From netbackup-admin@advanced.org Sun Sep 1 06:50:43 2002 Received: from mailhost.advanced.org (root@mailhost.advanced.org [209.211.239.227]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id GAA13921 for ; Sun, 1 Sep 2002 06:50:43 -0400 (EDT) Received: from mailhost.advanced.org (localhost [127.0.0.1]) by mailhost.advanced.org (8.12.4/8.12.4/Debian-4) with ESMTP id g81AqGiU020738 for ; Sun, 1 Sep 2002 06:52:16 -0400 Date: Sun, 01 Sep 2002 06:52:16 -0400 Message-ID: <20020901105216.19995.29923.Mailman@mailhost.advanced.org> Subject: advanced.org mailing list memberships reminder From: mailman-owner@advanced.org To: ippm-archive@ietf.org X-No-Archive: yes List-Id: Multiple lists at advanced.org X-Ack: no Sender: netbackup-admin@advanced.org Errors-To: netbackup-admin@advanced.org X-BeenThere: netbackup@mailhost.advanced.org X-Mailman-Version: 2.0.12 Precedence: bulk This is a reminder, sent out once a month, about your advanced.org mailing list memberships. It includes your subscription info and how to use it to change it or unsubscribe from a list. You can visit the URLs to change your membership status or configuration, including unsubscribing, setting digest-style delivery or disabling delivery altogether (e.g., for a vacation), and so on. In addition to the URL interfaces, you can also use email to make such changes. For more info, send a message to the '-request' address of the list (for example, ippm-request@advanced.org) containing just the word 'help' in the message body, and an email message will be sent to you with instructions. If you have questions, problems, comments, etc, send them to mailman-owner@advanced.org. Thanks! Passwords for ippm-archive@lists.ietf.org: List Password // URL ---- -------- ippm@advanced.org jTPE http://mailhost.advanced.org/mailman/options/ippm/ippm-archive%40lists.ietf.org From ippm-admin@advanced.org Tue Sep 3 07:40:11 2002 Received: from mailhost.advanced.org (root@mailhost.advanced.org [209.211.239.227]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id HAA22345 for ; Tue, 3 Sep 2002 07:40:10 -0400 (EDT) Received: from mailhost.advanced.org (localhost [127.0.0.1]) by mailhost.advanced.org (8.12.4/8.12.4/Debian-4) with ESMTP id g83BaAiU024684; Tue, 3 Sep 2002 07:36:10 -0400 Received: from ietf.org (odin.ietf.org [132.151.1.176]) by mailhost.advanced.org (8.12.4/8.12.4/Debian-4) with ESMTP id g83BZQiU024119 for ; Tue, 3 Sep 2002 07:35:30 -0400 Received: from CNRI.Reston.VA.US (localhost [127.0.0.1]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id HAA21934; Tue, 3 Sep 2002 07:33:51 -0400 (EDT) Message-Id: <200209031133.HAA21934@ietf.org> Mime-Version: 1.0 Content-Type: Multipart/Mixed; Boundary="NextPart" To: IETF-Announce: ; Cc: ippm@advanced.org From: Internet-Drafts@ietf.org Reply-to: Internet-Drafts@ietf.org Subject: [ippm] I-D ACTION:draft-ietf-ippm-owdp-reqs-04.txt Sender: ippm-admin@advanced.org Errors-To: ippm-admin@advanced.org X-BeenThere: ippm@advanced.org X-Mailman-Version: 2.0.12 Precedence: bulk List-Help: List-Post: List-Subscribe: , List-Id: IETF IP Performance Metrics Working Group List List-Unsubscribe: , List-Archive: Date: Tue, 03 Sep 2002 07:33:50 -0400 --NextPart A New Internet-Draft is available from the on-line Internet-Drafts directories. This draft is a work item of the IP Performance Metrics Working Group of the IETF. Title : A One-way Active Measurement Protocol Requirements Author(s) : S. Shalunov, B. Teitelbaum Filename : draft-ietf-ippm-owdp-reqs-04.txt Pages : 10 Date : 2002-8-30 With growing availability of good time sources to network nodes, it becomes increasingly possible to measure one-way IP performance metrics with high precision. To do so in an interoperable manner, a common protocol for such measurements is required. This document specifies requirements for a one-way active measurement protocol (OWAMP) standard. The protocol can measure one-way delay, as well as other unidirectional characteristics, such as one-way loss. A URL for this Internet-Draft is: http://www.ietf.org/internet-drafts/draft-ietf-ippm-owdp-reqs-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-ippm-owdp-reqs-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-ippm-owdp-reqs-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: <2002-8-30145152.I-D@ietf.org> ENCODING mime FILE /internet-drafts/draft-ietf-ippm-owdp-reqs-04.txt --OtherAccess Content-Type: Message/External-body; name="draft-ietf-ippm-owdp-reqs-04.txt"; site="ftp.ietf.org"; access-type="anon-ftp"; directory="internet-drafts" Content-Type: text/plain Content-ID: <2002-8-30145152.I-D@ietf.org> --OtherAccess-- --NextPart-- _______________________________________________ ippm mailing list ippm@advanced.org http://mailhost.advanced.org/mailman/listinfo/ippm From ippm-admin@advanced.org Tue Sep 3 08:25:48 2002 Received: from mailhost.advanced.org (root@mailhost.advanced.org [209.211.239.227]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id HAA22344 for ; Tue, 3 Sep 2002 07:40:10 -0400 (EDT) Received: from mailhost.advanced.org (localhost [127.0.0.1]) by mailhost.advanced.org (8.12.4/8.12.4/Debian-4) with ESMTP id g83Ba4iU024312; Tue, 3 Sep 2002 07:36:04 -0400 Received: from ietf.org (odin.ietf.org [132.151.1.176]) by mailhost.advanced.org (8.12.4/8.12.4/Debian-4) with ESMTP id g83BZLiU023933 for ; Tue, 3 Sep 2002 07:35:22 -0400 Received: from CNRI.Reston.VA.US (localhost [127.0.0.1]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id HAA21918; Tue, 3 Sep 2002 07:33:46 -0400 (EDT) Message-Id: <200209031133.HAA21918@ietf.org> Mime-Version: 1.0 Content-Type: Multipart/Mixed; Boundary="NextPart" To: IETF-Announce: ; Cc: ippm@advanced.org From: Internet-Drafts@ietf.org Reply-to: Internet-Drafts@ietf.org Subject: [ippm] I-D ACTION:draft-ietf-ippm-owdp-05.txt Sender: ippm-admin@advanced.org Errors-To: ippm-admin@advanced.org X-BeenThere: ippm@advanced.org X-Mailman-Version: 2.0.12 Precedence: bulk List-Help: List-Post: List-Subscribe: , List-Id: IETF IP Performance Metrics Working Group List List-Unsubscribe: , List-Archive: Date: Tue, 03 Sep 2002 07:33:45 -0400 --NextPart A New Internet-Draft is available from the on-line Internet-Drafts directories. This draft is a work item of the IP Performance Metrics Working Group of the IETF. Title : A One-way Active Measurement Protocol Author(s) : S. Shalunov, B. Teitelbaum, M. Zekauskas Filename : draft-ietf-ippm-owdp-05.txt Pages : 27 Date : 2002-8-30 The IETF IP Performance Metrics (IPPM) working group has proposed draft standard metrics for one-way packet delay [RFC2679] and loss [RFC 2680] across Internet paths. Although there are now several measurement platforms that implement collection of these metrics [SURVEYOR], [RIPE], there is not currently a standard that would permit initiation of test streams or exchange of packets to collect singleton metrics in an interoperable manner. With the increasingly wide availability of affordable global positioning system (GPS) and CDMA based time sources, hosts increasingly have available to them very accurate time sources--either directly or through their proximity to NTP primary (stratum 1) time servers. By standardizing a technique for collecting IPPM one-way active measurements, we hope to create an environment where IPPM metrics may be collected across a far broader mesh of Internet paths than is currently possible. One particularly compelling vision is of widespread deployment of open OWAMP servers that would make measurement of one-way delay as commonplace as measurement of round-trip time using an ICMP-based tool like ping. Additional design goals of OWAMP include being hard to detect and manipulate, security, logical separation of control and test functionality, and support for small test packets. OWAMP test traffic is hard to detect, because it is simply a stream of UDP packets from and to negotiated port numbers with potentially nothing static in the packets (size is negotiated, too). Additionally, OWAMP supports an encrypted mode, that further obscures the traffic, at the same time making it impossible to alter timestamps undetectably. Security features include optional authentication and/or encryption of control and test messages. These features may be useful to prevent unauthorized access to results or man-in-the-middle attackers who attempt to provide special treatment to OWAMP test streams or who attempt to modify sender A URL for this Internet-Draft is: http://www.ietf.org/internet-drafts/draft-ietf-ippm-owdp-05.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-ippm-owdp-05.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-ippm-owdp-05.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: <2002-8-30145142.I-D@ietf.org> ENCODING mime FILE /internet-drafts/draft-ietf-ippm-owdp-05.txt --OtherAccess Content-Type: Message/External-body; name="draft-ietf-ippm-owdp-05.txt"; site="ftp.ietf.org"; access-type="anon-ftp"; directory="internet-drafts" Content-Type: text/plain Content-ID: <2002-8-30145142.I-D@ietf.org> --OtherAccess-- --NextPart-- _______________________________________________ ippm mailing list ippm@advanced.org http://mailhost.advanced.org/mailman/listinfo/ippm From ippm-admin@advanced.org Sat Sep 7 08:16:50 2002 Received: from mailhost.advanced.org (root@mailhost.advanced.org [209.211.239.227]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id IAA11766 for ; Sat, 7 Sep 2002 08:16:49 -0400 (EDT) Received: from mailhost.advanced.org (localhost [127.0.0.1]) by mailhost.advanced.org (8.12.4/8.12.4/Debian-4) with ESMTP id g87CG3iU011038; Sat, 7 Sep 2002 08:16:03 -0400 Received: from netlx009.civ.utwente.nl (netlx009.civ.utwente.nl [130.89.1.91]) by mailhost.advanced.org (8.12.4/8.12.4/Debian-4) with ESMTP id g87CFQiU010849 for ; Sat, 7 Sep 2002 08:15:27 -0400 Received: from henrik (inbel241182.inbel.utwente.nl [130.89.241.182]) by netlx009.civ.utwente.nl (8.11.4/HKD) with SMTP id g87CFLf30328 for ; Sat, 7 Sep 2002 14:15:22 +0200 Message-Id: <200209071215.g87CFLf30328@netlx009.civ.utwente.nl> X-Sender: s9605363@mail.student.utwente.nl (Unverified) X-Mailer: QUALCOMM Windows Eudora Pro Version 4.0.1 To: ippm@advanced.org From: "H.E.Holland" Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" X-UTwente-MailScanner: Found to be clean Subject: [ippm] metric requirements Sender: ippm-admin@advanced.org Errors-To: ippm-admin@advanced.org X-BeenThere: ippm@advanced.org X-Mailman-Version: 2.0.12 Precedence: bulk List-Help: List-Post: List-Subscribe: , List-Id: IETF IP Performance Metrics Working Group List List-Unsubscribe: , List-Archive: Date: Sat, 07 Sep 2002 14:13:38 +0200 Dear all, Where might I find quantitative metric requirements for services such as VoIP? E.g. what are recommended values for one-way delay and packet loss for a VoIP service? Regard, Henrik. _______________________________________________ ippm mailing list ippm@advanced.org http://mailhost.advanced.org/mailman/listinfo/ippm From ippm-admin@advanced.org Sun Sep 8 04:39:41 2002 Received: from mailhost.advanced.org (root@mailhost.advanced.org [209.211.239.227]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id EAA04107 for ; Sun, 8 Sep 2002 04:39:41 -0400 (EDT) Received: from mailhost.advanced.org (localhost [127.0.0.1]) by mailhost.advanced.org (8.12.4/8.12.4/Debian-4) with ESMTP id g888d4iU009113; Sun, 8 Sep 2002 04:39:04 -0400 Received: from mgw-x4.nokia.com (mgw-x4.nokia.com [131.228.20.27]) by mailhost.advanced.org (8.12.4/8.12.4/Debian-4) with ESMTP id g888ctiV009019 (version=TLSv1/SSLv3 cipher=EDH-RSA-DES-CBC3-SHA bits=168 verify=FAIL) for ; Sun, 8 Sep 2002 04:38:57 -0400 Received: from esvir03nok.nokia.com (esvir03nokt.ntc.nokia.com [172.21.143.35]) by mgw-x4.nokia.com (Switch-2.2.1/Switch-2.2.0) with ESMTP id g888d1725375 for ; Sun, 8 Sep 2002 11:39:01 +0300 (EET DST) Received: from esebh001.NOE.Nokia.com (unverified) by esvir03nok.nokia.com (Content Technologies SMTPRS 4.2.5) with ESMTP id ; Sun, 8 Sep 2002 11:38:53 +0300 Received: from esebh005.NOE.Nokia.com ([172.21.138.86]) by esebh001.NOE.Nokia.com with Microsoft SMTPSVC(5.0.2195.5329); Sun, 8 Sep 2002 11:38:53 +0300 Received: from esebe004.NOE.Nokia.com ([172.21.138.44]) by esebh005.NOE.Nokia.com with Microsoft SMTPSVC(5.0.2195.5329); Sun, 8 Sep 2002 11:38:53 +0300 X-MimeOLE: Produced By Microsoft Exchange V6.0.6249.0 content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Subject: RE: [ippm] metric requirements Message-ID: <009CA59D1752DD448E07F8EB2F911757019F6EDE@esebe004.ntc.nokia.com> Thread-Topic: [ippm] metric requirements Thread-Index: AcJWaQt0ArEpF/AVS5ueMiO81zEqngAqVBKQ From: To: , X-OriginalArrivalTime: 08 Sep 2002 08:38:53.0461 (UTC) FILETIME=[29A72450:01C25713] Content-Transfer-Encoding: 8bit X-MIME-Autoconverted: from quoted-printable to 8bit by mailhost.advanced.org id g888ctiV009019 Sender: ippm-admin@advanced.org Errors-To: ippm-admin@advanced.org X-BeenThere: ippm@advanced.org X-Mailman-Version: 2.0.12 Precedence: bulk List-Help: List-Post: List-Subscribe: , List-Id: IETF IP Performance Metrics Working Group List List-Unsubscribe: , List-Archive: Date: Sun, 8 Sep 2002 11:38:52 +0300 Content-Transfer-Encoding: 8bit Greetings, ETSI project TIPHON has done work on this, please take a look at WG5 documents at http://docbox.etsi.org/tiphon/tiphon/07-drafts/wg5/_Published/. The relevant document series is numbered 101 329-X, where "X" is one or more digits. (Funky naming, that.) In those document I think there are pointers to other sources such as ITU-T. BR, Vilho > -----Original Message----- > From: ext H.E.Holland [mailto:h.e.holland@student.utwente.nl] > Sent: 07 September 2002 15:14 > To: ippm@advanced.org > Subject: [ippm] metric requirements > > > Dear all, > > Where might I find quantitative metric requirements for > services such as > VoIP? E.g. what are recommended values for one-way delay and > packet loss > for a VoIP service? > > Regard, > Henrik. > > _______________________________________________ > ippm mailing list > ippm@advanced.org > http://mailhost.advanced.org/mailman/listinfo/ippm > _______________________________________________ ippm mailing list ippm@advanced.org http://mailhost.advanced.org/mailman/listinfo/ippm From ippm-admin@advanced.org Sun Sep 8 22:46:05 2002 Received: from mailhost.advanced.org (root@mailhost.advanced.org [209.211.239.227]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id WAA14242 for ; Sun, 8 Sep 2002 22:46:05 -0400 (EDT) Received: from mailhost.advanced.org (localhost [127.0.0.1]) by mailhost.advanced.org (8.12.4/8.12.4/Debian-4) with ESMTP id g892k3iU026898; Sun, 8 Sep 2002 22:46:03 -0400 Received: from almso2.proxy.att.com (almso2.att.com [192.128.166.71]) by mailhost.advanced.org (8.12.4/8.12.4/Debian-4) with ESMTP id g892jViU026804 for ; Sun, 8 Sep 2002 22:45:31 -0400 Received: from hogpa.mt.att.com ([135.16.74.2]) by almso2.proxy.att.com (AT&T IPNS/MSO-4.0) with SMTP id g892jU3Q014755; Sun, 8 Sep 2002 22:45:30 -0400 (EDT) Received: from acmortonw.att.com by hogpa.mt.att.com (SMI-8.6/ATTEMS-1.4.1 sol2) id WAA04233; Sun, 8 Sep 2002 22:45:28 -0400 Message-Id: <5.1.0.14.0.20020908224052.02c43750@hogpa.mt.att.com> X-Sender: acm1@hogpa.mt.att.com X-Mailer: QUALCOMM Windows Eudora Version 5.1 To: , , From: Al Morton Subject: RE: [ippm] metric requirements In-Reply-To: <009CA59D1752DD448E07F8EB2F911757019F6EDE@esebe004.ntc.noki a.com> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii"; format=flowed Sender: ippm-admin@advanced.org Errors-To: ippm-admin@advanced.org X-BeenThere: ippm@advanced.org X-Mailman-Version: 2.0.12 Precedence: bulk List-Help: List-Post: List-Subscribe: , List-Id: IETF IP Performance Metrics Working Group List List-Unsubscribe: , List-Archive: Date: Sun, 08 Sep 2002 22:45:26 -0400 One ITU-T doc that Vilho refers to can be obtained in near-final form here: ftp://ftp.t1.org/T1A1/2001-a13/1a130351.ps (covers VoIP and more) Al At 11:38 AM 09/08/2002 +0300, vilho.raisanen@nokia.com wrote: >Greetings, > >ETSI project TIPHON has done work on this, please take a look at WG5 >documents at >http://docbox.etsi.org/tiphon/tiphon/07-drafts/wg5/_Published/. The >relevant document series is numbered 101 329-X, where "X" is one or more >digits. (Funky naming, that.) In those document I think there are pointers >to other sources such as ITU-T. > > BR, > Vilho > > > -----Original Message----- > > From: ext H.E.Holland [mailto:h.e.holland@student.utwente.nl] > > Sent: 07 September 2002 15:14 > > To: ippm@advanced.org > > Subject: [ippm] metric requirements > > > > > > Dear all, > > > > Where might I find quantitative metric requirements for > > services such as > > VoIP? E.g. what are recommended values for one-way delay and > > packet loss > > for a VoIP service? > > > > Regard, > > Henrik. > > > > _______________________________________________ > > ippm mailing list > > ippm@advanced.org > > http://mailhost.advanced.org/mailman/listinfo/ippm > > > >_______________________________________________ >ippm mailing list >ippm@advanced.org >http://mailhost.advanced.org/mailman/listinfo/ippm _______________________________________________ ippm mailing list ippm@advanced.org http://mailhost.advanced.org/mailman/listinfo/ippm From ippm-admin@advanced.org Wed Sep 11 18:52:46 2002 Received: from mailhost.advanced.org (root@mailhost.advanced.org [209.211.239.227]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id SAA27639 for ; Wed, 11 Sep 2002 18:52:46 -0400 (EDT) Received: from mailhost.advanced.org (localhost [127.0.0.1]) by mailhost.advanced.org (8.12.4/8.12.4/Debian-4) with ESMTP id g8BMr4iU004899; Wed, 11 Sep 2002 18:53:04 -0400 Received: from brixcorp2.brixnet.com ([63.122.27.34]) by mailhost.advanced.org (8.12.4/8.12.4/Debian-4) with ESMTP id g8BKUFiU007675 for ; Wed, 11 Sep 2002 16:30:15 -0400 Received: by mail.brixnet.com with Internet Mail Service (5.5.2656.59) id ; Wed, 11 Sep 2002 16:30:13 -0400 Message-ID: <07B0D4912B83D31188F300A0C9F62EBB01BE8041@mail.brixnet.com> From: "Hedayat, Kaynam" To: "'H.E.Holland'" , ippm@advanced.org Subject: RE: [ippm] metric requirements MIME-Version: 1.0 X-Mailer: Internet Mail Service (5.5.2656.59) Content-Type: text/plain; charset="iso-8859-1" Sender: ippm-admin@advanced.org Errors-To: ippm-admin@advanced.org X-BeenThere: ippm@advanced.org X-Mailman-Version: 2.0.12 Precedence: bulk List-Help: List-Post: List-Subscribe: , List-Id: IETF IP Performance Metrics Working Group List List-Unsubscribe: , List-Archive: Date: Wed, 11 Sep 2002 16:30:11 -0400 You will find recommended one-way latency values in the IUT-T G.114 recommendation. Kaynam Hedayat Brix Networks -----Original Message----- From: H.E.Holland [mailto:h.e.holland@student.utwente.nl] Sent: Saturday, September 07, 2002 8:14 AM To: ippm@advanced.org Subject: [ippm] metric requirements Dear all, Where might I find quantitative metric requirements for services such as VoIP? E.g. what are recommended values for one-way delay and packet loss for a VoIP service? Regard, Henrik. _______________________________________________ ippm mailing list ippm@advanced.org http://mailhost.advanced.org/mailman/listinfo/ippm _______________________________________________ ippm mailing list ippm@advanced.org http://mailhost.advanced.org/mailman/listinfo/ippm From ippm-admin@advanced.org Wed Sep 11 19:37:14 2002 Received: from mailhost.advanced.org (root@mailhost.advanced.org [209.211.239.227]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id TAA28644 for ; Wed, 11 Sep 2002 19:37:14 -0400 (EDT) Received: from mailhost.advanced.org (localhost [127.0.0.1]) by mailhost.advanced.org (8.12.4/8.12.4/Debian-4) with ESMTP id g8BNW3iU009731; Wed, 11 Sep 2002 19:32:03 -0400 Received: from green.its.bldrdoc.gov (green.its.bldrdoc.gov [132.163.64.205]) by mailhost.advanced.org (8.12.4/8.12.4/Debian-4) with ESMTP id g8BNVGiU009458 for ; Wed, 11 Sep 2002 19:31:19 -0400 Received: from flagstaff (flagstaff.its.bldrdoc.gov [132.163.64.249]) by green.its.bldrdoc.gov (8.9.2/8.9.2) with SMTP id RAA14387; Wed, 11 Sep 2002 17:29:52 -0600 (MDT) Message-ID: <012401c259eb$b4d77180$f940a384@its.bldrdoc.gov> From: "R. S. Bloomfield" To: "Hedayat, Kaynam" , "'H.E.Holland'" , References: <07B0D4912B83D31188F300A0C9F62EBB01BE8041@mail.brixnet.com> Subject: Re: [ippm] metric requirements MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit X-Priority: 3 X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook Express 5.50.4522.1200 X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4522.1200 Sender: ippm-admin@advanced.org Errors-To: ippm-admin@advanced.org X-BeenThere: ippm@advanced.org X-Mailman-Version: 2.0.12 Precedence: bulk List-Help: List-Post: List-Subscribe: , List-Id: IETF IP Performance Metrics Working Group List List-Unsubscribe: , List-Archive: Date: Wed, 11 Sep 2002 17:33:56 -0600 Content-Transfer-Encoding: 7bit Perhaps of interest to you would be ITU-T Recommendation G.1010 (2001), "End-user multimedia QoS categories," and its normative (ITU-T F.700 (2000), G.131 (1996), G.114 (2000), and ETSI TS 101329-2 (2002)) and informative (Appendix II/Bibliography) references. Also of possible interest would be Y.1541 (2002), "Network performance objectives for IP-based services" (ftp://ftp.t1.org/Pub/T1S1/T1S1.0/2002/2s101490.pdf). Regards, - Randy Bloomfield, NTIA/ITS.P ----- Original Message ----- From: "Hedayat, Kaynam" To: "'H.E.Holland'" ; Sent: Wednesday, September 11, 2002 2:30 PM Subject: RE: [ippm] metric requirements You will find recommended one-way latency values in the IUT-T G.114 recommendation. Kaynam Hedayat Brix Networks -----Original Message----- From: H.E.Holland [mailto:h.e.holland@student.utwente.nl] Sent: Saturday, September 07, 2002 8:14 AM To: ippm@advanced.org Subject: [ippm] metric requirements Dear all, Where might I find quantitative metric requirements for services such as VoIP? E.g. what are recommended values for one-way delay and packet loss for a VoIP service? Regard, Henrik. _______________________________________________ ippm mailing list ippm@advanced.org http://mailhost.advanced.org/mailman/listinfo/ippm _______________________________________________ ippm mailing list ippm@advanced.org http://mailhost.advanced.org/mailman/listinfo/ippm _______________________________________________ ippm mailing list ippm@advanced.org http://mailhost.advanced.org/mailman/listinfo/ippm From ippm-admin@advanced.org Fri Sep 13 05:05:01 2002 Received: from mailhost.advanced.org (root@mailhost.advanced.org [209.211.239.227]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id FAA29113 for ; Fri, 13 Sep 2002 05:05:00 -0400 (EDT) Received: from mailhost.advanced.org (localhost [127.0.0.1]) by mailhost.advanced.org (8.12.4/8.12.4/Debian-4) with ESMTP id g8D8o5iU024973; Fri, 13 Sep 2002 04:50:05 -0400 Received: from parsmtp1.rd.francetelecom.com (parsmtp1.rd.francetelecom.com [194.167.105.13]) by mailhost.advanced.org (8.12.4/8.12.4/Debian-4) with ESMTP id g8D8n1iU024698 for ; Fri, 13 Sep 2002 04:49:02 -0400 Received: from LANMHS20.rd.francetelecom.fr ([10.193.21.60]) by parsmtp1.rd.francetelecom.com with Microsoft SMTPSVC(5.0.2195.4905); Fri, 13 Sep 2002 10:49:00 +0200 content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" X-MimeOLE: Produced By Microsoft Exchange V6.0.5762.3 Message-ID: Thread-Topic: Sending in the reporting MIB Thread-Index: AcIVvqMVwP/tgWTcT2m2AVoQB89mAwAjXUAAEQlwOcA= From: "STEPHAN Emile FTRD/DAC/LAN" To: "IPPM Working Group" , , , X-OriginalArrivalTime: 13 Sep 2002 08:49:00.0867 (UTC) FILETIME=[67C2B130:01C25B02] Content-Transfer-Encoding: 8bit X-MIME-Autoconverted: from quoted-printable to 8bit by mailhost.advanced.org id g8D8n1iU024698 Subject: [ippm] draft-stephan-ippm-spatial-metrics-00.txt Sender: ippm-admin@advanced.org Errors-To: ippm-admin@advanced.org X-BeenThere: ippm@advanced.org X-Mailman-Version: 2.0.12 Precedence: bulk List-Help: List-Post: List-Subscribe: , List-Id: IETF IP Performance Metrics Working Group List List-Unsubscribe: , List-Archive: Date: Fri, 13 Sep 2002 10:49:00 +0200 Content-Transfer-Encoding: 8bit Hello, An new I-D related to measurement is available online http://www.ietf.org/internet-drafts/draft-stephan-ippm-spatial-metrics-00.txt Abstract: The IETF IP Performance Metrics (IPPM) working group has standardized metrics for measuring end to end performance. Measurements system scope is often limited to administrative boundaries. This memo defines spatial metrics both for measuring end-to-end network performance using aggregation of sequence of network measures and for measuring the performance of segment of an IP path trajectory. It distinguishes clearly the decomposition of one end-to-end measure result in a sequence of per hop results from the aggregation of a sequence of per hop measure results in an end-to-end result. best regards. Emile _______________________________________________ ippm mailing list ippm@advanced.org http://mailhost.advanced.org/mailman/listinfo/ippm From ippm-admin@advanced.org Tue Sep 17 17:45:20 2002 Received: from mailhost.advanced.org (root@mailhost.advanced.org [209.211.239.227]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id RAA01623 for ; Tue, 17 Sep 2002 17:45:19 -0400 (EDT) Received: from mailhost.advanced.org (localhost [127.0.0.1]) by mailhost.advanced.org (8.12.4/8.12.4/Debian-4) with ESMTP id g8HLh3iU024048; Tue, 17 Sep 2002 17:43:03 -0400 Received: from netlx010.civ.utwente.nl (netlx010.civ.utwente.nl [130.89.1.92]) by mailhost.advanced.org (8.12.4/8.12.4/Debian-4) with ESMTP id g8HLgRiU023856 for ; Tue, 17 Sep 2002 17:42:28 -0400 Received: from henrik (inbel241182.inbel.utwente.nl [130.89.241.182]) by netlx010.civ.utwente.nl (8.11.4/HKD) with SMTP id g8HLgJQ19312; Tue, 17 Sep 2002 23:42:19 +0200 Message-Id: <200209172142.g8HLgJQ19312@netlx010.civ.utwente.nl> X-Sender: s9605363@mail.student.utwente.nl X-Mailer: QUALCOMM Windows Eudora Pro Version 4.0.1 To: stanislav shalunov , victor.reijs@surfnet.nl From: "H.E.Holland" Subject: Re: [ippm] concatenation of QoS parameters Cc: IPPM Working Group In-Reply-To: <87pu389erp.fsf@cain.internet2.edu> References: <3C6B8CED.72D8475D@surfnet.nl> <3C6B8CED.72D8475D@surfnet.nl> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" X-UTwente-MailScanner: Found to be clean Sender: ippm-admin@advanced.org Errors-To: ippm-admin@advanced.org X-BeenThere: ippm@advanced.org X-Mailman-Version: 2.0.12 Precedence: bulk List-Help: List-Post: List-Subscribe: , List-Id: IETF IP Performance Metrics Working Group List List-Unsubscribe: , List-Archive: Date: Tue, 17 Sep 2002 23:40:19 +0200 Dear all, I found this discussion about concatenating parameters from a while back. Little is said however about concatenating distributions. For example, I asume that I can simply add medians of delay samples for adjacent independently measured links. And the percentiles. Is this correct? The mail below speaks of one-way delay "averages" being added over various domains where average signifies "mean", but mean "can't" be calculated (officially anyway) since losses aren't acounted for. Thanks, Henrik. At 12:03 14-2-02 -0500, stanislav shalunov wrote: >This is probably outside of scope of IPPM WG activities (and in the >realm of a statistics), but: Everything hinges on the assumption of >independent distribution of the delay variables for different domains. >On one hand, since the next domain doesn't explicitly know about >treatment in the previous one you could say that the independence >assumption should be true. On the other hand, if there exist global >congestion oscillation phenomena (and why wouldn't they?) the >assumption would be incorrect. > >Further, you do not specify your metrics. Averages, for example, >would be much easier to analyze in this way than medians. > >Victor Reijs writes: > >> . OWD. I assume, I can just add figures in case of concatenation of >> domains. > >Sum is correct for averages regardless of independence. > >> . IPDV. This is a difficult one. I would tend to say one can not just >> added the figures. I once heard a convolution is needed here. What are >> your ideas? > >Sum is correct for variances provided that there is independence. > >> . IP Packet Loss. > >Loss probability, assuming independence, is, of course, 1-\Pi(1-p_i). >Accidentally, for small values of p_i the first order approximation of >this quantity happens to be \Sigma p_i. > >> . IP Loss patterns. > >No numeric parameter. > >> . IP packet reordering: > >No defined metric and no defined model. > >> The above was for the millisec range, perhaps other ideas are needed >> in microsec range. > >Scale doesn't matter. > >-- >Stanislav Shalunov http://www.internet2.edu/~shalunov/ > >"You wake me up early in the morning to tell me I am right? Please >wait until I am wrong." -- John von Neumann, on being phoned at 10 a.m. >_______________________________________________ >ippm mailing list >ippm@advanced.org >http://mailhost.advanced.org/mailman/listinfo/ippm > _______________________________________________ ippm mailing list ippm@advanced.org http://mailhost.advanced.org/mailman/listinfo/ippm From ippm-admin@advanced.org Tue Sep 17 21:24:35 2002 Received: from mailhost.advanced.org (root@mailhost.advanced.org [209.211.239.227]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id VAA05422 for ; Tue, 17 Sep 2002 21:24:35 -0400 (EDT) Received: from mailhost.advanced.org (localhost [127.0.0.1]) by mailhost.advanced.org (8.12.4/8.12.4/Debian-4) with ESMTP id g8I1I4iU029551; Tue, 17 Sep 2002 21:18:04 -0400 Received: from telchemy.com ([66.155.69.251]) by mailhost.advanced.org (8.12.4/8.12.4/Debian-4) with ESMTP id g8I1HbiU029361 for ; Tue, 17 Sep 2002 21:17:37 -0400 Received: from alansnotebook [66.156.11.103] by telchemy.com with ESMTP (SMTPD32-7.11) id A46330201F4; Tue, 17 Sep 2002 21:18:27 -0400 From: "Alan Clark" To: "H.E.Holland" , "stanislav shalunov" , Cc: "IPPM Working Group" Subject: RE: [ippm] concatenation of QoS parameters Message-ID: MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit X-Priority: 3 (Normal) X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook IMO, Build 9.0.2416 (9.0.2911.0) X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4133.2400 Importance: Normal In-Reply-To: <200209172142.g8HLgJQ19312@netlx010.civ.utwente.nl> Sender: ippm-admin@advanced.org Errors-To: ippm-admin@advanced.org X-BeenThere: ippm@advanced.org X-Mailman-Version: 2.0.12 Precedence: bulk List-Help: List-Post: List-Subscribe: , List-Id: IETF IP Performance Metrics Working Group List List-Unsubscribe: , List-Archive: Date: Tue, 17 Sep 2002 21:21:03 -0400 Content-Transfer-Encoding: 7bit My apologies for restating the obvious, if already stated .. typical traces of loss and jitter are clearly non-stationary. Jitter often appears to have a roughly constant "background" level with superimposed short term delay variations (100mS to several seconds) due to transient congestion on LAN and access links. Network packet loss is due to transient congestion, route changes, link failures and similar events. Contrary to popular belief, bursts are often long and sparse - e.g. several seconds of 20% loss - rather than the short bursts of consecutive loss often described. For multimedia applications this is a critical distinction as long bursts of 20-30% loss can be more objectionable than short dense loss periods. If you assume that both loss and jitter have some fairly constant background rate/ level with transient events then it would seem to be necessary to separately combine the background and transient parameters. e.g. for packet loss: (i) compute gap/background loss probability using independence assumption (ii) compute burst probability, assuming that the probability of bursts occurring on network segments is independent, and combine burst density/length stats. Alan Clark -----Original Message----- From: ippm-admin@advanced.org [mailto:ippm-admin@advanced.org]On Behalf Of H.E.Holland Sent: Tuesday, September 17, 2002 5:40 PM To: stanislav shalunov; victor.reijs@surfnet.nl Cc: IPPM Working Group Subject: Re: [ippm] concatenation of QoS parameters Dear all, I found this discussion about concatenating parameters from a while back. Little is said however about concatenating distributions. For example, I asume that I can simply add medians of delay samples for adjacent independently measured links. And the percentiles. Is this correct? The mail below speaks of one-way delay "averages" being added over various domains where average signifies "mean", but mean "can't" be calculated (officially anyway) since losses aren't acounted for. Thanks, Henrik. At 12:03 14-2-02 -0500, stanislav shalunov wrote: >This is probably outside of scope of IPPM WG activities (and in the >realm of a statistics), but: Everything hinges on the assumption of >independent distribution of the delay variables for different domains. >On one hand, since the next domain doesn't explicitly know about >treatment in the previous one you could say that the independence >assumption should be true. On the other hand, if there exist global >congestion oscillation phenomena (and why wouldn't they?) the >assumption would be incorrect. > >Further, you do not specify your metrics. Averages, for example, >would be much easier to analyze in this way than medians. > >Victor Reijs writes: > >> . OWD. I assume, I can just add figures in case of concatenation of >> domains. > >Sum is correct for averages regardless of independence. > >> . IPDV. This is a difficult one. I would tend to say one can not just >> added the figures. I once heard a convolution is needed here. What are >> your ideas? > >Sum is correct for variances provided that there is independence. > >> . IP Packet Loss. > >Loss probability, assuming independence, is, of course, 1-\Pi(1-p_i). >Accidentally, for small values of p_i the first order approximation of >this quantity happens to be \Sigma p_i. > >> . IP Loss patterns. > >No numeric parameter. > >> . IP packet reordering: > >No defined metric and no defined model. > >> The above was for the millisec range, perhaps other ideas are needed >> in microsec range. > >Scale doesn't matter. > >-- >Stanislav Shalunov http://www.internet2.edu/~shalunov/ > >"You wake me up early in the morning to tell me I am right? Please >wait until I am wrong." -- John von Neumann, on being phoned at 10 a.m. >_______________________________________________ >ippm mailing list >ippm@advanced.org >http://mailhost.advanced.org/mailman/listinfo/ippm > _______________________________________________ ippm mailing list ippm@advanced.org http://mailhost.advanced.org/mailman/listinfo/ippm _______________________________________________ ippm mailing list ippm@advanced.org http://mailhost.advanced.org/mailman/listinfo/ippm From ippm-admin@advanced.org Wed Sep 18 01:42:11 2002 Received: from mailhost.advanced.org (root@mailhost.advanced.org [209.211.239.227]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id BAA09852 for ; Wed, 18 Sep 2002 01:42:11 -0400 (EDT) Received: from mailhost.advanced.org (localhost [127.0.0.1]) by mailhost.advanced.org (8.12.4/8.12.4/Debian-4) with ESMTP id g8I5g3iU002038; Wed, 18 Sep 2002 01:42:03 -0400 Received: from smtp.noos.fr (racine.noos.net [212.198.2.71]) by mailhost.advanced.org (8.12.4/8.12.4/Debian-4) with ESMTP id g8I5fWiV001944 (version=TLSv1/SSLv3 cipher=EDH-RSA-DES-CBC3-SHA bits=168 verify=FAIL) for ; Wed, 18 Sep 2002 01:41:33 -0400 Received: (qmail 30182375 invoked by uid 0); 18 Sep 2002 05:41:30 -0000 Received: from unknown (HELO LIP60MU1PINK59) ([212.198.208.249]) (envelope-sender ) by 212.198.2.71 (qmail-ldap-1.03) with SMTP for ; 18 Sep 2002 05:41:30 -0000 From: =?us-ascii?Q?Kave_Salamatian?= To: "Alan Clark" , "H.E.Holland" , "stanislav shalunov" , Cc: "IPPM Working Group" Subject: RE: [ippm] concatenation of QoS parameters Message-ID: MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit X-Priority: 3 (Normal) X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook IMO, Build 9.0.2416 (9.0.2911.0) X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000 In-Reply-To: Importance: Normal Sender: ippm-admin@advanced.org Errors-To: ippm-admin@advanced.org X-BeenThere: ippm@advanced.org X-Mailman-Version: 2.0.12 Precedence: bulk List-Help: List-Post: List-Subscribe: , List-Id: IETF IP Performance Metrics Working Group List List-Unsubscribe: , List-Archive: Date: Wed, 18 Sep 2002 07:41:06 +0200 Content-Transfer-Encoding: 7bit I agree with Alan about the structure of losses. Losses can be described by period of high loss probability followed by period of low loss probability. This is explained in "Hidden Markov Modeling for network communication channels, M.R Salamatian, S. Vaton, SIGMETRICS 2001, Cambridge, Massachusett." (can be downloaded from www-rp.lip6.fr). For delay we have a similar description. That is under writing. However I don't agree with Alan about Non stationnarity at all scales. Globally speaking, stationarity cannot be proven of disproven when you work with empirical data. Stationarity is only a property of the model that you decide to use and that is best suited to your application. In fact stationarity is a bet about the fact that what you have observed into your observation window will be reproduced (with same statistical parameters) out of the observation window. A single observation may be described as well by a stationnary model as a non stationnary. If one go for a stationnary model he may have to manage Long Range dependences that will be necessary to explain mean level variation. If he go for non stationarity he need to track the change of the underlying process, but he may use short memory model with memory scale at the level of his tracking agility. Which one is better depend on the application ! Meaning that loss and jitter are not non-stationnary by essence, but they may be subjectively stationnary or not. Kave Salamatian -----Message d'origine----- De : ippm-admin@advanced.org [mailto:ippm-admin@advanced.org]De la part de Alan Clark Envoye : mercredi 18 septembre 2002 03:21 A : H.E.Holland; stanislav shalunov; victor.reijs@surfnet.nl Cc : IPPM Working Group Objet : RE: [ippm] concatenation of QoS parameters My apologies for restating the obvious, if already stated .. typical traces of loss and jitter are clearly non-stationary. Jitter often appears to have a roughly constant "background" level with superimposed short term delay variations (100mS to several seconds) due to transient congestion on LAN and access links. Network packet loss is due to transient congestion, route changes, link failures and similar events. Contrary to popular belief, bursts are often long and sparse - e.g. several seconds of 20% loss - rather than the short bursts of consecutive loss often described. For multimedia applications this is a critical distinction as long bursts of 20-30% loss can be more objectionable than short dense loss periods. If you assume that both loss and jitter have some fairly constant background rate/ level with transient events then it would seem to be necessary to separately combine the background and transient parameters. e.g. for packet loss: (i) compute gap/background loss probability using independence assumption (ii) compute burst probability, assuming that the probability of bursts occurring on network segments is independent, and combine burst density/length stats. Alan Clark -----Original Message----- From: ippm-admin@advanced.org [mailto:ippm-admin@advanced.org]On Behalf Of H.E.Holland Sent: Tuesday, September 17, 2002 5:40 PM To: stanislav shalunov; victor.reijs@surfnet.nl Cc: IPPM Working Group Subject: Re: [ippm] concatenation of QoS parameters Dear all, I found this discussion about concatenating parameters from a while back. Little is said however about concatenating distributions. For example, I asume that I can simply add medians of delay samples for adjacent independently measured links. And the percentiles. Is this correct? The mail below speaks of one-way delay "averages" being added over various domains where average signifies "mean", but mean "can't" be calculated (officially anyway) since losses aren't acounted for. Thanks, Henrik. At 12:03 14-2-02 -0500, stanislav shalunov wrote: >This is probably outside of scope of IPPM WG activities (and in the >realm of a statistics), but: Everything hinges on the assumption of >independent distribution of the delay variables for different domains. >On one hand, since the next domain doesn't explicitly know about >treatment in the previous one you could say that the independence >assumption should be true. On the other hand, if there exist global >congestion oscillation phenomena (and why wouldn't they?) the >assumption would be incorrect. > >Further, you do not specify your metrics. Averages, for example, >would be much easier to analyze in this way than medians. > >Victor Reijs writes: > >> . OWD. I assume, I can just add figures in case of concatenation of >> domains. > >Sum is correct for averages regardless of independence. > >> . IPDV. This is a difficult one. I would tend to say one can not just >> added the figures. I once heard a convolution is needed here. What are >> your ideas? > >Sum is correct for variances provided that there is independence. > >> . IP Packet Loss. > >Loss probability, assuming independence, is, of course, 1-\Pi(1-p_i). >Accidentally, for small values of p_i the first order approximation of >this quantity happens to be \Sigma p_i. > >> . IP Loss patterns. > >No numeric parameter. > >> . IP packet reordering: > >No defined metric and no defined model. > >> The above was for the millisec range, perhaps other ideas are needed >> in microsec range. > >Scale doesn't matter. > >-- >Stanislav Shalunov http://www.internet2.edu/~shalunov/ > >"You wake me up early in the morning to tell me I am right? Please >wait until I am wrong." -- John von Neumann, on being phoned at 10 a.m. >_______________________________________________ >ippm mailing list >ippm@advanced.org >http://mailhost.advanced.org/mailman/listinfo/ippm > _______________________________________________ ippm mailing list ippm@advanced.org http://mailhost.advanced.org/mailman/listinfo/ippm _______________________________________________ ippm mailing list ippm@advanced.org http://mailhost.advanced.org/mailman/listinfo/ippm _______________________________________________ ippm mailing list ippm@advanced.org http://mailhost.advanced.org/mailman/listinfo/ippm From ippm-admin@advanced.org Wed Sep 18 07:04:30 2002 Received: from mailhost.advanced.org (root@mailhost.advanced.org [209.211.239.227]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id HAA08234 for ; Wed, 18 Sep 2002 07:04:30 -0400 (EDT) Received: from mailhost.advanced.org (localhost [127.0.0.1]) by mailhost.advanced.org (8.12.4/8.12.4/Debian-4) with ESMTP id g8IAu3iU013114; Wed, 18 Sep 2002 06:56:03 -0400 Received: from telchemy.com ([66.155.69.251]) by mailhost.advanced.org (8.12.4/8.12.4/Debian-4) with ESMTP id g8IAtCiU013014 for ; Wed, 18 Sep 2002 06:55:12 -0400 Received: from alansnotebook [66.156.8.160] by telchemy.com with ESMTP (SMTPD32-7.11) id ABC14BC01D8; Wed, 18 Sep 2002 06:56:01 -0400 From: "Alan Clark" To: "Kave Salamatian" , "H.E.Holland" , "stanislav shalunov" , Cc: "IPPM Working Group" Subject: RE: [ippm] concatenation of QoS parameters Message-ID: MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit X-Priority: 3 (Normal) X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook IMO, Build 9.0.2416 (9.0.2911.0) X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4133.2400 In-Reply-To: Importance: Normal Sender: ippm-admin@advanced.org Errors-To: ippm-admin@advanced.org X-BeenThere: ippm@advanced.org X-Mailman-Version: 2.0.12 Precedence: bulk List-Help: List-Post: List-Subscribe: , List-Id: IETF IP Performance Metrics Working Group List List-Unsubscribe: , List-Archive: Date: Wed, 18 Sep 2002 06:58:36 -0400 Content-Transfer-Encoding: 7bit Kave I had some difficulty finding the paper you reference on the lip6 web site - please verify it's location. My IPtel 2001 paper describes the use of a 4 state Markov Model for describing burst packet loss distribution and may be of interest to you http://www.iptel.org/2001/pg/final_program/21.pdf Whilst I have to agree with your philosophical observation concerning stationarity I think you are agreeing that traces do show transient packet loss and short term delay changes. The main subject of the posting was to address the issue of how you would combine QoS parameters from two network segments - it would be intersting to hear your views on this Regards Alan -----Original Message----- From: Kave Salamatian [mailto:salamat@rp.lip6.fr] Sent: Wednesday, September 18, 2002 1:41 AM To: Alan Clark; H.E.Holland; stanislav shalunov; victor.reijs@surfnet.nl Cc: IPPM Working Group Subject: RE: [ippm] concatenation of QoS parameters I agree with Alan about the structure of losses. Losses can be described by period of high loss probability followed by period of low loss probability. This is explained in "Hidden Markov Modeling for network communication channels, M.R Salamatian, S. Vaton, SIGMETRICS 2001, Cambridge, Massachusett." (can be downloaded from www-rp.lip6.fr). For delay we have a similar description. That is under writing. However I don't agree with Alan about Non stationnarity at all scales. Globally speaking, stationarity cannot be proven of disproven when you work with empirical data. Stationarity is only a property of the model that you decide to use and that is best suited to your application. In fact stationarity is a bet about the fact that what you have observed into your observation window will be reproduced (with same statistical parameters) out of the observation window. A single observation may be described as well by a stationnary model as a non stationnary. If one go for a stationnary model he may have to manage Long Range dependences that will be necessary to explain mean level variation. If he go for non stationarity he need to track the change of the underlying process, but he may use short memory model with memory scale at the level of his tracking agility. Which one is better depend on the application ! Meaning that loss and jitter are not non-stationnary by essence, but they may be subjectively stationnary or not. Kave Salamatian -----Message d'origine----- De : ippm-admin@advanced.org [mailto:ippm-admin@advanced.org]De la part de Alan Clark Envoye : mercredi 18 septembre 2002 03:21 A : H.E.Holland; stanislav shalunov; victor.reijs@surfnet.nl Cc : IPPM Working Group Objet : RE: [ippm] concatenation of QoS parameters My apologies for restating the obvious, if already stated .. typical traces of loss and jitter are clearly non-stationary. Jitter often appears to have a roughly constant "background" level with superimposed short term delay variations (100mS to several seconds) due to transient congestion on LAN and access links. Network packet loss is due to transient congestion, route changes, link failures and similar events. Contrary to popular belief, bursts are often long and sparse - e.g. several seconds of 20% loss - rather than the short bursts of consecutive loss often described. For multimedia applications this is a critical distinction as long bursts of 20-30% loss can be more objectionable than short dense loss periods. If you assume that both loss and jitter have some fairly constant background rate/ level with transient events then it would seem to be necessary to separately combine the background and transient parameters. e.g. for packet loss: (i) compute gap/background loss probability using independence assumption (ii) compute burst probability, assuming that the probability of bursts occurring on network segments is independent, and combine burst density/length stats. Alan Clark -----Original Message----- From: ippm-admin@advanced.org [mailto:ippm-admin@advanced.org]On Behalf Of H.E.Holland Sent: Tuesday, September 17, 2002 5:40 PM To: stanislav shalunov; victor.reijs@surfnet.nl Cc: IPPM Working Group Subject: Re: [ippm] concatenation of QoS parameters Dear all, I found this discussion about concatenating parameters from a while back. Little is said however about concatenating distributions. For example, I asume that I can simply add medians of delay samples for adjacent independently measured links. And the percentiles. Is this correct? The mail below speaks of one-way delay "averages" being added over various domains where average signifies "mean", but mean "can't" be calculated (officially anyway) since losses aren't acounted for. Thanks, Henrik. At 12:03 14-2-02 -0500, stanislav shalunov wrote: >This is probably outside of scope of IPPM WG activities (and in the >realm of a statistics), but: Everything hinges on the assumption of >independent distribution of the delay variables for different domains. >On one hand, since the next domain doesn't explicitly know about >treatment in the previous one you could say that the independence >assumption should be true. On the other hand, if there exist global >congestion oscillation phenomena (and why wouldn't they?) the >assumption would be incorrect. > >Further, you do not specify your metrics. Averages, for example, >would be much easier to analyze in this way than medians. > >Victor Reijs writes: > >> . OWD. I assume, I can just add figures in case of concatenation of >> domains. > >Sum is correct for averages regardless of independence. > >> . IPDV. This is a difficult one. I would tend to say one can not just >> added the figures. I once heard a convolution is needed here. What are >> your ideas? > >Sum is correct for variances provided that there is independence. > >> . IP Packet Loss. > >Loss probability, assuming independence, is, of course, 1-\Pi(1-p_i). >Accidentally, for small values of p_i the first order approximation of >this quantity happens to be \Sigma p_i. > >> . IP Loss patterns. > >No numeric parameter. > >> . IP packet reordering: > >No defined metric and no defined model. > >> The above was for the millisec range, perhaps other ideas are needed >> in microsec range. > >Scale doesn't matter. > >-- >Stanislav Shalunov http://www.internet2.edu/~shalunov/ > >"You wake me up early in the morning to tell me I am right? Please >wait until I am wrong." -- John von Neumann, on being phoned at 10 a.m. >_______________________________________________ >ippm mailing list >ippm@advanced.org >http://mailhost.advanced.org/mailman/listinfo/ippm > _______________________________________________ ippm mailing list ippm@advanced.org http://mailhost.advanced.org/mailman/listinfo/ippm _______________________________________________ ippm mailing list ippm@advanced.org http://mailhost.advanced.org/mailman/listinfo/ippm _______________________________________________ ippm mailing list ippm@advanced.org http://mailhost.advanced.org/mailman/listinfo/ippm From ippm-admin@advanced.org Wed Sep 18 07:42:45 2002 Received: from mailhost.advanced.org (root@mailhost.advanced.org [209.211.239.227]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id HAA08791 for ; Wed, 18 Sep 2002 07:42:45 -0400 (EDT) Received: from mailhost.advanced.org (localhost [127.0.0.1]) by mailhost.advanced.org (8.12.4/8.12.4/Debian-4) with ESMTP id g8IBe4iU020213; Wed, 18 Sep 2002 07:40:04 -0400 Received: from smtp.noos.fr (racine.noos.net [212.198.2.71]) by mailhost.advanced.org (8.12.4/8.12.4/Debian-4) with ESMTP id g8IBd1iV020079 (version=TLSv1/SSLv3 cipher=EDH-RSA-DES-CBC3-SHA bits=168 verify=FAIL) for ; Wed, 18 Sep 2002 07:39:02 -0400 Received: (qmail 30261992 invoked by uid 0); 18 Sep 2002 11:38:58 -0000 Received: from unknown (HELO LIP60MU1PINK59) ([212.198.208.249]) (envelope-sender ) by 212.198.2.71 (qmail-ldap-1.03) with SMTP for ; 18 Sep 2002 11:38:58 -0000 From: =?us-ascii?Q?Kave_Salamatian?= To: "Alan Clark" , "H.E.Holland" , "stanislav shalunov" , Cc: "IPPM Working Group" Subject: RE: [ippm] concatenation of QoS parameters Message-ID: MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit X-Priority: 3 (Normal) X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook IMO, Build 9.0.2416 (9.0.2911.0) X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000 In-Reply-To: Importance: Normal Sender: ippm-admin@advanced.org Errors-To: ippm-admin@advanced.org X-BeenThere: ippm@advanced.org X-Mailman-Version: 2.0.12 Precedence: bulk List-Help: List-Post: List-Subscribe: , List-Id: IETF IP Performance Metrics Working Group List List-Unsubscribe: , List-Archive: Date: Wed, 18 Sep 2002 13:38:32 +0200 Content-Transfer-Encoding: 7bit Sorry for the incomplete url. The complete one is www-rp.lip6.fr/~salamat. For the question about how combining QoS parameter from two network segments it is indeed difficult. If you can suppose that the network segment are independent you may combine the Qos measure. But if they are dependent you may not combine them easily as the moement of the combinaison of two dependent Random variable is not straightforwardly a simple combinaison of the individual moments. Meaning that in a general scenario (without independence hypothesis) you may not be able to easily combine QoS parameter from two networks. Bests kv -----Message d'origine----- De : Alan Clark [mailto:alan@telchemy.com] Envoye : mercredi 18 septembre 2002 12:59 A : Kave Salamatian; H.E.Holland; stanislav shalunov; victor.reijs@surfnet.nl Cc : IPPM Working Group Objet : RE: [ippm] concatenation of QoS parameters Kave I had some difficulty finding the paper you reference on the lip6 web site - please verify it's location. My IPtel 2001 paper describes the use of a 4 state Markov Model for describing burst packet loss distribution and may be of interest to you http://www.iptel.org/2001/pg/final_program/21.pdf Whilst I have to agree with your philosophical observation concerning stationarity I think you are agreeing that traces do show transient packet loss and short term delay changes. The main subject of the posting was to address the issue of how you would combine QoS parameters from two network segments - it would be intersting to hear your views on this Regards Alan -----Original Message----- From: Kave Salamatian [mailto:salamat@rp.lip6.fr] Sent: Wednesday, September 18, 2002 1:41 AM To: Alan Clark; H.E.Holland; stanislav shalunov; victor.reijs@surfnet.nl Cc: IPPM Working Group Subject: RE: [ippm] concatenation of QoS parameters I agree with Alan about the structure of losses. Losses can be described by period of high loss probability followed by period of low loss probability. This is explained in "Hidden Markov Modeling for network communication channels, M.R Salamatian, S. Vaton, SIGMETRICS 2001, Cambridge, Massachusett." (can be downloaded from www-rp.lip6.fr). For delay we have a similar description. That is under writing. However I don't agree with Alan about Non stationnarity at all scales. Globally speaking, stationarity cannot be proven of disproven when you work with empirical data. Stationarity is only a property of the model that you decide to use and that is best suited to your application. In fact stationarity is a bet about the fact that what you have observed into your observation window will be reproduced (with same statistical parameters) out of the observation window. A single observation may be described as well by a stationnary model as a non stationnary. If one go for a stationnary model he may have to manage Long Range dependences that will be necessary to explain mean level variation. If he go for non stationarity he need to track the change of the underlying process, but he may use short memory model with memory scale at the level of his tracking agility. Which one is better depend on the application ! Meaning that loss and jitter are not non-stationnary by essence, but they may be subjectively stationnary or not. Kave Salamatian -----Message d'origine----- De : ippm-admin@advanced.org [mailto:ippm-admin@advanced.org]De la part de Alan Clark Envoye : mercredi 18 septembre 2002 03:21 A : H.E.Holland; stanislav shalunov; victor.reijs@surfnet.nl Cc : IPPM Working Group Objet : RE: [ippm] concatenation of QoS parameters My apologies for restating the obvious, if already stated .. typical traces of loss and jitter are clearly non-stationary. Jitter often appears to have a roughly constant "background" level with superimposed short term delay variations (100mS to several seconds) due to transient congestion on LAN and access links. Network packet loss is due to transient congestion, route changes, link failures and similar events. Contrary to popular belief, bursts are often long and sparse - e.g. several seconds of 20% loss - rather than the short bursts of consecutive loss often described. For multimedia applications this is a critical distinction as long bursts of 20-30% loss can be more objectionable than short dense loss periods. If you assume that both loss and jitter have some fairly constant background rate/ level with transient events then it would seem to be necessary to separately combine the background and transient parameters. e.g. for packet loss: (i) compute gap/background loss probability using independence assumption (ii) compute burst probability, assuming that the probability of bursts occurring on network segments is independent, and combine burst density/length stats. Alan Clark -----Original Message----- From: ippm-admin@advanced.org [mailto:ippm-admin@advanced.org]On Behalf Of H.E.Holland Sent: Tuesday, September 17, 2002 5:40 PM To: stanislav shalunov; victor.reijs@surfnet.nl Cc: IPPM Working Group Subject: Re: [ippm] concatenation of QoS parameters Dear all, I found this discussion about concatenating parameters from a while back. Little is said however about concatenating distributions. For example, I asume that I can simply add medians of delay samples for adjacent independently measured links. And the percentiles. Is this correct? The mail below speaks of one-way delay "averages" being added over various domains where average signifies "mean", but mean "can't" be calculated (officially anyway) since losses aren't acounted for. Thanks, Henrik. At 12:03 14-2-02 -0500, stanislav shalunov wrote: >This is probably outside of scope of IPPM WG activities (and in the >realm of a statistics), but: Everything hinges on the assumption of >independent distribution of the delay variables for different domains. >On one hand, since the next domain doesn't explicitly know about >treatment in the previous one you could say that the independence >assumption should be true. On the other hand, if there exist global >congestion oscillation phenomena (and why wouldn't they?) the >assumption would be incorrect. > >Further, you do not specify your metrics. Averages, for example, >would be much easier to analyze in this way than medians. > >Victor Reijs writes: > >> . OWD. I assume, I can just add figures in case of concatenation of >> domains. > >Sum is correct for averages regardless of independence. > >> . IPDV. This is a difficult one. I would tend to say one can not just >> added the figures. I once heard a convolution is needed here. What are >> your ideas? > >Sum is correct for variances provided that there is independence. > >> . IP Packet Loss. > >Loss probability, assuming independence, is, of course, 1-\Pi(1-p_i). >Accidentally, for small values of p_i the first order approximation of >this quantity happens to be \Sigma p_i. > >> . IP Loss patterns. > >No numeric parameter. > >> . IP packet reordering: > >No defined metric and no defined model. > >> The above was for the millisec range, perhaps other ideas are needed >> in microsec range. > >Scale doesn't matter. > >-- >Stanislav Shalunov http://www.internet2.edu/~shalunov/ > >"You wake me up early in the morning to tell me I am right? Please >wait until I am wrong." -- John von Neumann, on being phoned at 10 a.m. >_______________________________________________ >ippm mailing list >ippm@advanced.org >http://mailhost.advanced.org/mailman/listinfo/ippm > _______________________________________________ ippm mailing list ippm@advanced.org http://mailhost.advanced.org/mailman/listinfo/ippm _______________________________________________ ippm mailing list ippm@advanced.org http://mailhost.advanced.org/mailman/listinfo/ippm _______________________________________________ ippm mailing list ippm@advanced.org http://mailhost.advanced.org/mailman/listinfo/ippm From ippm-admin@advanced.org Wed Sep 18 08:09:40 2002 Received: from mailhost.advanced.org (root@mailhost.advanced.org [209.211.239.227]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id IAA09540 for ; Wed, 18 Sep 2002 08:09:40 -0400 (EDT) Received: from mailhost.advanced.org (localhost [127.0.0.1]) by mailhost.advanced.org (8.12.4/8.12.4/Debian-4) with ESMTP id g8IC83iU025157; Wed, 18 Sep 2002 08:08:03 -0400 Received: from emerson.torrentnet.com (emerson.torrentnet.com [198.78.51.110]) by mailhost.advanced.org (8.12.4/8.12.4/Debian-4) with ESMTP id g8IC78iU024746 for ; Wed, 18 Sep 2002 08:07:09 -0400 Received: from imperial.torrentnet.com (imperial.torrentnet.com [198.78.51.109]) by emerson.torrentnet.com (8.11.2/8.11.2) with ESMTP id g8IC5lP10365; Wed, 18 Sep 2002 08:05:47 -0400 (EDT) Received: from malibu.torrentnet.com (malibu.torrentnet.com [198.78.51.100]) by imperial.torrentnet.com (8.11.2/8.11.2) with ESMTP id g8IC5lQ41942; Wed, 18 Sep 2002 08:05:47 -0400 (EDT) Received: from malibu.torrentnet.com (chimento@localhost) by malibu.torrentnet.com (8.11.2/8.11.2) with ESMTP id g8IC5kG10300; Wed, 18 Sep 2002 08:05:46 -0400 (EDT) Message-Id: <200209181205.g8IC5kG10300@malibu.torrentnet.com> X-Mailer: exmh version 2.3.1 01/18/2001 with nmh-1.0.4 To: =?us-ascii?Q?Kave_Salamatian?= cc: "Alan Clark" , "H.E.Holland" , "stanislav shalunov" , victor.reijs@surfnet.nl, "IPPM Working Group" Subject: Re: [ippm] concatenation of QoS parameters In-Reply-To: Message from =?us-ascii?Q?Kave_Salamatian?= of "Wed, 18 Sep 2002 13:38:32 +0200." Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii From: Philip Chimento Sender: ippm-admin@advanced.org Errors-To: ippm-admin@advanced.org X-BeenThere: ippm@advanced.org X-Mailman-Version: 2.0.12 Precedence: bulk List-Help: List-Post: List-Subscribe: , List-Id: IETF IP Performance Metrics Working Group List List-Unsubscribe: , List-Archive: Date: Wed, 18 Sep 2002 08:05:46 -0400 Once again: means may always be combined (added or in linear combinations) whether or not the underlying random variables are independent. As Stanislav pointed out, variances must be of independent random variables to be able simply to add them. While Stanislav's statement is correct on this point, it does not apply to either IPDV or delay in successive segments in a real network. While you can build a model based on this assumption, your model will at best only approximate the real situation (and you must then work on bounds). I would NOT recommend trying to write an SLA based on higher moments or distributions of combinations of delay or IPDV. Regards, Phil Chimento -- Phil Chimento Ericsson IPI Phone: 240-314-3597 7301 Calhoun Place chimento@torrentnet.com Rockville, Md. 20855 ----------------------------------------------------- _______________________________________________ ippm mailing list ippm@advanced.org http://mailhost.advanced.org/mailman/listinfo/ippm From ippm-admin@advanced.org Wed Sep 18 09:12:02 2002 Received: from mailhost.advanced.org (root@mailhost.advanced.org [209.211.239.227]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id JAA11551 for ; Wed, 18 Sep 2002 09:12:02 -0400 (EDT) Received: from mailhost.advanced.org (localhost [127.0.0.1]) by mailhost.advanced.org (8.12.4/8.12.4/Debian-4) with ESMTP id g8IDC3iU001928; Wed, 18 Sep 2002 09:12:03 -0400 Received: from emerson.torrentnet.com (emerson.torrentnet.com [198.78.51.110]) by mailhost.advanced.org (8.12.4/8.12.4/Debian-4) with ESMTP id g8IDBJiU001829 for ; Wed, 18 Sep 2002 09:11:20 -0400 Received: from imperial.torrentnet.com (imperial.torrentnet.com [198.78.51.109]) by emerson.torrentnet.com (8.11.2/8.11.2) with ESMTP id g8IDBH014762; Wed, 18 Sep 2002 09:11:17 -0400 (EDT) Received: from malibu.torrentnet.com (malibu.torrentnet.com [198.78.51.100]) by imperial.torrentnet.com (8.11.2/8.11.2) with ESMTP id g8IDBHw46711; Wed, 18 Sep 2002 09:11:17 -0400 (EDT) Received: from malibu.torrentnet.com (chimento@localhost) by malibu.torrentnet.com (8.11.2/8.11.2) with ESMTP id g8IDBGH11436; Wed, 18 Sep 2002 09:11:16 -0400 (EDT) Message-Id: <200209181311.g8IDBGH11436@malibu.torrentnet.com> X-Mailer: exmh version 2.3.1 01/18/2001 with nmh-1.0.4 To: salamat@rp.lip6.fr, ippm@advanced.org Subject: Re: [ippm] concatenation of QoS parameters In-Reply-To: Message from =?us-ascii?Q?Kave_Salamatian?= of "Wed, 18 Sep 2002 14:33:12 +0200." Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii From: Philip Chimento Sender: ippm-admin@advanced.org Errors-To: ippm-admin@advanced.org X-BeenThere: ippm@advanced.org X-Mailman-Version: 2.0.12 Precedence: bulk List-Help: List-Post: List-Subscribe: , List-Id: IETF IP Performance Metrics Working Group List List-Unsubscribe: , List-Archive: Date: Wed, 18 Sep 2002 09:11:16 -0400 Hi Kave: > I agree with Philipp about the first moment additivity property. But once > again a mean without any indication of dispersion (at least variance) is > meaningless ! a 10% loss rate mean can be one packet each 10 packets or > 10000 consecutive packets each 100000 packets. Even if the delay in two > consecutive segment is the sum of the two delay but the jitter (which is the > variance of the delay) is not the sum of the two jitter. > The problem of giving means without variances is well known. About 15 years ago Ward Whitt wrote an article (I think that it was published in Management Science) titled "The horrors of the waiting room..." that should be required reading. I certainly was not suggesting that you ignore variability. > I don't mean that we have to integrate higher order moment and or > distribution directly into SLA. Nevertheless SLA should define something > that is statistically correct. > I think the last sentance is a statement everyone can agree with :) > I think that the good way of defining an SLA will be to define it as : > 1- a mean "Easy". > 2- an upper and lower bound > 3- A probability of crossing the bound This is where you run into trouble. How are you going to determine these ? How can you guarantee that the bounds will not be so wide as to be useless ? I can give you bounds on delay right now: 0 picoseconds and 5 years. Probability to cross the lower bound is 0. Probability to cross the upper bound is less than 10^(-100). My guess is that to come up with tight bounds is on the same order of difficulty as doing the convolutions of the pdfs. After you come up with the bounds, they may be easier to compute, but... > 4- a scale of time for calculating the mean and the previous probability > > All the difficulties with other moments and pdf and so will be hidden into > the relationship between the bounds and the probability of crossing. > > Regards > > Kv > > Regards, Phil -- Phil Chimento Ericsson IPI Phone: 240-314-3597 7301 Calhoun Place chimento@torrentnet.com Rockville, Md. 20855 ----------------------------------------------------- > -----Message d'origine----- > De : Philip Chimento [mailto:chimento@torrentnet.com] > Envoye : mercredi 18 septembre 2002 14:06 > A : Kave Salamatian > Cc : Alan Clark; H.E.Holland; stanislav shalunov; > victor.reijs@surfnet.nl; IPPM Working Group > Objet : Re: [ippm] concatenation of QoS parameters > > > > Once again: means may always be combined (added or in linear combinations) > whether or not the underlying random variables are independent. As Stanislav > pointed out, variances must be of independent random variables to be able > simply to add them. While Stanislav's statement is correct on this point, it > does not apply to either IPDV or delay in successive segments in a real > network. While you can build a model based on this assumption, your model > will > at best only approximate the real situation (and you must then work on > bounds). I would NOT recommend trying to write an SLA based on higher > moments > or distributions of combinations of delay or IPDV. > > Regards, Phil Chimento > -- > Phil Chimento Ericsson IPI > Phone: 240-314-3597 7301 Calhoun Place > chimento@torrentnet.com Rockville, Md. 20855 > ----------------------------------------------------- > _______________________________________________ ippm mailing list ippm@advanced.org http://mailhost.advanced.org/mailman/listinfo/ippm From ippm-admin@advanced.org Sat Sep 21 09:37:09 2002 Received: from mailhost.advanced.org (root@mailhost.advanced.org [209.211.239.227]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id JAA11131 for ; Sat, 21 Sep 2002 09:37:09 -0400 (EDT) Received: from mailhost.advanced.org (localhost [127.0.0.1]) by mailhost.advanced.org (8.12.4/8.12.4/Debian-4) with ESMTP id g8LDV3iU027856; Sat, 21 Sep 2002 09:31:03 -0400 Received: from mail.internet2.edu (mail.internet2.edu [209.211.239.218]) by mailhost.advanced.org (8.12.4/8.12.4/Debian-4) with ESMTP id g8LDUfiU027752 for ; Sat, 21 Sep 2002 09:30:41 -0400 Received: from localhost (localhost [127.0.0.1]) by mail.internet2.edu (Postfix) with ESMTP id 8EC365EE8E; Sat, 21 Sep 2002 09:30:41 -0400 (EDT) Received: from localhost (localhost [127.0.0.1]) by mail.internet2.edu (Postfix) with ESMTP id 7D69A5EE87; Sat, 21 Sep 2002 09:30:40 -0400 (EDT) To: "H.E.Holland" Cc: IPPM Working Group Subject: Re: [ippm] concatenation of QoS parameters References: <3C6B8CED.72D8475D@surfnet.nl> <3C6B8CED.72D8475D@surfnet.nl> <200209172142.g8HLgJQ19312@netlx010.civ.utwente.nl> From: stanislav shalunov In-Reply-To: <200209172142.g8HLgJQ19312@netlx010.civ.utwente.nl> Message-ID: <87znubii68.fsf@cain.internet2.edu> Lines: 14 X-Mailer: Gnus v5.7/Emacs 20.4 Sender: ippm-admin@advanced.org Errors-To: ippm-admin@advanced.org X-BeenThere: ippm@advanced.org X-Mailman-Version: 2.0.12 Precedence: bulk List-Help: List-Post: List-Subscribe: , List-Id: IETF IP Performance Metrics Working Group List List-Unsubscribe: , List-Archive: Date: 21 Sep 2002 09:30:39 -0400 "H.E.Holland" writes: > Little is said however about concatenating distributions. For example, I > asume that I can simply add medians of delay samples for adjacent > independently measured links. And the percentiles. You *cannot* add medians, etc. For that, you'll have to do a convolution (which requires both complete distributions, not just one or five percentiles). -- Stanislav Shalunov http://www.internet2.edu/~shalunov/ Sex is the mathematics urge sublimated. -- M. C. Reed. _______________________________________________ ippm mailing list ippm@advanced.org http://mailhost.advanced.org/mailman/listinfo/ippm From ippm-admin@advanced.org Sat Sep 21 17:41:43 2002 Received: from mailhost.advanced.org (root@mailhost.advanced.org [209.211.239.227]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id RAA17842 for ; Sat, 21 Sep 2002 17:41:43 -0400 (EDT) Received: from mailhost.advanced.org (localhost [127.0.0.1]) by mailhost.advanced.org (8.12.4/8.12.4/Debian-4) with ESMTP id g8LLf4iU020492; Sat, 21 Sep 2002 17:41:04 -0400 Received: from netlx009.civ.utwente.nl (netlx009.civ.utwente.nl [130.89.1.91]) by mailhost.advanced.org (8.12.4/8.12.4/Debian-4) with ESMTP id g8LLeviU020398 for ; Sat, 21 Sep 2002 17:40:58 -0400 Received: from henrik (inbel242216.inbel.utwente.nl [130.89.242.216]) by netlx009.civ.utwente.nl (8.11.4/HKD) with SMTP id g8LLeoO22302; Sat, 21 Sep 2002 23:40:50 +0200 Message-Id: <200209212140.g8LLeoO22302@netlx009.civ.utwente.nl> X-Sender: s9605363@mail.student.utwente.nl X-Mailer: QUALCOMM Windows Eudora Pro Version 4.0.1 To: Matthew Roughan From: "H.E.Holland" Subject: Re: [ippm] concatenation of QoS parameters Cc: ippm@advanced.org In-Reply-To: <3D888692.7050607@research.att.com> References: <3C6B8CED.72D8475D@surfnet.nl> <3C6B8CED.72D8475D@surfnet.nl> <200209172142.g8HLgJQ19312@netlx010.civ.utwente.nl> <200209181222.g8ICMj430863@netlx009.civ.utwente.nl> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" X-UTwente-MailScanner: Found to be clean Sender: ippm-admin@advanced.org Errors-To: ippm-admin@advanced.org X-BeenThere: ippm@advanced.org X-Mailman-Version: 2.0.12 Precedence: bulk List-Help: List-Post: List-Subscribe: , List-Id: IETF IP Performance Metrics Working Group List List-Unsubscribe: , List-Archive: Date: Sat, 21 Sep 2002 23:38:46 +0200 Hi Matthew, So I guess one of the hardest parts is acquiring of "fitting" a complete distrubution function. How would you do this? Furthermore, maybe a Laplace transformation to make calculations easier? Henrik. At 09:58 18-9-02 -0400, you wrote: >Hi, > >its actually quite tricky. The simplest case is where you know the >distribution functions (or can fit simple distribution functions) and >then you integrate to do the convolution. > >Mathematically, if you are taking the sum X = X_1 + X_2 (where X_1 and >X_2 are non-negative), the distribution function of X is > F(T) = p(X<=T) > = int_0^T P(X_1 <= S) P(X_2 <= T-S) dS > = int_0^T F_1(S) F_2(T-S) dS >where F_1 and F_2 are the distribution functions of X_1 and X_2. > >There are neat ways to compute these for some distributions by taking >the characteristic functions of the distributions. There are also >special cases like the Gaussian distribution where the distribution is >completely characterized by the mean and variance so adding independent >RVs is easy. > >I assume you can do the integral above numerically for empirical >distributions, but I don't have any experience doing that, and don't >know what the numerical issues are, but you had better have fairly good >data before you try something like that. > >Matt > >H.E.Holland wrote: >> Matt, >> >> Can you tell me from a practical point of view how I do this? With the >> normal discrete time convolution sum? >> >> Henrik. >> >> At 21:28 17-9-02 -0400, you wrote: >> >>>No, you can't add medians or other percentiles. You have to convolve the >>>distributions. >>> >>>Matt >>> >>>H.E.Holland wrote: >>> >>> >>>>Dear all, >>>> >>>>I found this discussion about concatenating parameters from a while back. >>>>Little is said however about concatenating distributions. For example, I >>>>asume that I can simply add medians of delay samples for adjacent >>>>independently measured links. And the percentiles. Is this correct? The >>>>mail below speaks of one-way delay "averages" being added over various >>>>domains where average signifies "mean", but mean "can't" be calculated >>>>(officially anyway) since losses aren't acounted for. >>>> >>>>Thanks, >>>>Henrik. >>>> >>>> >>>>At 12:03 14-2-02 -0500, stanislav shalunov wrote: >>>> >>>> >>>> >>>>>This is probably outside of scope of IPPM WG activities (and in the >>>>>realm of a statistics), but: Everything hinges on the assumption of >>>>>independent distribution of the delay variables for different domains. >>>>>On one hand, since the next domain doesn't explicitly know about >>>>>treatment in the previous one you could say that the independence >>>>>assumption should be true. On the other hand, if there exist global >>>>>congestion oscillation phenomena (and why wouldn't they?) the >>>>>assumption would be incorrect. >>>>> >>>>>Further, you do not specify your metrics. Averages, for example, >>>>>would be much easier to analyze in this way than medians. >>>>> >>>>>Victor Reijs writes: >>>>> >>>>> >>>>> >>>>> >>>>>>. OWD. I assume, I can just add figures in case of concatenation of >>>>>>domains. >>>>>> >>>>>> >>>>> >>>>>Sum is correct for averages regardless of independence. >>>>> >>>>> >>>>> >>>>> >>>>>>. IPDV. This is a difficult one. I would tend to say one can not just >>>>>>added the figures. I once heard a convolution is needed here. What are >>>>>>your ideas? >>>>>> >>>>>> >>>>> >>>>>Sum is correct for variances provided that there is independence. >>>>> >>>>> >>>>> >>>>> >>>>>>. IP Packet Loss. >>>>>> >>>>>> >>>>> >>>>>Loss probability, assuming independence, is, of course, 1-\Pi(1-p_i). >>>>>Accidentally, for small values of p_i the first order approximation of >>>>>this quantity happens to be \Sigma p_i. >>>>> >>>>> >>>>> >>>>> >>>>>>. IP Loss patterns. >>>>>> >>>>>> >>>>> >>>>>No numeric parameter. >>>>> >>>>> >>>>> >>>>> >>>>>>. IP packet reordering: >>>>>> >>>>>> >>>>> >>>>>No defined metric and no defined model. >>>>> >>>>> >>>>> >>>>> >>>>>>The above was for the millisec range, perhaps other ideas are needed >>>>>>in microsec range. >>>>>> >>>>>> >>>>> >>>>>Scale doesn't matter. >>>>> >>>>>-- >>>>>Stanislav Shalunov http://www.internet2.edu/~shalunov/ >>>>> >>>>>"You wake me up early in the morning to tell me I am right? Please >>>>>wait until I am wrong." -- John von Neumann, on being phoned at 10 a.m. >>>>>_______________________________________________ >>>>>ippm mailing list >>>>>ippm@advanced.org >>>>>http://mailhost.advanced.org/mailman/listinfo/ippm >>>>> >>>>> >>>>> >>>> >>>> >>>>_______________________________________________ >>>>ippm mailing list >>>>ippm@advanced.org >>>>http://mailhost.advanced.org/mailman/listinfo/ippm >>>> >>>> >>> >> >> > _______________________________________________ ippm mailing list ippm@advanced.org http://mailhost.advanced.org/mailman/listinfo/ippm From ippm-admin@advanced.org Tue Sep 24 09:07:27 2002 Received: from mailhost.advanced.org (root@mailhost.advanced.org [209.211.239.227]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id JAA25816 for ; Tue, 24 Sep 2002 09:07:27 -0400 (EDT) Received: from mailhost.advanced.org (localhost [127.0.0.1]) by mailhost.advanced.org (8.12.4/8.12.4/Debian-4) with ESMTP id g8OCw3iU030282; Tue, 24 Sep 2002 08:58:03 -0400 Received: from emerson.torrentnet.com (emerson.torrentnet.com [198.78.51.110]) by mailhost.advanced.org (8.12.4/8.12.4/Debian-4) with ESMTP id g8OCvNiU029857 for ; Tue, 24 Sep 2002 08:57:24 -0400 Received: from imperial.torrentnet.com (imperial.torrentnet.com [198.78.51.109]) by emerson.torrentnet.com (8.11.2/8.11.2) with ESMTP id g8OCu6h89609; Tue, 24 Sep 2002 08:56:06 -0400 (EDT) Received: from malibu.torrentnet.com (malibu.torrentnet.com [198.78.51.100]) by imperial.torrentnet.com (8.11.2/8.11.2) with ESMTP id g8OCu6E52694; Tue, 24 Sep 2002 08:56:06 -0400 (EDT) Received: from malibu.torrentnet.com (chimento@localhost) by malibu.torrentnet.com (8.11.2/8.11.2) with ESMTP id g8OCu5P12622; Tue, 24 Sep 2002 08:56:05 -0400 (EDT) Message-Id: <200209241256.g8OCu5P12622@malibu.torrentnet.com> X-Mailer: exmh version 2.3.1 01/18/2001 with nmh-1.0.4 To: "H.E.Holland" cc: Matthew Roughan , ippm@advanced.org Subject: Re: [ippm] concatenation of QoS parameters In-Reply-To: Message from "H.E.Holland" of "Sat, 21 Sep 2002 23:38:46 +0200." <200209212140.g8LLeoO22302@netlx009.civ.utwente.nl> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii From: Philip Chimento Sender: ippm-admin@advanced.org Errors-To: ippm-admin@advanced.org X-BeenThere: ippm@advanced.org X-Mailman-Version: 2.0.12 Precedence: bulk List-Help: List-Post: List-Subscribe: , List-Id: IETF IP Performance Metrics Working Group List List-Unsubscribe: , List-Archive: Date: Tue, 24 Sep 2002 08:56:04 -0400 Henrik: Fitting distributions is equally tricky. There are many different statistical tests to use for fitting distributions, but in general it is an art of approximation. Further if you have purely empirical distributions, Laplace transforms won't help. If you fit the data to a known distribution (approximately) and convolve the known distribution then you will get an approximate answer. But, remember, if the distributions are not independent (and they are NOT in this case) you have a more complicated convolution integral than the one that Matt gave you. In that case, you wind up with an integral that looks like (for Z = X+Y, the sum of only two random variables) f(z) = int_0^infinity{f(s, z-s) ds} to get the density (see, Trivedi, Probability & Statistics with Reliability, Queueing and Computer Science Applications for the derivation). Note that the bivariate function (f(x,y)) does not resolve into a product of the marginals because they are not independent. Basically, if you don't fit a distribution function, you have to compute this numerically. Regards, Phil Chimento -- Phil Chimento Ericsson IPI Phone: 240-314-3597 7301 Calhoun Place chimento@torrentnet.com Rockville, Md. 20855 ----------------------------------------------------- _______________________________________________ ippm mailing list ippm@advanced.org http://mailhost.advanced.org/mailman/listinfo/ippm From ippm-admin@advanced.org Tue Sep 24 10:47:25 2002 Received: from mailhost.advanced.org (root@mailhost.advanced.org [209.211.239.227]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id KAA00274 for ; Tue, 24 Sep 2002 10:47:25 -0400 (EDT) Received: from mailhost.advanced.org (localhost [127.0.0.1]) by mailhost.advanced.org (8.12.4/8.12.4/Debian-4) with ESMTP id g8OEl3iU004402; Tue, 24 Sep 2002 10:47:03 -0400 Received: from mail-blue.research.att.com (H-135-207-30-102.research.att.com [135.207.30.102]) by mailhost.advanced.org (8.12.4/8.12.4/Debian-4) with ESMTP id g8ODT7iU007535 for ; Tue, 24 Sep 2002 09:29:08 -0400 Received: from bigmail.research.att.com (bigmail.research.att.com [135.207.30.101]) by mail-blue.research.att.com (Postfix) with ESMTP id 432664CE4B; Tue, 24 Sep 2002 09:29:07 -0400 (EDT) Received: from research.att.com (roughan@wombat.research.att.com [135.207.26.100]) by bigmail.research.att.com (8.8.8/8.8.8) with ESMTP id JAA02108; Tue, 24 Sep 2002 09:29:06 -0400 (EDT) Message-ID: <3D9068BF.2040304@research.att.com> From: Matthew Roughan Organization: AT&T Research Labs User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.1) Gecko/20020826 X-Accept-Language: en-us, en MIME-Version: 1.0 To: Philip Chimento Cc: "H.E.Holland" , ippm@advanced.org Subject: Re: [ippm] concatenation of QoS parameters References: <200209241256.g8OCu5P12622@malibu.torrentnet.com> Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit Sender: ippm-admin@advanced.org Errors-To: ippm-admin@advanced.org X-BeenThere: ippm@advanced.org X-Mailman-Version: 2.0.12 Precedence: bulk List-Help: List-Post: List-Subscribe: , List-Id: IETF IP Performance Metrics Working Group List List-Unsubscribe: , List-Archive: Date: Tue, 24 Sep 2002 09:29:35 -0400 Content-Transfer-Encoding: 7bit All good points. Particularly as delay distributions can easily be multimodal, and fitting such distributions can be challenging. Matt Philip Chimento wrote: > Henrik: > Fitting distributions is equally tricky. There are many different > statistical tests to use for fitting distributions, but in general > it is an art of approximation. Further if you have purely empirical > distributions, Laplace transforms won't help. If you fit the data to > a known distribution (approximately) and convolve the known distribution > then you will get an approximate answer. But, remember, if the distributions > are not independent (and they are NOT in this case) you have a more > complicated convolution integral than the one that Matt gave you. In that > case, you wind up with an integral that looks like (for Z = X+Y, the sum of > only two random variables) > f(z) = int_0^infinity{f(s, z-s) ds} to get the density (see, Trivedi, > Probability & Statistics with Reliability, Queueing and Computer Science > Applications for the derivation). Note that the bivariate function (f(x,y)) > does not resolve into a product of the marginals because they are not > independent. Basically, if you don't fit a distribution function, you have to > compute this numerically. > Regards, Phil Chimento _______________________________________________ ippm mailing list ippm@advanced.org http://mailhost.advanced.org/mailman/listinfo/ippm From ippm-admin@advanced.org Sat Sep 28 22:31:52 2002 Received: from mailhost.advanced.org (root@mailhost.advanced.org [209.211.239.227]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id WAA16345 for ; Sat, 28 Sep 2002 22:31:52 -0400 (EDT) Received: from mailhost.advanced.org (localhost [127.0.0.1]) by mailhost.advanced.org (8.12.4/8.12.4/Debian-4) with ESMTP id g8T2T3iU021634; Sat, 28 Sep 2002 22:29:03 -0400 Received: from web15101.mail.bjs.yahoo.com (web15101.mail.bjs.yahoo.com [61.135.128.13]) by mailhost.advanced.org (8.12.4/8.12.4/Debian-4) with SMTP id g8T2S2iU021528 for ; Sat, 28 Sep 2002 22:28:03 -0400 Message-ID: <20020929022801.52939.qmail@web15101.mail.bjs.yahoo.com> Received: from [61.174.139.165] by web15101.mail.bjs.yahoo.com via HTTP; Sun, 29 Sep 2002 10:28:01 CST From: =?gb2312?q?Shen=20Jing?= To: end2end-interest@postel.org Cc: ippm@advanced.org MIME-Version: 1.0 Content-Type: text/plain; charset=gb2312 Content-Transfer-Encoding: 8bit Subject: [ippm] Mathematical analysis of e2e lable switching path Sender: ippm-admin@advanced.org Errors-To: ippm-admin@advanced.org X-BeenThere: ippm@advanced.org X-Mailman-Version: 2.0.12 Precedence: bulk List-Help: List-Post: List-Subscribe: , List-Id: IETF IP Performance Metrics Working Group List List-Unsubscribe: , List-Archive: Date: Sun, 29 Sep 2002 10:28:01 +0800 (CST) Content-Transfer-Encoding: 8bit Hi there, I've some question on e2e performance analysis of label switching path. There has been many research on e2e performance analysis for IP routing. Nearly all of them take the model of "network of queue", and under assumpiton of Possion Arrival, exponential flow size etc. To my limited knowledge, this is not to the state of Internet which experiences LRD. On the other hand, different router archtecture must have different effect on transmission performance. As MPLS proceeds to be the primary backbone technology, I think there must be something new introduced into the e2e performance in internet. So, I want to know, whether there is some work with methematical modeling of LSP networks? and, are there anyone would do me a favor to recommend some refrence book, web page, research paper or the like ? Thank you very much. Best regards ===== Jing Shen State Key Lab of CAD&CG ZheJiang University(YuQuan) HangZhou, ZheJiang Province 310027 P.R.China _________________________________________________________ Do You Yahoo!? "发短信赢手机,快来参加雅虎巨星秀!" http://cn.ent.yahoo.com/star/midautumn/index.html _______________________________________________ ippm mailing list ippm@advanced.org http://mailhost.advanced.org/mailman/listinfo/ippm From ippm-admin@advanced.org Mon Sep 30 12:10:36 2002 Received: from mailhost.advanced.org (root@mailhost.advanced.org [209.211.239.227]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id MAA10087 for ; Mon, 30 Sep 2002 12:10:35 -0400 (EDT) Received: from mailhost.advanced.org (localhost [127.0.0.1]) by mailhost.advanced.org (8.12.4/8.12.4/Debian-4) with ESMTP id g8UGB4iU016666; Mon, 30 Sep 2002 12:11:04 -0400 Received: from mail.internet2.edu (mail.internet2.edu [209.211.239.218]) by mailhost.advanced.org (8.12.4/8.12.4/Debian-4) with ESMTP id g8UGA6iU016388 for ; Mon, 30 Sep 2002 12:10:06 -0400 Received: from localhost (localhost [127.0.0.1]) by mail.internet2.edu (Postfix) with ESMTP id 152125EEA1; Mon, 30 Sep 2002 12:10:06 -0400 (EDT) Received: from localhost (localhost [127.0.0.1]) by mail.internet2.edu (Postfix) with ESMTP id 1C4DB5EE8E; Mon, 30 Sep 2002 12:10:05 -0400 (EDT) To: h.e.holland@student.utwente.nl Cc: ippm@advanced.org Subject: Re: [ippm] metric requirements References: <200209071215.g87CFLf30328@netlx009.civ.utwente.nl> From: stanislav shalunov In-Reply-To: <200209071215.g87CFLf30328@netlx009.civ.utwente.nl> Message-ID: <87it0no3vn.fsf@cain.internet2.edu> Lines: 19 X-Mailer: Gnus v5.7/Emacs 20.4 Sender: ippm-admin@advanced.org Errors-To: ippm-admin@advanced.org X-BeenThere: ippm@advanced.org X-Mailman-Version: 2.0.12 Precedence: bulk List-Help: List-Post: List-Subscribe: , List-Id: IETF IP Performance Metrics Working Group List List-Unsubscribe: , List-Archive: Date: 30 Sep 2002 12:10:04 -0400 "H.E.Holland" writes: > Where might I find quantitative metric requirements for services > such as VoIP? E.g. what are recommended values for one-way delay and > packet loss for a VoIP service? While others pointed out the ITU documents for one-way delay, I don't think we answered the `loss' part of the question. There can possibly be no answer, because codecs have various degrees of tolerance and redundancy. Relevant gedanken Experiment: Imagine that your favorite codec were to send all its packets in triplicate; that would change the threshold of tolerance for loss, wouldn't it? -- Stanislav Shalunov http://www.internet2.edu/~shalunov/ "I didn't attend the funeral, but I sent a nice letter saying that I approved of it." -- Mark Twain _______________________________________________ ippm mailing list ippm@advanced.org http://mailhost.advanced.org/mailman/listinfo/ippm From ippm-admin@advanced.org Mon Sep 30 15:20:29 2002 Received: from mailhost.advanced.org (root@mailhost.advanced.org [209.211.239.227]) by ietf.org (8.9.1a/8.9.1a) with ESMTP id PAA18071 for ; Mon, 30 Sep 2002 15:20:29 -0400 (EDT) Received: from mailhost.advanced.org (localhost [127.0.0.1]) by mailhost.advanced.org (8.12.4/8.12.4/Debian-4) with ESMTP id g8UJL3iU028936; Mon, 30 Sep 2002 15:21:03 -0400 Received: from telchemy.com ([66.155.69.251]) by mailhost.advanced.org (8.12.4/8.12.4/Debian-4) with ESMTP id g8UJKuiU028841 for ; Mon, 30 Sep 2002 15:20:57 -0400 Received: from TELWS104 [209.186.148.73] by telchemy.com with ESMTP (SMTPD32-7.11) id A4478AD0180; Mon, 30 Sep 2002 15:21:43 -0400 Reply-To: From: "Alan Clark" To: "'stanislav shalunov'" , Cc: Subject: RE: [ippm] metric requirements Message-ID: <002f01c268b6$2ed91370$6801a8c0@telchemy.com> MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit X-Priority: 3 (Normal) X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook CWS, Build 9.0.2416 (9.0.2911.0) Importance: Normal X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000 In-Reply-To: <87it0no3vn.fsf@cain.internet2.edu> Sender: ippm-admin@advanced.org Errors-To: ippm-admin@advanced.org X-BeenThere: ippm@advanced.org X-Mailman-Version: 2.0.12 Precedence: bulk List-Help: List-Post: List-Subscribe: , List-Id: IETF IP Performance Metrics Working Group List List-Unsubscribe: , List-Archive: Date: Mon, 30 Sep 2002 15:18:38 -0400 Content-Transfer-Encoding: 7bit Fortunately many codecs behave in a consistent way and don't generate triple the amount of traffic in order to improve resilience!! Work done within ITU SG12, ETSI TIPHON and a few other groups has characterized the behavior of a wide range of codecs under packet loss conditions up to 20-30%. It can however be a little tricky deducing if packet loss concealment is in use from the info available in RTP headers! One key point when considering the effects of packet loss on VoIP is the distribution of packet loss. This does not behave, as many would suggest, as a simple burst channel exhibiting only consecutive loss. Bursts of loss are often sparse (e.g. 20%) and extend for periods of seconds - largely caused by periods of congestion. You may also see the extreme case of bursts of consecutive loss extending for hundreds of packets, due to link failures. The effect of this is that call quality will be fine for some period of time, and then experience several seconds of badly degraded quality before improving again. Alan Clark -----Original Message----- From: ippm-admin@advanced.org [mailto:ippm-admin@advanced.org]On Behalf Of stanislav shalunov Sent: Monday, September 30, 2002 12:10 PM To: h.e.holland@student.utwente.nl Cc: ippm@advanced.org Subject: Re: [ippm] metric requirements "H.E.Holland" writes: > Where might I find quantitative metric requirements for services > such as VoIP? E.g. what are recommended values for one-way delay and > packet loss for a VoIP service? While others pointed out the ITU documents for one-way delay, I don't think we answered the `loss' part of the question. There can possibly be no answer, because codecs have various degrees of tolerance and redundancy. Relevant gedanken Experiment: Imagine that your favorite codec were to send all its packets in triplicate; that would change the threshold of tolerance for loss, wouldn't it? -- Stanislav Shalunov http://www.internet2.edu/~shalunov/ "I didn't attend the funeral, but I sent a nice letter saying that I approved of it." -- Mark Twain _______________________________________________ ippm mailing list ippm@advanced.org http://mailhost.advanced.org/mailman/listinfo/ippm _______________________________________________ ippm mailing list ippm@advanced.org http://mailhost.advanced.org/mailman/listinfo/ippm