
>>>>>>>>>>>>> the
>>>>>>>>>>>>> sender
>>>>>>>>>>>>> decides to no longer transmit and/or retransmit a specific
>>>>>>>>>>>>> data
>>>>>>>>>>>>> chunk.  How
>>>>>>>>>>>>> about the term "abandoned" in place of "dropped"?)
>>>>>>>>>>>>
>>>>>>>>>>>>
>>>>>>>>>>>> Sounds good..
>>>>>>>>>>>>
>>>>>>>>>>>> note, Michael R also agrees .. or seems to .. we will talk
>>>>>>>>>>>> more..
>>>>>>>>>>>>
>>>>>>>>>>>> R
>>>>>>>>>>>>
>>>>>>>>>>>>
>>>>>>>>>>>>>
>>>>>>>>>>>>> /ptc
>>>>>>>>>>>>>
>>>>>>>>>>>>> At 11:44 AM 6/5/2002 -0500, Randall Stewart wrote:
>>>>>>>>>>>>>> Brian/Qiaobing:
>>>>>>>>>>>>>>
>>>>>>>>>>>>>> I have read through this interesting debate. Qiaobing and
>>>>>>>>>>>>>> I had a major discussion on this way back when it was
>>>>>>>>>>>>>> considered U-SCTP. Renee Revis first pointed this out to 
>>>>>>>>>>>>>> me..
>>>>>>>>>>>>>> but
>>>>>>>>>>>>>> from a different persepective i.e. the receiver being able 
>>>>>>>>>>>>>> to
>>>>>>>>>>>>>> efficently process stuck TSN's ...
>>>>>>>>>>>>>>
>>>>>>>>>>>>>> After a lengthy debate I agreed to not have the SSN/SID
>>>>>>>>>>>>>> listed
>>>>>>>>>>>>>> in the FWD TSN... I did NOT think of the HOL blocking 
>>>>>>>>>>>>>> issue..
>>>>>>>>>>>>>> or
>>>>>>>>>>>>>> I would have used it as a issue (I was the one arguing for
>>>>>>>>>>>>>> the
>>>>>>>>>>>>>> SSN/SID in the FWD-TSN :->).
>>>>>>>>>>>>>>
>>>>>>>>>>>>>> After watching this debate.. I think I have come to the
>>>>>>>>>>>>>> conclusion
>>>>>>>>>>>>>> that
>>>>>>>>>>>>>> Brian is correct :)
>>>>>>>>>>>>>>
>>>>>>>>>>>>>> Michael R. is here... I will discuss it with him.. I want 
>>>>>>>>>>>>>> to
>>>>>>>>>>>>>> get
>>>>>>>>>>>>>> the other co-author's take on this one... If the others 
>>>>>>>>>>>>>> agree
>>>>>>>>>>>>>> with
>>>>>>>>>>>>>> Brian (and me :>) then I think we will put this in.... 
>>>>>>>>>>>>>> also I
>>>>>>>>>>>>>> would
>>>>>>>>>>>>>> like to hear what the rest of the WG feels on this???
>>>>>>>>>>>>>>
>>>>>>>>>>>>>>
>>>>>>>>>>>>>> I think it is reasonable and not that much overhead to
>>>>>>>>>>>>>> include
>>>>>>>>>>>>>> JUST
>>>>>>>>>>>>>> the dropped streams NEXT expected SSN... You would be
>>>>>>>>>>>>>> adding 4
>>>>>>>>>>>>>> bytes
>>>>>>>>>>>>>> for EVERY stream dropping one or more TSN's... note that 
>>>>>>>>>>>>>> this
>>>>>>>>>>>>>> does
>>>>>>>>>>>>>> NOT mean 4 bytes per TSN but 4 bytes per where one or more
>>>>>>>>>>>>>> TSN's
>>>>>>>>>>>>>> were
>>>>>>>>>>>>>> dropped..
>>>>>>>>>>>>>>
>>>>>>>>>>>>>>
>>>>>>>>>>>>>> I think this is reasonable.. it improves processing AND
>>>>>>>>>>>>>> avoids
>>>>>>>>>>>>>> HOL
>>>>>>>>>>>>>> blocking..
>>>>>>>>>>>>>>
>>>>>>>>>>>>>>
>>>>>>>>>>>>>> Michael T/Phil? what do you think... Michael R.. I will 
>>>>>>>>>>>>>> talk
>>>>>>>>>>>>>> to you
>>>>>>>>>>>>>> at
>>>>>>>>>>>>>> the
>>>>>>>>>>>>>> next class break..
>>>>>>>>>>>>>>
>>>>>>>>>>>>>>
>>>>>>>>>>>>>>
>>>>>>>>>>>>>> R
>>>>>>>>>>>>>>
>>>>>>>>>>>>>> --
>>>>>>>>>>>>>> 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)
>>>>>>>>>>>
>>>>>>>>>>> --
>>>>>>>>>>> Brian F. G. Bidulock    ¦ The reasonable man adapts himself to
>>>>>>>>>>> the ¦
>>>>>>>>>>> bidulock@openss7.org    ¦ world; the unreasonable one persists
>>>>>>>>>>> in  ¦
>>>>>>>>>>> http://www.openss7.org/ ¦ trying  to adapt the  world  to
>>>>>>>>>>> himself. ¦
>>>>>>>>>>>                         ¦ Therefore  all  progress  depends on
>>>>>>>>>>> the ¦
>>>>>>>>>>>                         ¦ unreasonable man. -- George Bernard
>>>>>>>>>>> Shaw ¦
>>>>>>>>>>
>>>>>>>>>> --
>>>>>>>>>> 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
>>>>>>>>>>
>>>>>>>>>>
Michael.Tuexen@micmac.franken.de
>>>>>>>
>>>>>>> --
>>>>>>> Brian F. G. Bidulock    ¦ The reasonable man adapts himself to 
>>>>>>> the ¦
>>>>>>> bidulock@openss7.org    ¦ world; the unreasonable one persists 
>>>>>>> in  ¦
>>>>>>> http://www.openss7.org/ ¦ trying  to adapt the  world  to 
>>>>>>> himself. ¦
>>>>>>>                         ¦ Therefore  all  progress  depends on 
>>>>>>> the ¦
>>>>>>>                         ¦ unreasonable man. -- George Bernard 
>>>>>>> Shaw ¦
>>>>>
>>>>> --
>>>>> Brian F. G. Bidulock    ¦ The reasonable man adapts himself to the ¦
>>>>> bidulock@openss7.org    ¦ world; the unreasonable one persists in  ¦
>>>>> http://www.openss7.org/ ¦ trying  to adapt the  world  to himself. ¦
>>>>>                         ¦ Therefore  all  progress  depends on the ¦
>>>>>                         ¦ unreasonable man. -- George Bernard Shaw ¦
>>>>>
>>>
>>> --
>>> Brian F. G. Bidulock    ¦ The reasonable man adapts himself to the ¦
>>> bidulock@openss7.org    ¦ world; the unreasonable one persists in  ¦
>>> http://www.openss7.org/ ¦ trying  to adapt the  world  to himself. ¦
>>>                         ¦ Therefore  all  progress  depends on the ¦
>>>                         ¦ unreasonable man. -- George Bernard Shaw ¦
>>>
>
> --
> Brian F. G. Bidulock    ¦ The reasonable man adapts himself to the ¦
> bidulock@openss7.org    ¦ world; the unreasonable one persists in  ¦
> http://www.openss7.org/ ¦ trying  to adapt the  world  to himself. ¦
>                         ¦ Therefore  all  progress  depends on the ¦
>                         ¦ unreasonable man. -- George Bernard Shaw ¦
>


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



From daemon@optimus.ietf.org  Sat Jun  8 17:29:23 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 RAA12126
	for <tsvwg-archive@odin.ietf.org>; Sat, 8 Jun 2002 17:29:23 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id RAA12980
	for tsvwg-archive@odin.ietf.org; Sat, 8 Jun 2002 17:29:57 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id QAA11260;
	Sat, 8 Jun 2002 16:57:51 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id QAA11232
	for <tsvwg@optimus.ietf.org>; Sat, 8 Jun 2002 16:57:49 -0400 (EDT)
Received: from bfgbhome.inetint.com (IDENT:root@tnt-dal-42-211.dallas.net [209.44.42.211])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id QAA11726
	for <tsvwg@ietf.org>; Sat, 8 Jun 2002 16:57:14 -0400 (EDT)
Received: (from brian@localhost)
	by bfgbhome.inetint.com (8.9.3/8.9.3) id PAA22509;
	Sat, 8 Jun 2002 15:56:26 -0500
Date: Sat, 8 Jun 2002 15:56:26 -0500
From: "Brian F. G. Bidulock" <bidulock@openss7.org>
To: Michael Tuexen <Michael.Tuexen@lurchi.franken.de>
Cc: Qiaobing Xie <Qiaobing.Xie@motorola.com>,
        Randall Stewart <randall@stewart.chicago.il.us>,
        Phillip Conrad <conrad@acm.org>,
        "Armando L. Caro Jr." <acaro@mail.eecis.udel.edu>, tsvwg@ietf.org,
        "Michael A. Ramalho" <mramalho@cisco.com>
Subject: Re: [Tsvwg] More on PRSCTP-00
Message-ID: <20020608155626.E20649@openss7.org>
Reply-To: bidulock@openss7.org
References: <20020608142612.D20649@openss7.org> <B8DDC5CE-7B1E-11D6-9BA0-0030654C1AB6@lurchi.franken.de>
Mime-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
User-Agent: Mutt/1.2i
In-Reply-To: <B8DDC5CE-7B1E-11D6-9BA0-0030654C1AB6@lurchi.franken.de>; from Michael.Tuexen@lurchi.franken.de on Sat, Jun 08, 2002 at 10:31:33PM +0200
Organization: http://www.openss7.org/
Dsn-Notification-To: <bidulock@openss7.org>
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

Michael,

Good.  Now if we can convince Qiaobing it will be unanimous.

--brian

On Sat, 08 Jun 2002, Michael Tuexen wrote:

[snip]
> >>
> >> So what will happen:
> >> 8. The retransmission timer will run off for packet3.
> >> 9. packet3 gets retransmitted, and the user data of packet3 and packet4
> >> can be delivered.
> >>
> >> I agree with you that there is HOL, but it is the same as there would 
> >> be
> >> if you
> >> would not use PR-SCTP. So PR-SCTP, as it is currently being defined 
> >> does
> >> not perform
> >> better than basic SCTP in terms of user-to-user message delay. It does
> >> perform better in terms
> >> of buffer and bandwidth usage.
> >
> > I disagree with that analysis.  Normal SCTP only has to wait for
> > retransmission of the SSN 2 packet, whereas PR-SCTP in this scenario
> > has to wait for retransmission of all the TSN gaps subsequent to
> > the FORWARD TSN.  These (e.g. TSN 119) might have 16k worth of data
> > in it making it subject to many retransmissions (the network is already
> > marginally in congestion).  The SSN 2 packet might contain 256 bytes
> > meaning that retransmission of TSNs 110-118 could be successful when
> > TSN 119 is going to be unsuccessful for some time.
> >
> > It is NOT the situation in regular SCTP that a stream must block
> > awaiting the retransmission of a TSN from a different stream.  For
> > PR-SCTP, streams 1 through 9 are blocked awaiting the retransmission
> > of a data chunk on stream 0.  This is the unacceptable HOL blocking.
> > It is inter-stream, not intra-stream.
> Hmmm. I see. The user data of packet4 can be delivered to the upper layer
> after the FWD-TSN (something like a retransmission of packet2) and 
> packet3
> have been received. In normal SCTP only packet2 needs to be received.
> Since the user-to-user delay of PR-SCTP should not be greater than for
> SCTP, I think we should put the SSN/SID info in the FWD-TSN chunk.
> But only for the affected streams. Qiaobing, what do you think?
> Best regards
> Michael
[snip]

-- 
Brian F. G. Bidulock    ¦ The reasonable man adapts himself to the ¦
bidulock@openss7.org    ¦ world; the unreasonable one persists in  ¦
http://www.openss7.org/ ¦ trying  to adapt the  world  to himself. ¦
                        ¦ Therefore  all  progress  depends on the ¦
                        ¦ unreasonable man. -- George Bernard Shaw ¦

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



From daemon@ns.ietf.org  Mon Jun 10 14:07:49 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 OAA16887
	for <tsvwg-archive@odin.ietf.org>; Mon, 10 Jun 2002 14:07:49 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id OAA20267
	for tsvwg-archive@odin.ietf.org; Mon, 10 Jun 2002 14:08:23 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id NAA18425;
	Mon, 10 Jun 2002 13:41:33 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id NAA18391
	for <tsvwg@ns.ietf.org>; Mon, 10 Jun 2002 13:41:31 -0400 (EDT)
Received: from motgate2.mot.com (motgate2.mot.com [136.182.1.10])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id NAA15639
	for <tsvwg@ietf.org>; Mon, 10 Jun 2002 13:40:56 -0400 (EDT)
Received: [from pobox3.mot.com (pobox3.mot.com [10.64.251.242]) by motgate2.mot.com (motgate2 2.1) with ESMTP id KAA17109; Mon, 10 Jun 2002 10:41:31 -0700 (MST)]
Received: [from relay1.cig.mot.com (relay1.cig.mot.com [136.182.15.23]) by pobox3.mot.com (MOT-pobox3 2.0) with ESMTP id KAA00248; Mon, 10 Jun 2002 10:41:00 -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 g5AHfLb23038;
	Mon, 10 Jun 2002 12:41:21 -0500 (CDT)
Received: from motorola.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 MAA12369;
	Mon, 10 Jun 2002 12:41:45 -0500 (CDT)
Message-ID: <3D04E575.6E332C21@motorola.com>
Date: Mon, 10 Jun 2002 12:44:21 -0500
From: Qiaobing Xie <Qiaobing.Xie@motorola.com>
X-Mailer: Mozilla 4.77 [en] (Win98; U)
X-Accept-Language: en
MIME-Version: 1.0
To: bidulock@openss7.org
CC: Michael Tuexen <Michael.Tuexen@lurchi.franken.de>,
        Randall Stewart <randall@stewart.chicago.il.us>,
        Phillip Conrad <conrad@acm.org>,
        "Armando L. Caro Jr." <acaro@mail.eecis.udel.edu>, tsvwg@ietf.org,
        "Michael A. Ramalho" <mramalho@cisco.com>
Subject: Re: [Tsvwg] More on PRSCTP-00
References: <20020608142612.D20649@openss7.org> <B8DDC5CE-7B1E-11D6-9BA0-0030654C1AB6@lurchi.franken.de> <20020608155626.E20649@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

"Brian F. G. Bidulock" wrote:
> 
> Michael,
> 
> Good.  Now if we can convince Qiaobing it will be unanimous.

Brian,

As I said before, I have no strong objection to the change. Still,
Randy's adding SNN info to make the implementation easier sounds a much
better argument to me. 

The reason that I encouraged you to fill in the precise details to your
initial example should become apparent now. As demonstrated by the
exercise you and Michael went through, to show that the absence of SNN
info in Fwd TSN can really cause a meaningful delay at the receiver end
is not as easy as you might originally think. That is my point - yes, it
can cause delay but you need to assume quite a few things happening
simultaneously for the delay to become meaningful. 

-Qiaobing

> 
> --brian
>

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



From daemon@optimus.ietf.org  Wed Jun 12 09:13:25 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 JAA26817
	for <tsvwg-archive@odin.ietf.org>; Wed, 12 Jun 2002 09:13:24 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id JAA27281
	for tsvwg-archive@odin.ietf.org; Wed, 12 Jun 2002 09:14:01 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id IAA25437;
	Wed, 12 Jun 2002 08:51:33 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id IAA25405
	for <tsvwg@optimus.ietf.org>; Wed, 12 Jun 2002 08:51:31 -0400 (EDT)
Received: from mx10.radisys.com (mx10.radisys.com [206.102.10.36])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id IAA25977
	for <tsvwg@ietf.org>; Wed, 12 Jun 2002 08:50:54 -0400 (EDT)
From: Simon.Zhuang@radisys.com
Received: by mx10.radisys.com (Postfix, from userid 5)
	id F2C24142DFC; Wed, 12 Jun 2002 05:51:25 -0700 (PDT)
Received: from UNKNOWN(206.103.52.220), claiming to be "notes.radisys.com"
 via SMTP by mx10, id smtpdAAAumUKD_; Wed Jun 12 05:51:15 2002
To: tsvwg@ietf.org
X-Mailer: Lotus Notes Release 5.0.8  June 18, 2001
Message-ID: <OFB0F2B3B0.5772B65D-ON85256BD6.00441CCD@radisys.com>
Date: Wed, 12 Jun 2002 08:36:08 -0400
X-MIMETrack: Serialize by Router on HQ_ACS_1/Radisys_Corporation/US(Release 5.0.9a |January
 7, 2002) at 06/12/2002 05:41:23 AM
MIME-Version: 1.0
Content-type: text/plain; charset=us-ascii
Subject: [Tsvwg] Question about special case for changing an address in SCTP-AddIP
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 List,

In <draft-ietf-tsvwg-addip-sctp-05.txt>, we have:

4.3.2 A special case for changing an address.

The receiver of such a chunk MUST always first use the source address
    found in the IP header in looking up the association.  The
    receiver should attempt to use the address found in the Address
    Bytes field only if the lookup fails using the source address from
    the IP header.  The receiver MUST reply to the source address
    of the packet in this case which is the new address that
    was added by the ASCONF (since the old address is no longer a part
    of the association after processing).

My question is: why the sender of this ASCONF can use the new address as
source
address before it receives ASCONF-ACK?


Simon


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



From daemon@optimus.ietf.org  Wed Jun 12 12:05:29 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 MAA02970
	for <tsvwg-archive@odin.ietf.org>; Wed, 12 Jun 2002 12:05:29 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id MAA08677
	for tsvwg-archive@odin.ietf.org; Wed, 12 Jun 2002 12:06:03 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id LAA05386;
	Wed, 12 Jun 2002 11:28:41 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id LAA05356
	for <tsvwg@optimus.ietf.org>; Wed, 12 Jun 2002 11:28:39 -0400 (EDT)
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 LAA01365
	for <tsvwg@ietf.org>; Wed, 12 Jun 2002 11:28:04 -0400 (EDT)
Received: from mira-sjc5-2.cisco.com (IDENT:mirapoint@mira-sjc5-2.cisco.com [171.71.163.16])
	by sj-msg-core-4.cisco.com (8.12.2/8.12.2) with ESMTP id g5CFQnIZ027511;
	Wed, 12 Jun 2002 08:28:05 -0700 (PDT)
Received: from stewart.chicago.il.us (dhcp-128-107-163-27.cisco.com [128.107.163.27])
	by mira-sjc5-2.cisco.com (Mirapoint)
	with ESMTP id ACQ35221;
	Wed, 12 Jun 2002 08:26:48 -0700 (PDT)
Message-ID: <3D076836.679D2E43@stewart.chicago.il.us>
Date: Wed, 12 Jun 2002 10:26:46 -0500
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: Simon.Zhuang@radisys.com
CC: tsvwg@ietf.org
Subject: Re: [Tsvwg] Question about special case for changing an address in 
 SCTP-AddIP
References: <OFB0F2B3B0.5772B65D-ON85256BD6.00441CCD@radisys.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

Simon:

One very good example of this would be where you have an
IPv6 renumber occur... and the former address is no
longer valid to source from. In this case you would stuff
the old address in to the address TLV so the peer has
a lookup mechanism. In fact one could even think about
using this address to always lookup the association the
more I think on it.. since you would never have a
lookup/miss-lookup/find..


There are other places it can be used as well... Michael Tuexen as
enumerated a few in his draft :>

R

Simon.Zhuang@radisys.com wrote:
> 
> Hi List,
> 
> In <draft-ietf-tsvwg-addip-sctp-05.txt>, we have:
> 
> 4.3.2 A special case for changing an address.
> 
> The receiver of such a chunk MUST always first use the source address
>     found in the IP header in looking up the association.  The
>     receiver should attempt to use the address found in the Address
>     Bytes field only if the lookup fails using the source address from
>     the IP header.  The receiver MUST reply to the source address
>     of the packet in this case which is the new address that
>     was added by the ASCONF (since the old address is no longer a part
>     of the association after processing).
> 
> My question is: why the sender of this ASCONF can use the new address as
> source
> address before it receives ASCONF-ACK?
> 
> Simon
> 
> _______________________________________________
> 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 Jun 13 09:20:53 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 JAA10895
	for <tsvwg-archive@odin.ietf.org>; Thu, 13 Jun 2002 09:20:53 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id JAA18369
	for tsvwg-archive@odin.ietf.org; Thu, 13 Jun 2002 09:21:29 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id JAA17126;
	Thu, 13 Jun 2002 09:00:44 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id JAA17087
	for <tsvwg@ns.ietf.org>; Thu, 13 Jun 2002 09:00:42 -0400 (EDT)
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 JAA10289
	for <tsvwg@ietf.org>; Thu, 13 Jun 2002 09:00:05 -0400 (EDT)
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 PAA21763;
	Thu, 13 Jun 2002 15:00:33 +0200 (MET DST)
Received: from mchh274e.demchh201e.icn.siemens.de ([139.21.200.84])
	by moody.mchh.siemens.de (8.9.1/8.9.1) with ESMTP id PAA27625;
	Thu, 13 Jun 2002 15:00:33 +0200 (MET DST)
Received: by MCHH274E with Internet Mail Service (5.5.2653.19)
	id <MY34LSJF>; Thu, 13 Jun 2002 15:00:18 +0200
Message-ID: <F92978223B01D311ACFF0008C71EE19D015DF84D@MCHH225E>
From: Tuexen Michael <Michael.Tuexen@icn.siemens.de>
To: Simon.Zhuang@radisys.com
Cc: tsvwg@ietf.org
Subject: RE: [Tsvwg] Question about special case for changing an address i
	n  SCTP-AddIP
Date: Thu, 13 Jun 2002 15:00:14 +0200
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

Simon,

you want to look at
http://www.ietf.org/internet-drafts/draft-riegel-tuexen-mobile-sctp-00.txt

If you have further questions, do not hesitate to contact me.

Best regards
Michael

Michael Tuexen    Tel.:   +49 89 722 47210
Siemens AG        Fax:    +49 89 722 48212
ICN WN CC SE 7    e-Mail: Michael.Tuexen@icn.siemens.de
 

> -----Original Message-----
> From: Randall Stewart [mailto:randall@stewart.chicago.il.us]
> Sent: Wednesday, June 12, 2002 5:27 PM
> To: Simon.Zhuang@radisys.com
> Cc: tsvwg@ietf.org
> Subject: Re: [Tsvwg] Question about special case for changing 
> an address
> in SCTP-AddIP
> 
> 
> Simon:
> 
> One very good example of this would be where you have an
> IPv6 renumber occur... and the former address is no
> longer valid to source from. In this case you would stuff
> the old address in to the address TLV so the peer has
> a lookup mechanism. In fact one could even think about
> using this address to always lookup the association the
> more I think on it.. since you would never have a
> lookup/miss-lookup/find..
> 
> 
> There are other places it can be used as well... Michael Tuexen as
> enumerated a few in his draft :>
> 
> R
> 
> Simon.Zhuang@radisys.com wrote:
> > 
> > Hi List,
> > 
> > In <draft-ietf-tsvwg-addip-sctp-05.txt>, we have:
> > 
> > 4.3.2 A special case for changing an address.
> > 
> > The receiver of such a chunk MUST always first use the 
> source address
> >     found in the IP header in looking up the association.  The
> >     receiver should attempt to use the address found in the Address
> >     Bytes field only if the lookup fails using the source 
> address from
> >     the IP header.  The receiver MUST reply to the source address
> >     of the packet in this case which is the new address that
> >     was added by the ASCONF (since the old address is no 
> longer a part
> >     of the association after processing).
> > 
> > My question is: why the sender of this ASCONF can use the 
> new address as
> > source
> > address before it receives ASCONF-ACK?
> > 
> > Simon
> > 
> > _______________________________________________
> > 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@optimus.ietf.org  Fri Jun 14 14:49:27 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 OAA10273
	for <tsvwg-archive@odin.ietf.org>; Fri, 14 Jun 2002 14:49:27 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id OAA00109
	for tsvwg-archive@odin.ietf.org; Fri, 14 Jun 2002 14:50:03 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id OAA29381;
	Fri, 14 Jun 2002 14:32:47 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id OAA29350
	for <tsvwg@optimus.ietf.org>; Fri, 14 Jun 2002 14:32:45 -0400 (EDT)
Received: from loquat.bbn.com (crodrigues.bbn.com [128.89.72.49])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id OAA09641
	for <tsvwg@ietf.org>; Fri, 14 Jun 2002 14:32:08 -0400 (EDT)
Received: (from crodrigu@localhost)
	by loquat.bbn.com (8.11.2/8.11.2) id g5EIVSw23549;
	Fri, 14 Jun 2002 14:31:28 -0400
Date: Fri, 14 Jun 2002 14:31:28 -0400
From: Craig Rodrigues <crodrigu@bbn.com>
To: Randall Stewart at work <rrs@cisco.com>
Cc: qxie1@email.mot.com, piggy@acm.org, jonwood@speakeasy.net,
        kacheong.poon@sun.com, fken@cd.jp.nec.com, tsvwg@ietf.org
Message-ID: <20020614143128.A23394@bbn.com>
References: <20020211115601.A32269@bbn.com> <3C682748.C66FF5C5@cisco.com> <20020613101028.A15591@bbn.com>
Mime-Version: 1.0
Content-Type: multipart/mixed; boundary="fdj2RfSjLxBAspz7"
Content-Disposition: inline
User-Agent: Mutt/1.2.5i
In-Reply-To: <20020613101028.A15591@bbn.com>; from crodrigu@bbn.com on Thu, Jun 13, 2002 at 10:10:28AM -0400
Subject: [Tsvwg] draft-ietf-tsvwg-sctpsocket-04.txt proposed corrections
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


--fdj2RfSjLxBAspz7
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline

Hi,

Here is a patch to draft-ietf-tsvwg-sctpsocket-04.txt to
change a few of the types in some of the function prototypes
on the socket calls to use socklen_t where appropriate.

Thanks.
-- 
Craig Rodrigues        Distributed Systems and Logistics, Office 6/304
crodrigu@bbn.com       BBN Technologies, a Verizon company
(617) 873-4725         Cambridge, MA


--fdj2RfSjLxBAspz7
Content-Type: text/plain; charset=us-ascii
Content-Disposition: attachment; filename="draft-ietf-tsvwg-sctpsocket-04.txt.diff"

--- draft-ietf-tsvwg-sctpsocket-04.txt.orig	Fri Jun 14 13:48:43 2002
+++ draft-ietf-tsvwg-sctpsocket-04.txt	Fri Jun 14 14:14:00 2002
@@ -308,7 +308,7 @@
 
     The syntax of bind() is,
 
-     ret = bind(int sd, struct sockaddr *addr, int addrlen);
+     ret = bind(int sd, struct sockaddr *addr, socklen_t addrlen);
 
      sd      - the socket descriptor returned by socket().
      addr    - the address structure (struct sockaddr_in or struct
@@ -478,7 +478,7 @@
 
     The syntax is
 
-      ret = connect(int sd, const struct sockaddr *nam, int len);
+      ret = connect(int sd, const struct sockaddr *nam, socklen_t len);
 
       sd      - the socket descriptor to have a new association added
                 to.
@@ -658,7 +658,7 @@
 
     The syntax is:
 
-     int bind(int sd, struct sockaddr *addr, int addrlen);
+     int bind(int sd, struct sockaddr *addr, socklen_t addrlen);
 
       sd      - the socket descriptor returned by socket() call.
       addr    - the address structure (either struct sockaddr_in or struct
@@ -736,7 +736,7 @@
 
     The syntax is
 
-     int connect(int sd, const struct sockaddr *addr, int addrlen);
+     int connect(int sd, const struct sockaddr *addr, socklen_t addrlen);
 
       sd      - the socket descriptor of the endpoint.
       addr    - the peer's address.
@@ -1826,7 +1826,7 @@
 
       ssize_t send(int sd, connst void *msg, size_t len, int flags);
       ssize_t sendto(int sd, const void *msg, size_t len, int flags,
-	             const struct sockaddr *to, int tolen);
+	             const struct sockaddr *to, socklen_t tolen);
 
 Stewart et.al.                                                 [Page 31]
 
@@ -1834,7 +1834,7 @@
 
       ssize_t recv(int sd, void *buf, size_t len, int flags);
       ssize_t recvfrom(int sd, void *buf, size_t len, int flags,
-                       struct sockaddr *from, int *fromlen);
+                       struct sockaddr *from, socklen_t *fromlen);
 
        sd      - the socket descriptor of an SCTP endpoint.
        msg     - the message to be sent.
@@ -1896,9 +1896,9 @@
     The syntax is:
 
     ret = getsockopt(int sd, int level, int optname, void *optval,
-                     size_t *optlen);
+                     socklen_t *optlen);
     ret = setsockopt(int sd, int level, int optname, const void *optval,
-                     size_t optlen);
+                     socklen_t optlen);
 
       sd      - the socket descript.
       level   - set to IPPROTO_SCTP for all SCTP options.

--fdj2RfSjLxBAspz7--

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



From daemon@optimus.ietf.org  Sat Jun 15 17:19:49 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 RAA14777
	for <tsvwg-archive@odin.ietf.org>; Sat, 15 Jun 2002 17:19:49 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id RAA05822
	for tsvwg-archive@odin.ietf.org; Sat, 15 Jun 2002 17:20:27 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id RAA05650;
	Sat, 15 Jun 2002 17:08:36 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id HAA05230
	for <tsvwg@optimus.ietf.org>; Fri, 14 Jun 2002 07:31:49 -0400 (EDT)
Received: from CNRI.Reston.VA.US (localhost [127.0.0.1])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id HAA25042;
	Fri, 14 Jun 2002 07:31:12 -0400 (EDT)
Message-Id: <200206141131.HAA25042@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: Fri, 14 Jun 2002 07:31:12 -0400
Subject: [Tsvwg] I-D ACTION:draft-ietf-tsvwg-initwin-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		: Increasing TCP's Initial Window
	Author(s)	: M. Allman, S. Floyd, C. Partridge
	Filename	: draft-ietf-tsvwg-initwin-04.txt
	Pages		: 13
	Date		: 13-Jun-02
	
This document specifies an optional standard for TCP to increase the
permitted initial window from one segment to roughly 4K bytes,
replacing RFC 2414.  This document discusses the advantages and
disadvantages of the higher initial window.  The document includes
discussion of experiments and simulations showing that the higher
initial window does not lead to congestion collapse. Finally, the
document provides guidance on implementation issues.

A URL for this Internet-Draft is:
http://www.ietf.org/internet-drafts/draft-ietf-tsvwg-initwin-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-initwin-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-initwin-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:	<20020613143615.I-D@ietf.org>

ENCODING mime
FILE /internet-drafts/draft-ietf-tsvwg-initwin-04.txt

--OtherAccess
Content-Type: Message/External-body;
	name="draft-ietf-tsvwg-initwin-04.txt";
	site="ftp.ietf.org";
	access-type="anon-ftp";
	directory="internet-drafts"

Content-Type: text/plain
Content-ID:	<20020613143615.I-D@ietf.org>

--OtherAccess--

--NextPart--




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



From daemon@optimus.ietf.org  Sat Jun 15 17:20:02 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 RAA14795
	for <tsvwg-archive@odin.ietf.org>; Sat, 15 Jun 2002 17:20:01 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id RAA05836
	for tsvwg-archive@odin.ietf.org; Sat, 15 Jun 2002 17:20:39 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id RAA05634;
	Sat, 15 Jun 2002 17:08:35 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id NAA26770
	for <tsvwg@optimus.ietf.org>; Fri, 14 Jun 2002 13:45:45 -0400 (EDT)
Received: from cougar.icir.org (cougar.icir.org [192.150.187.76])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id NAA07813
	for <tsvwg@ietf.org>; Fri, 14 Jun 2002 13:45:08 -0400 (EDT)
Received: from cougar.icir.org (localhost [127.0.0.1])
	by cougar.icir.org (8.11.6/8.11.3) with ESMTP id g5EHjhB75994
	for <tsvwg@ietf.org>; Fri, 14 Jun 2002 10:45:43 -0700 (PDT)
	(envelope-from floyd@cougar.icir.org)
Message-Id: <200206141745.g5EHjhB75994@cougar.icir.org>
To: tsvwg@ietf.org
From: Sally Floyd <floyd@icir.org>
Date: Fri, 14 Jun 2002 10:45:43 -0700
Subject: [Tsvwg] HighSpeed TCP
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 new internet-drafts are just out about HighSpeed TCP.  This is
going to be discussed in tsvwg in Yokohama.  I have created a web
page for this at "http://www.icir.org/floyd/hstcp.html", and any
feedback would be appreciated.

Many thanks,
- Sally
http://www.icir.org/floyd/

------- Forwarded Messages

Date:    Fri, 14 Jun 2002 07:30:01 -0400
From:    Internet-Drafts@ietf.org
To:      IETF-Announce: ;
Subject: I-D ACTION:draft-floyd-tcp-highspeed-00.txt,.ps

--NextPart

A New Internet-Draft is available from the on-line Internet-Drafts directories.


	Title		: HighSpeed TCP for Large Congestion Windows
	Author(s)	: S. Floyd
	Filename	: draft-floyd-tcp-highspeed-00.txt,.ps
	Pages		: 16
	Date		: 13-Jun-02
	
This document proposes HighSpeed TCP, a modification to TCP's
congestion control mechanism for use with TCP connections with large
congestion windows.  The congestion control mechanisms of the
current, Standard TCP constrains the congestion windows that can be
achieved by TCP in realistic environments.  For example, for a
Standard TCP connection with 1500-byte packets and a 100 ms round-
trip time, achieving a steady-state throughput of 10 Gbps would
require an average congestion window of 83,333 segments, and a packet
drop rate of at most one congestion event every 5,000,000,000 packet
(or equivalently, at most one congestion event every 1 2/3 hours).
We do not consider this a realistic condition.  To address this
limitation of TCP, this document proposes HighSpeed TCP, and solicits
experimentation and feedback from the wider community.

A URL for this Internet-Draft is:
http://www.ietf.org/internet-drafts/draft-floyd-tcp-highspeed-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-floyd-tcp-highspeed-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-floyd-tcp-highspeed-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:	<20020613143000.I-D@ietf.org>

ENCODING mime
FILE /internet-drafts/draft-floyd-tcp-highspeed-00.txt

--OtherAccess
Content-Type: Message/External-body;
	name="draft-floyd-tcp-highspeed-00.txt";
	site="ftp.ietf.org";
	access-type="anon-ftp";
	directory="internet-drafts"

Content-Type: text/plain
Content-ID:	<20020613143000.I-D@ietf.org>

--OtherAccess--

--NextPart--



------- Message 2

Date:    Fri, 14 Jun 2002 07:30:25 -0400
From:    Internet-Drafts@ietf.org
To:      IETF-Announce: ;
Subject: I-D ACTION:draft-floyd-tcp-slowstart-00.txt,.ps

--NextPart

A New Internet-Draft is available from the on-line Internet-Drafts directories.


	Title		: Limited Slow-Start for TCP with Large Congestion 
                          Windows
	Author(s)	: S. Floyd
	Filename	: draft-floyd-tcp-slowstart-00.txt,.ps
	Pages		: 5
	Date		: 13-Jun-02
	
This note proposes a modification for TCP's slow-start for use with
TCP connections with large congestion windows.  For TCP connections
that are able to use congestion windows of thousands (or tens of
thousands) of MSS-sized segments (for MSS the sender's MAXIMUM
SEGMENT SIZE), the current slow-start procedure can result in
increasing the congestion window by thousands of segments in a single
round-trip time.  Such an increase can easily result in thousands of
packets being dropped in one round-trip time.  This is often counter-
productive for the TCP flow itself, and is also hard on the rest of
the traffic sharing the congested link.  This note proposes Limited
Slow-Start, limiting the number of segments by which the congestion
window is increased for one window of data during slow-start, in
order to improve performance for TCP connections with large
congestion windows.

A URL for this Internet-Draft is:
http://www.ietf.org/internet-drafts/draft-floyd-tcp-slowstart-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-floyd-tcp-slowstart-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-floyd-tcp-slowstart-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:	<20020613143436.I-D@ietf.org>

ENCODING mime
FILE /internet-drafts/draft-floyd-tcp-slowstart-00.txt

--OtherAccess
Content-Type: Message/External-body;
	name="draft-floyd-tcp-slowstart-00.txt";
	site="ftp.ietf.org";
	access-type="anon-ftp";
	directory="internet-drafts"

Content-Type: text/plain
Content-ID:	<20020613143436.I-D@ietf.org>

--OtherAccess--

--NextPart--



------- End of Forwarded Messages



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



From daemon@optimus.ietf.org  Mon Jun 24 15:30:44 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 PAA26348
	for <tsvwg-archive@odin.ietf.org>; Mon, 24 Jun 2002 15:30:44 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id PAA06158
	for tsvwg-archive@odin.ietf.org; Mon, 24 Jun 2002 15:31:27 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id PAA04678;
	Mon, 24 Jun 2002 15:04:16 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id OAA03322
	for <tsvwg@optimus.ietf.org>; Mon, 24 Jun 2002 14:48:32 -0400 (EDT)
Received: from cougar.icir.org (cougar.icir.org [192.150.187.76])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id OAA24082
	for <tsvwg@ietf.org>; Mon, 24 Jun 2002 14:47:48 -0400 (EDT)
Received: from cougar.icir.org (localhost [127.0.0.1])
	by cougar.icir.org (8.11.6/8.11.3) with ESMTP id g5OImUB00534;
	Mon, 24 Jun 2002 11:48:30 -0700 (PDT)
	(envelope-from floyd@cougar.icir.org)
Message-Id: <200206241848.g5OImUB00534@cougar.icir.org>
To: tsvwg@ietf.org, mankin@isi.edu, Scott Bradner <sob@harvard.edu>
cc: "Amit Jain" <ajain@netscaler.com>
From: Sally Floyd <floyd@icir.org>
Date: Mon, 24 Jun 2002 11:48:30 -0700
Subject: [Tsvwg] Quick-Start for TCP and IP
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

Amit Jain and I have a new internet-draft on
Quick-Start for TCP and IP, available at:

http://www.icir.org/floyd/papers/draft-amit-quick-start-00b.txt

We would appreciate any feedback from this communitu.

This is something that it would take some time to deploy, and it
might or might not be a good idea, but it could significantly reduce
transfer times for moderate-sized best-effort traffic over
underutilized, high-bandwidth infrastructures...

We would note that this is not (at the moment) an official tsvwg
work item, or even an official internet-draft (since we missed the
submission deadline).

Thanks,
- Sally

Abstract:

   This draft outlines an optional Quick-Start mechanism for transport
   protocols to determine an optional allowed initial congestion window
   or initial sending rate at the start of a data transfer.  This
   mechanism is designed to be used by a range of transport protocols;
   however, in this document we only describe its use with TCP and IPv4.
   By using Quick-Start, a TCP host, say, host A, would indicate its
   desired initial sending rate in packets per second in a Quick Start
   Request option in the IP header of the initial TCP SYN or SYN/ACK
   packet. Each router in turn could either approve the specified
   initial rate, reduce the specified initial rate, or indicate that
   nothing above the default initial rate for that protocol would be
   allowed.  The Quick-Start mechanism also can determine if there are
   routers along the path that do not understand the Quick Start Request
   option, or have not agreed to the initial rate described in the
   option. If all of the routers along the path have agreed to the
   initial rate in the Quick-Start Request, then TCP host B communicates
   this to TCP host A in a transport-level Quick-Start Response in the
   answering SYN/ACK or ACK packet.  Quick-Start is designed to allow
   TCP connections to use high initial windows in circumstances when
   there is significant unused bandwidth along the path, and all of the
   routers along the path support the Quick-Start Request.  We note that
   this is currently not a mature proposal, but an outline of one
   possible path of development in the IP protocol, and a request for
   feedback from the community.

------- Forwarded Message

Date:    Mon, 24 Jun 2002 09:31:48 -0700
From:    Sally Floyd <floyd@icir.org>
To:      internet-drafts@ietf.org
cc:      Amit Jain <ajain@netscaler.com>
Subject: Internet drafts submission of draft-amit-quick-start-00a.txt

I would like to submit this internet-draft.

http://www.icir.org/floyd/papers/draft-amit-quick-start-00a.txt
draft-amit-quick-sta.. 23-Jun-2002 22:13    36k  

I unfornately just realized this morning that the deadline was
9AM ET instead of 5PM ET, as I had been assuming, so probably
you are not able to accept this submission.  But I thought I
would try nevertheless: the file on the web server is dated
last night, and therefore the draft itself was completed by
the deadline...

- Sally

------- End of Forwarded Message



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



From daemon@optimus.ietf.org  Tue Jun 25 04:01:25 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 EAA25837
	for <tsvwg-archive@odin.ietf.org>; Tue, 25 Jun 2002 04:01:25 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id EAA24379
	for tsvwg-archive@odin.ietf.org; Tue, 25 Jun 2002 04:02:08 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id DAA23223;
	Tue, 25 Jun 2002 03:48:46 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id DAA23194
	for <tsvwg@optimus.ietf.org>; Tue, 25 Jun 2002 03:48:44 -0400 (EDT)
Received: from hindon.hss.co.in ([202.54.26.202])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id DAA25489
	for <tsvwg@ietf.org>; Tue, 25 Jun 2002 03:48:00 -0400 (EDT)
From: samahajan@hss.hns.com
Received: from ultra.hss.co.in (ultra.hss.hns.com [192.168.100.5])
	by hindon.hss.co.in (8.10.0/8.10.0) with ESMTP id g5P7mlY04970
	for <tsvwg@ietf.org>; Tue, 25 Jun 2002 13:18:47 +0530 (IST)
Received: from sandesh.hss.hns.com (localhost [127.0.0.1])
	by ultra.hss.co.in (8.10.0/8.10.0) with SMTP id g5P7nCM09488
	for <tsvwg@ietf.org>; Tue, 25 Jun 2002 13:19:12 +0530 (IST)
Received: by sandesh.hss.hns.com(Lotus SMTP MTA v4.6.3  (733.2 10-16-1998))  id 65256BE3.002A8728 ; Tue, 25 Jun 2002 13:14:31 +0530
X-Lotus-FromDomain: HSS
To: tsvwg@ietf.org
Message-ID: <65256BE3.002A856E.00@sandesh.hss.hns.com>
Date: Tue, 25 Jun 2002 13:14:20 +0530
Mime-Version: 1.0
Content-type: text/plain; charset=us-ascii
Content-Disposition: inline
Subject: [Tsvwg] SCTP: INITACK handling doubt
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 a very basic doubt about the handling of INITACK message. Please consider
the following scenario.
Side A send a INIT message to side B with src_port = Port A, src_addr = IP A,
dst_port = Port B, dst_addr= IP B, Init-Tag = Tag A.
Side sends back INITACK but changes its port number and/or IP address, So
INITACK has src_addr= IP C,  src_port = PORT C, dst_port = PORT A, dst_addr = IP
A. V-Tag = Tag A.

Questions :-
a) Is it allowed by SCTP specification that SCTP at side A while processing
INITACK message uses only dst_port, dst_addr and V-Tag parameters to correlate
received INITACK message with TCB of association sending INIT messag and ignores
src_port and src_address parameters ?
b) If it is allowed, is it a mandatory feature, So is it correct to assume that
if some implementation sends such INITACK, other side will be processing it ?
c) If it is a optional feature, Does any existing implementations supports this
?

Best Regards.
-Sandeep


DISCLAIMER: This message is proprietary to Hughes Software Systems
Limited (HSS) and is intended solely for the use of the individual
to whom it is addressed. It may contain  privileged or confidential
information  and should not be circulated or used for any purpose other
than for what it is intended. If you have received this message in error,
please notify the originator immediately. If you are not the intended
recipient, you are notified that you are strictly prohibited from using,
copying, altering, or disclosing the contents of this message. HSS accepts
no responsibility for loss or damage arising from the use of the information
transmitted by this email including damage from virus.



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



From daemon@optimus.ietf.org  Tue Jun 25 04:02:49 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 EAA25873
	for <tsvwg-archive@odin.ietf.org>; Tue, 25 Jun 2002 04:02:49 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id EAA24442
	for tsvwg-archive@odin.ietf.org; Tue, 25 Jun 2002 04:03:32 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id DAA23512;
	Tue, 25 Jun 2002 03:52:45 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id DAA23485
	for <tsvwg@optimus.ietf.org>; Tue, 25 Jun 2002 03:52:44 -0400 (EDT)
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 DAA25591
	for <tsvwg@ietf.org>; Tue, 25 Jun 2002 03:52:00 -0400 (EDT)
From: john.loughney@nokia.com
Received: from esvir03nok.nokia.com (esvir03nokt.ntc.nokia.com [172.21.143.35])
	by mgw-x2.nokia.com (Switch-2.2.1/Switch-2.2.0) with ESMTP id g5P7rBi17734
	for <tsvwg@ietf.org>; Tue, 25 Jun 2002 10:53:11 +0300 (EET DST)
Received: from esebh001.NOE.Nokia.com (unverified) by esvir03nok.nokia.com
 (Content Technologies SMTPRS 4.2.5) with ESMTP id <T5bb34f9ab3ac158f23077@esvir03nok.nokia.com>;
 Tue, 25 Jun 2002 10:52:42 +0300
Received: from esebe004.NOE.Nokia.com ([172.21.138.44]) by esebh001.NOE.Nokia.com with Microsoft SMTPSVC(5.0.2195.4905);
	 Tue, 25 Jun 2002 10:52:42 +0300
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: Tue, 25 Jun 2002 10:52:41 +0300
Message-ID: <0C1353ABB1DEB74DB067ADFF749C4EEFC65514@esebe004.NOE.Nokia.com>
Thread-Topic: New draft on using SCTP for signaling protocols
Thread-Index: AcIcHUgl2GqmaMHmQwWm8aGSXk9Kow==
To: <tsvwg@ietf.org>, <sigtran@ietf.com>
X-OriginalArrivalTime: 25 Jun 2002 07:52:42.0389 (UTC) FILETIME=[48FBDC50:01C21C1D]
Content-Transfer-Encoding: 8bit
X-MIME-Autoconverted: from quoted-printable to 8bit by optimus.ietf.org id DAA23486
Subject: [Tsvwg] New draft on using SCTP for signaling protocols
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,

Sorry for the cross-posting, but I have just submitted the following draft:

SCTP Layer for Transporting Signaling Protocols 

Abstract 
    
    
   The Stream Control Transport Protocol (SCTP) has been standardized to 
   provide a reliable transport services for signaling protocols.  SCTP 
   features a number of new features that are useful for transporting 
   signaling protocols, but may require additional information on how to 
   use them to transport existing signaling protocols. 

Which can be found here until it gets through the queue:

http://www-nrc.nokia.com/sua/draft-loughney-sctp-sig-prot-00.txt

=== 

Basically, this is the outcome of some internal protocol development.  We
found that using an existing SIGTRAN UA for peer-to-peer signaling
is often quite heavy and not needed.  However, adapting existing signaling
protocols to run directly over SCTP sockets may not always be feasible
without re-standardizing these protocols.

Currently, the draft details some issues for consideration.  We do have
much of this already running in code, but I thought it best to start
more at the basics before jumping into solutions.

I'd appreciate comments on this, if folks feel that it is useful or not.

best regards,
John

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



From daemon@optimus.ietf.org  Tue Jun 25 04:44:04 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 EAA26777
	for <tsvwg-archive@odin.ietf.org>; Tue, 25 Jun 2002 04:44:04 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id EAA27030
	for tsvwg-archive@odin.ietf.org; Tue, 25 Jun 2002 04:44:47 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id EAA25443;
	Tue, 25 Jun 2002 04:26:43 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id EAA25416
	for <tsvwg@optimus.ietf.org>; Tue, 25 Jun 2002 04:26:41 -0400 (EDT)
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 EAA26326
	for <tsvwg@ietf.org>; Tue, 25 Jun 2002 04:25:57 -0400 (EDT)
From: Pasi.Sarolahti@nokia.com
Received: from esvir02nok.ntc.nokia.com (esvir02nokt.ntc.nokia.com [172.21.143.34])
	by mgw-x2.nokia.com (Switch-2.2.1/Switch-2.2.0) with ESMTP id g5P8R8i16117
	for <tsvwg@ietf.org>; Tue, 25 Jun 2002 11:27:08 +0300 (EET DST)
Received: from esebh001.NOE.Nokia.com (unverified) by esvir02nok.ntc.nokia.com
 (Content Technologies SMTPRS 4.2.5) with ESMTP id <T5bb36ea974ac158f22076@esvir02nok.ntc.nokia.com>;
 Tue, 25 Jun 2002 11:26:37 +0300
Received: from esebe008.NOE.Nokia.com ([172.21.138.48]) by esebh001.NOE.Nokia.com with Microsoft SMTPSVC(5.0.2195.4905);
	 Tue, 25 Jun 2002 11:26:37 +0300
Received: from esebe020.NOE.Nokia.com ([172.21.138.59]) by esebe008.NOE.Nokia.com with Microsoft SMTPSVC(5.0.2195.4905);
	 Tue, 25 Jun 2002 11:26:37 +0300
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: Tue, 25 Jun 2002 11:26:36 +0300
Message-ID: <0142A94A281F654BB2AF6A0B6235D6270D7827@esebe020.NOE.Nokia.com>
Thread-Topic: Internet-Draft on avoiding unnecessary retransmissions after TCP RTO
Thread-Index: AcIcIgT9pncO2SfqTlaGPsR4x7PEeA==
To: <tsvwg@ietf.org>
Cc: <sob@harvard.edu>, <mankin@isi.edu>
X-OriginalArrivalTime: 25 Jun 2002 08:26:37.0479 (UTC) FILETIME=[05FDF370:01C21C22]
Content-Transfer-Encoding: 8bit
X-MIME-Autoconverted: from quoted-printable to 8bit by optimus.ietf.org id EAA25417
Subject: [Tsvwg] Internet-Draft on avoiding unnecessary retransmissions after TCP RTO
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,

We have a new internet-draft titled "F-RTO: A TCP RTO Recovery Algorithm for Avoiding Unnecessary Retransmissions" available at
http://www.cs.Helsinki.FI/u/sarolaht/papers/draft-sarolahti-tsvwg-tcp-frto-00.txt

The draft describes an algorithm for avoiding unnecessary retransmissions after spurious TCP RTOs. We believe that the F-RTO algorithm can be safely employed as an alternative RTO recovery algorithm for TCP, and thought it would be interesting to this group. We would appreciate any feedback on the draft.

Unfortunately I missed the submission deadline for Yokohama, but the draft can be found in the location given above until it is resubmitted after Yokohama.

Abstract:
   Spurious retransmission timeouts (RTOs) cause suboptimal TCP
   performance, because they often result in unnecessary retransmission
   of the last window of data. This document describes the "Forward RTO
   Recovery" (F-RTO) algorithm for recovering from the TCP RTOs
   efficiently even when the RTO was spurious. F-RTO is a TCP sender
   only algorithm that does not require any TCP options to operate.
   After retransmitting the first unacknowledged segment triggered by an
   RTO, the F-RTO algorithm at a TCP sender monitors the incoming
   acknowledgements to determine whether the timeout was spurious and to
   decide whether to send new segments or retransmit unacknowledged
   segments. The algorithm effectively avoids additional unnecessary
   retransmissions and thereby improves TCP performance in case of a
   spurious timeout.

Regards,
- Pasi

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



From daemon@optimus.ietf.org  Tue Jun 25 08:02:49 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 IAA01557
	for <tsvwg-archive@odin.ietf.org>; Tue, 25 Jun 2002 08:02:49 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id IAA06284
	for tsvwg-archive@odin.ietf.org; Tue, 25 Jun 2002 08:03:26 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id HAA05435;
	Tue, 25 Jun 2002 07:48:04 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id HAA05405
	for <tsvwg@optimus.ietf.org>; Tue, 25 Jun 2002 07:48:02 -0400 (EDT)
Received: from ilsa.franken.de (postfix@ilsa.franken.de [193.175.24.42])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id HAA01068
	for <tsvwg@ietf.org>; Tue, 25 Jun 2002 07:47:18 -0400 (EDT)
Received: from localhost (pD9E4F721.dip.t-dialin.net [217.228.247.33])
	by ilsa.franken.de (Postfix) with ESMTP
	id 9A57724618; Tue, 25 Jun 2002 13:47:44 +0200 (CEST)
Date: Tue, 25 Jun 2002 13:46:48 +0200
Subject: Re: [Tsvwg] SCTP: INITACK handling doubt
Content-Type: text/plain; charset=US-ASCII; format=flowed
Mime-Version: 1.0 (Apple Message framework v482)
Cc: tsvwg@ietf.org
To: samahajan@hss.hns.com
From: Michael Tuexen <Michael.Tuexen@icn.siemens.de>
In-Reply-To: <65256BE3.002A856E.00@sandesh.hss.hns.com>
Message-Id: <3B998DF7-8831-11D6-B128-0030654C1AB6@icn.siemens.de>
Content-Transfer-Encoding: 7bit
X-Mailer: Apple Mail (2.482)
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

Sandeep,

see my comments below.

Best regards
Michael

On Tuesday, June 25, 2002, at 09:44 AM, samahajan@hss.hns.com wrote:

>
>
>
> Hi All,
>
> I have a very basic doubt about the handling of INITACK message. Please 
> consider
> the following scenario.
> Side A send a INIT message to side B with src_port = Port A, src_addr = 
> IP A,
> dst_port = Port B, dst_addr= IP B, Init-Tag = Tag A.
> Side sends back INITACK but changes its port number and/or IP address, 
> So
> INITACK has src_addr= IP C,  src_port = PORT C, dst_port = PORT A, 
> dst_addr = IP
> A. V-Tag = Tag A.
>
> Questions :-
> a) Is it allowed by SCTP specification that SCTP at side A while 
> processing
> INITACK message uses only dst_port, dst_addr and V-Tag parameters to 
> correlate
> received INITACK message with TCB of association sending INIT messag 
> and ignores
> src_port and src_address parameters ?
>
This is not allowed in SCTP as defined in RFC 2960 and the Implementor 
guide.
Side A uses the
- the source port number
- destination port number (= port A)
- source address
- destination address
- addresses in the list of addresses in the INIT-ACK chunk
of the packet containing the INIT-ACK chunk to match the local TCB.
Therefore the source port of the INIT-ACK packet must be the destination 
port
of the INIT packet. The destination address of the INIT packet must be 
the
source address of the INIT-ACK or one of the addresses contained in the 
INIT-ACK.
This is clarified in the Implementor Guide.
> b) If it is allowed, is it a mandatory feature, So is it correct to 
> assume that
> if some implementation sends such INITACK, other side will be 
> processing it ?
It is OOTB.
> c) If it is a optional feature, Does any existing implementations 
> supports this
> ?
The correct behavior is OOTB handling. You MAY send an ABORT but B has no
TCB so sending makes only sense for debugging.

>
> Best Regards.
> -Sandeep
>
>
> DISCLAIMER: This message is proprietary to Hughes Software Systems
> Limited (HSS) and is intended solely for the use of the individual
> to whom it is addressed. It may contain  privileged or confidential
> information  and should not be circulated or used for any purpose other
> than for what it is intended. If you have received this message in 
> error,
> please notify the originator immediately. If you are not the intended
> recipient, you are notified that you are strictly prohibited from using,
> copying, altering, or disclosing the contents of this message. HSS 
> accepts
> no responsibility for loss or damage arising from the use of the 
> information
> transmitted by this email including damage from virus.
>
>
>
> _______________________________________________
> tsvwg mailing list
> tsvwg@ietf.org
> https://www1.ietf.org/mailman/listinfo/tsvwg
>
Michael.Tuexen@micmac.franken.de


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



From daemon@ns.ietf.org  Tue Jun 25 21:33:25 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 VAA05591
	for <tsvwg-archive@odin.ietf.org>; Tue, 25 Jun 2002 21:33:25 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id VAA00666
	for tsvwg-archive@odin.ietf.org; Tue, 25 Jun 2002 21:34:08 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id VAA29808;
	Tue, 25 Jun 2002 21:18:21 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id HAA05625
	for <tsvwg@optimus.ietf.org>; Tue, 25 Jun 2002 07:53:17 -0400 (EDT)
Received: from wiprom2mx1.wipro.com (wiprom2mx1.wipro.com [203.197.164.41])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id HAA01199
	for <tsvwg@ietf.org>; Tue, 25 Jun 2002 07:52:19 -0400 (EDT)
Received: from m2vwall5.wipro.com (m2vwall5.wipro.com [10.115.50.5])
	by wiprom2mx1.wipro.com (8.11.3/8.11.3) with SMTP id g5PBREe28161
	for <tsvwg@ietf.org>; Tue, 25 Jun 2002 16:57:15 +0530 (IST)
Received: from wipro.com ([192.168.188.173]) by
          ecmail.mail.wipro.com (Netscape Messaging Server 4.15) with
          ESMTP id GY9EH800.UFH; Tue, 25 Jun 2002 16:57:08 +0530 
Message-ID: <3D185489.652ED624@wipro.com>
Date: Tue, 25 Jun 2002 17:01:21 +0530
From: Mohammed Ashraf <mohammed.asharaf@wipro.com>
Organization: Wipro
X-Mailer: Mozilla 4.76 [en] (X11; U; Linux 2.4.2-2 i686)
X-Accept-Language: en
MIME-Version: 1.0
To: Sally Floyd <floyd@icir.org>
CC: tsvwg@ietf.org, Amit Jain <ajain@netscaler.com>
Subject: Re: [Tsvwg] Quick-Start for TCP and IP
References: <200206241848.g5OImUB00534@cougar.icir.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


I have some basic queries  on  Quick-Start for TCP and IP draft.  I
appreciate if  you could clarify this.

Here are my queries:
 1. Draft says that Quick start request option will be included as a part
of IP header, while  quick start response as part of
      TCP header. Is it because the  quick start response need not be
processed by IP?

2. I guess IP header option should keep 32 bit boundry (i.e multiple of 4
bytes ). But in the case of quick start request, it  has got 5 bytes. Is
the padding to be done here assumed?

Thanks in advance,
Ashraf


Sally Floyd wrote:

> Amit Jain and I have a new internet-draft on
> Quick-Start for TCP and IP, available at:
>
> http://www.icir.org/floyd/papers/draft-amit-quick-start-00b.txt
>
> We would appreciate any feedback from this communitu.
>
> This is something that it would take some time to deploy, and it
> might or might not be a good idea, but it could significantly reduce
> transfer times for moderate-sized best-effort traffic over
> underutilized, high-bandwidth infrastructures...
>
> We would note that this is not (at the moment) an official tsvwg
> work item, or even an official internet-draft (since we missed the
> submission deadline).
>
> Thanks,
> - Sally
>
> Abstract:
>
>    This draft outlines an optional Quick-Start mechanism for transport
>    protocols to determine an optional allowed initial congestion window
>    or initial sending rate at the start of a data transfer.  This
>    mechanism is designed to be used by a range of transport protocols;
>    however, in this document we only describe its use with TCP and IPv4.
>    By using Quick-Start, a TCP host, say, host A, would indicate its
>    desired initial sending rate in packets per second in a Quick Start
>    Request option in the IP header of the initial TCP SYN or SYN/ACK
>    packet. Each router in turn could either approve the specified
>    initial rate, reduce the specified initial rate, or indicate that
>    nothing above the default initial rate for that protocol would be
>    allowed.  The Quick-Start mechanism also can determine if there are
>    routers along the path that do not understand the Quick Start Request
>    option, or have not agreed to the initial rate described in the
>    option. If all of the routers along the path have agreed to the
>    initial rate in the Quick-Start Request, then TCP host B communicates
>    this to TCP host A in a transport-level Quick-Start Response in the
>    answering SYN/ACK or ACK packet.  Quick-Start is designed to allow
>    TCP connections to use high initial windows in circumstances when
>    there is significant unused bandwidth along the path, and all of the
>    routers along the path support the Quick-Start Request.  We note that
>    this is currently not a mature proposal, but an outline of one
>    possible path of development in the IP protocol, and a request for
>    feedback from the community.
>
> ----
> -- Forwarded Message
>
> Date:    Mon, 24 Jun 2002 09:31:48 -0700
> From:    Sally Floyd <floyd@icir.org>
> To:      internet-drafts@ietf.org
> cc:      Amit Jain <ajain@netscaler.com>
> Subject: Internet drafts submission of draft-amit-quick-start-00a.txt
>
> I would like to submit this internet-draft.
>
> http://www.icir.org/floyd/papers/draft-amit-quick-start-00a.txt
> draft-amit-quick-sta.. 23-Jun-2002 22:13    36k
>
> I unfornately just realized this morning that the deadline was
> 9AM ET instead of 5PM ET, as I had been assuming, so probably
> you are not able to accept this submission.  But I thought I
> would try nevertheless: the file on the web server is dated
> last night, and therefore the draft itself was completed by
> the deadline...
>
> - Sally
>
> ------- End of Forwarded Message
>
> _______________________________________________
> 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@optimus.ietf.org  Fri Jun 28 04:35:00 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 EAA14475
	for <tsvwg-archive@odin.ietf.org>; Fri, 28 Jun 2002 04:35:00 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id EAA05146
	for tsvwg-archive@odin.ietf.org; Fri, 28 Jun 2002 04:35:45 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id EAA03880;
	Fri, 28 Jun 2002 04:21:15 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id EAA03851
	for <tsvwg@optimus.ietf.org>; Fri, 28 Jun 2002 04:21:13 -0400 (EDT)
Received: from mgw-x1.nokia.com (mgw-x1.nokia.com [131.228.20.21])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id EAA14135
	for <tsvwg@ietf.org>; Fri, 28 Jun 2002 04:20:22 -0400 (EDT)
From: dajiang.zhang@nokia.com
Received: from esvir05nok.ntc.nokia.com (esvir05nokt.ntc.nokia.com [172.21.143.37])
	by mgw-x1.nokia.com (Switch-2.2.1/Switch-2.2.0) with ESMTP id g5S8Jfd11869
	for <tsvwg@ietf.org>; Fri, 28 Jun 2002 11:19:41 +0300 (EET DST)
Received: from esebh003.NOE.Nokia.com (unverified) by esvir05nok.ntc.nokia.com
 (Content Technologies SMTPRS 4.2.5) with ESMTP id <T5bc2dcadbeac158f25c04@esvir05nok.ntc.nokia.com> for <tsvwg@ietf.org>;
 Fri, 28 Jun 2002 11:21:06 +0300
Received: from beebh001.NOE.Nokia.com ([172.28.19.38]) by esebh003.NOE.Nokia.com with Microsoft SMTPSVC(5.0.2195.4905);
	 Fri, 28 Jun 2002 11:21:05 +0300
Received: from beebe001.NOE.Nokia.com ([172.28.19.42]) by beebh001.NOE.Nokia.com with Microsoft SMTPSVC(5.0.2195.4905);
	 Fri, 28 Jun 2002 16:14:55 +0800
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: Fri, 28 Jun 2002 16:02:44 +0800
Message-ID: <E373C6E4507012439877385BCEBEDD876498B9@beebe001.NOE.Nokia.com>
Thread-Topic: [Tsvwg] Quick-Start for TCP and IP
Thread-Index: AcIcr4MooH2kK9LnT1O5SAKXb8iQLgBxMr/g
To: <tsvwg@ietf.org>
X-OriginalArrivalTime: 28 Jun 2002 08:14:55.0285 (UTC) FILETIME=[E2B0E650:01C21E7B]
Content-Transfer-Encoding: 8bit
X-MIME-Autoconverted: from quoted-printable to 8bit by optimus.ietf.org id EAA03852
Subject: [Tsvwg] SCTP Error Cause Code  conflict
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

In Stream Control Transmission Protocol (SCTP) Implementers Guide, in section 2.6.2,  there is error code 11 for  Restart of an association  with new addresses.

In   Stream Control Transmission Protocol (SCTP) Dynamic Address Reconfiguration, in section 3.3,   there is error code 0x000B for         Request to Delete Last Remaining IP Address.

Do they conflict?

Br
Dajiang Zhang

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



From daemon@optimus.ietf.org  Fri Jun 28 04:46:52 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 EAA14755
	for <tsvwg-archive@odin.ietf.org>; Fri, 28 Jun 2002 04:46:52 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id EAA05534
	for tsvwg-archive@odin.ietf.org; Fri, 28 Jun 2002 04:47:37 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id EAA05202;
	Fri, 28 Jun 2002 04:36:56 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id EAA05167
	for <tsvwg@optimus.ietf.org>; Fri, 28 Jun 2002 04:36:54 -0400 (EDT)
Received: from ilsa.franken.de (postfix@ilsa.franken.de [193.175.24.42])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id EAA14505
	for <tsvwg@ietf.org>; Fri, 28 Jun 2002 04:36:08 -0400 (EDT)
Received: from localhost (pD9E4F79C.dip.t-dialin.net [217.228.247.156])
	by ilsa.franken.de (Postfix) with ESMTP
	id 62403245D7; Fri, 28 Jun 2002 10:36:50 +0200 (CEST)
Date: Fri, 28 Jun 2002 10:35:53 +0200
Subject: Re: [Tsvwg] SCTP Error Cause Code  conflict
Content-Type: text/plain; charset=US-ASCII; format=flowed
Mime-Version: 1.0 (Apple Message framework v482)
Cc: <tsvwg@ietf.org>
To: dajiang.zhang@nokia.com
From: Michael Tuexen <Michael.Tuexen@icn.siemens.de>
In-Reply-To: <E373C6E4507012439877385BCEBEDD876498B9@beebe001.NOE.Nokia.com>
Message-Id: <0EE869E5-8A72-11D6-99AE-0030654C1AB6@icn.siemens.de>
Content-Transfer-Encoding: 7bit
X-Mailer: Apple Mail (2.482)
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

Dajizang Zhang,

these values conflict and in draft-ietf-tsvwg-addip-sctp-06.txt they
will be changed to

     Cause Code
     Value          Cause Code
     ---------      ----------------
     0x0100          Request to Delete Last Remaining IP Address.
     0x0101          Operation Refused Due to Resource Shortage.
     0x0102          Request to Delete Source IP Address.
     0x0103          Association Aborted due to illegal ASCONF-ACK

           Table 4: New Error Causes

Best regards
Michael

On Friday, June 28, 2002, at 10:02 AM, dajiang.zhang@nokia.com wrote:

> Hi, all
>
> In Stream Control Transmission Protocol (SCTP) Implementers Guide, in 
> section 2.6.2,  there is error code 11 for  Restart of an association  
> with new addresses.
>
> In   Stream Control Transmission Protocol (SCTP) Dynamic Address 
> Reconfiguration, in section 3.3,   there is error code 0x000B 
> for         Request to Delete Last Remaining IP Address.
>
> Do they conflict?
>
> Br
> Dajiang Zhang
>
> _______________________________________________
> tsvwg mailing list
> tsvwg@ietf.org
> https://www1.ietf.org/mailman/listinfo/tsvwg
>
Michael.Tuexen@micmac.franken.de


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



From daemon@optimus.ietf.org  Fri Jun 28 20:19:14 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 UAA29892
	for <tsvwg-archive@odin.ietf.org>; Fri, 28 Jun 2002 20:19:14 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id UAA16374
	for tsvwg-archive@odin.ietf.org; Fri, 28 Jun 2002 20:20:00 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id UAA15906;
	Fri, 28 Jun 2002 20:03:24 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id GAA11958
	for <tsvwg@optimus.ietf.org>; Fri, 28 Jun 2002 06:37:27 -0400 (EDT)
Received: from CNRI.Reston.VA.US (localhost [127.0.0.1])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id GAA17428;
	Fri, 28 Jun 2002 06:36:39 -0400 (EDT)
Message-Id: <200206281036.GAA17428@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: Fri, 28 Jun 2002 06:36:39 -0400
Subject: [Tsvwg] I-D ACTION:draft-sarolahti-tsvwg-tcp-frto-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.


	Title		: F-RTO: A TCP RTO Recovery Algorithm for Avoiding 
                          Unnecessary Retransmissions
	Author(s)	: P. Sarolahti, M. Kojo
	Filename	: draft-sarolahti-tsvwg-tcp-frto-00.txt
	Pages		: 11
	Date		: 27-Jun-02
	
Spurious retransmission timeouts (RTOs) cause suboptimal TCP
performance, because they often result in unnecessary retransmission
of the last window of data. This document describes the 'Forward RTO
Recovery' (F-RTO) algorithm for recovering from the TCP RTOs
efficiently even when the RTO was spurious. F-RTO is a TCP sender
only algorithm that does not require any TCP options to operate.
After retransmitting the first unacknowledged segment triggered by an
RTO, the F-RTO algorithm at a TCP sender monitors the incoming
acknowledgements to determine whether the timeout was spurious and to
decide whether to send new segments or retransmit unacknowledged
segments. The algorithm effectively avoids additional unnecessary
retransmissions and thereby improves TCP performance in case of a
spurious timeout.

A URL for this Internet-Draft is:
http://www.ietf.org/internet-drafts/draft-sarolahti-tsvwg-tcp-frto-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-sarolahti-tsvwg-tcp-frto-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-sarolahti-tsvwg-tcp-frto-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:	<20020627133057.I-D@ietf.org>

ENCODING mime
FILE /internet-drafts/draft-sarolahti-tsvwg-tcp-frto-00.txt

--OtherAccess
Content-Type: Message/External-body;
	name="draft-sarolahti-tsvwg-tcp-frto-00.txt";
	site="ftp.ietf.org";
	access-type="anon-ftp";
	directory="internet-drafts"

Content-Type: text/plain
Content-ID:	<20020627133057.I-D@ietf.org>

--OtherAccess--

--NextPart--




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



From daemon@optimus.ietf.org  Fri Jun 28 20:19:29 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 UAA29908
	for <tsvwg-archive@odin.ietf.org>; Fri, 28 Jun 2002 20:19:29 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id UAA16387
	for tsvwg-archive@odin.ietf.org; Fri, 28 Jun 2002 20:20:15 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id UAA15829;
	Fri, 28 Jun 2002 20:02:06 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id LAA19449
	for <tsvwg@optimus.ietf.org>; Tue, 25 Jun 2002 11:55:27 -0400 (EDT)
Received: from cougar.icir.org (cougar.icir.org [192.150.187.76])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id LAA11860
	for <tsvwg@ietf.org>; Tue, 25 Jun 2002 11:54:44 -0400 (EDT)
Received: from cougar.icir.org (localhost [127.0.0.1])
	by cougar.icir.org (8.11.6/8.11.3) with ESMTP id g5PFtHB12543;
	Tue, 25 Jun 2002 08:55:17 -0700 (PDT)
	(envelope-from floyd@cougar.icir.org)
Message-Id: <200206251555.g5PFtHB12543@cougar.icir.org>
To: Mohammed Ashraf <mohammed.asharaf@wipro.com>
cc: tsvwg@ietf.org, Amit Jain <ajain@netscaler.com>
From: Sally Floyd <floyd@icir.org>
Subject: Re: [Tsvwg] Quick-Start for TCP and IP 
Date: Tue, 25 Jun 2002 08:55:17 -0700
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

Ashraf -

>Here are my queries:
> 1. Draft says that Quick start request option will be included as a part
>of IP header, while  quick start response as part of
>      TCP header. Is it because the  quick start response need not be
>processed by IP?

Yep.  This follows the pattern of ECN, with the ECN field in the IP
header for routers to modify, and the feedback from receiver to sender
in the transport header for routers to ignore.  We will add a sentence 
about the parallel to the draft.

>2. I guess IP header option should keep 32 bit boundry (i.e multiple of 4
>bytes ). But in the case of quick start request, it  has got 5 bytes. Is
>the padding to be done here assumed?

Good point.  We will say something about that in the revised version
of the draft.  

If the community ended up thinking that Quick-Start was a good thing
to pursue further, then we would have to see if the longer option
field was a problem.  One would get a four-byte option by having a
k-bit Initial Rate field and an (8-k)-bit QS Nonce, for example,
but that doesn't really strike me as a viable alternative.

- Sally



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



From daemon@optimus.ietf.org  Fri Jun 28 20:21:07 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 UAA29951
	for <tsvwg-archive@odin.ietf.org>; Fri, 28 Jun 2002 20:21:07 -0400 (EDT)
Received: (from daemon@localhost)
	by optimus.ietf.org (8.9.1a/8.9.1) id UAA16434
	for tsvwg-archive@odin.ietf.org; Fri, 28 Jun 2002 20:21:53 -0400 (EDT)
Received: from optimus.ietf.org (localhost [127.0.0.1])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id UAA15904;
	Fri, 28 Jun 2002 20:03:24 -0400 (EDT)
Received: from ietf.org (odin [132.151.1.176])
	by optimus.ietf.org (8.9.1a/8.9.1) with ESMTP id GAA12205
	for <tsvwg@optimus.ietf.org>; Fri, 28 Jun 2002 06:38:35 -0400 (EDT)
Received: from CNRI.Reston.VA.US (localhost [127.0.0.1])
	by ietf.org (8.9.1a/8.9.1a) with ESMTP id GAA17767;
	Fri, 28 Jun 2002 06:37:48 -0400 (EDT)
Message-Id: <200206281037.GAA17767@ietf.org>
Mime-Version: 1.0
Content-Type: Multipart/Mixed; Boundary="NextPart"
To: IETF-Announce: ;
CC: tsvwg@ietf.org, sigtran@ietf.org
From: Internet-Drafts@ietf.org
Reply-to: Internet-Drafts@ietf.org
Date: Fri, 28 Jun 2002 06:37:48 -0400
Subject: [Tsvwg] I-D ACTION:draft-loughney-sctp-sig-prot-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.


	Title		: SCTP Layer for Transporting Signaling Protocols
	Author(s)	: J. Loughney, R. Sanchez
	Filename	: draft-loughney-sctp-sig-prot-00.txt
	Pages		: 7
	Date		: 27-Jun-02
	
The Stream Control Transport Protocol (SCTP) has been standardized to 
provide a reliable transport services for signaling protocols.  SCTP 
features a number of new features that are useful for transporting 
signaling protocols, but may require additional information on how to 
use them to transport existing signaling protocols.

A URL for this Internet-Draft is:
http://www.ietf.org/internet-drafts/draft-loughney-sctp-sig-prot-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-loughney-sctp-sig-prot-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-loughney-sctp-sig-prot-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:	<20020627133308.I-D@ietf.org>

ENCODING mime
FILE /internet-drafts/draft-loughney-sctp-sig-prot-00.txt

--OtherAccess
Content-Type: Message/External-body;
	name="draft-loughney-sctp-sig-prot-00.txt";
	site="ftp.ietf.org";
	access-type="anon-ftp";
	directory="internet-drafts"

Content-Type: text/plain
Content-ID:	<20020627133308.I-D@ietf.org>

--OtherAccess--

--NextPart--




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



