Agenda Bashing by - Chair Persons (Matthew Zekauskas & Henk Uijterwaal) Chair person then went over administrative requirements. Jay agreed to take minutes ---------------------------------------- TWAMP By Kaynam Heyadat TWAMP version 6 is needed and planned to be released. Al: 6) Section 3.5 - Does the server being contacted with an unknown command and always responding create a security risk? Chair person Matt said since it is on top of TCP it cannot be a big risk ---------------------------------------- Al's Framework for Metric Composition and spatial composition of Metrics. Presented his slides. Slide 2 - version of spatial-composition should have been 5. Explained types of composition. None objected Al's response for comments from Alan Clark. - Waiting for deadline hurts. No one reads the drafts. - In response to Rexa Fardid's comments Al agreed to some of them. Al: Since we added the term "index" in this version, it was a source of concern for both Reza and Loki (and others mentioned this off-line). Loki's concern regards the explicit scope, clearly this was an attempt to expand the scope of the draft and make it useful beyond IPPM. We need to carefully consider this course, and get feedback from the WG Al wanted to say "no" to referencing RFC editor submissions - typo there. More comments from Reza. Al agreed. Al would respond to more comments on the list. Al would address more of Loki's comments. Loki envisions a case where some sub-path metrics are not standardized, they might be proprietary, or may be covered by IPR. What is the implication on the composed metric, if proprietary and or IPR-encumbered metrics are used to compose a standard metric? Al further added that it is besides the point and worth further discussion since Loki deals with these issues in particular. Al wanted more comments on Framework and needs volunteers for reviewing spatial composition as well. Chairperson - Henk is pulling a list of volunteers. He asked that Al sends mail to the folks who volunteered seeking comments. Chairperson thanked folks who commented. --------------------------------------- Al presented status of the multimetrics draft. Al gave a good editorial read-through. Encouraged the audience to read. Joe Kopena From Drexel university - Wanted to know if the draft could be split into 2 since it is so huge. Most contents would not be overlapping. Al mentioned he did a similar approach with p2p and p2mp with some other work and agreed with this comment. Al agreed it is getting long - especially according to the Joe Kopena that when spatial jitter metric is added. Al said this point is well taken and asked if any other wanted to see the draft split. No one raised their hands. Al mentioned he will figure this out. Al wanted to have WGLC on version 06 by mid January. Henk commented that there was not much point in the split and Al added that someone with interest in one part of a draft can just read and comment on that part. Also has other comments he would send the list but 1 comment in section 4 - indicating a loss and ..... Statement is not clear. It is not obvious that it is possible. Al agreed. Need some clarification there. Al agreed If loss and delay was not part of the vector it would have been simpler. ---------------------------------------- Delay variation Applicability statement By Al Beat Benoit Claise on a nightly basis to evolve this draft. Al described 2 methods to create delay metrics. Described inter packet delay var. Section 3.2 has de-jitter buffer size - clarified that section further. Got some comment on Appendixes but need more. Requested audience to look at the recent version to figure out if their comments were addressed. Al summarized Loki'S comments. What is done in the draft is within the scope of this WG. Section 8.6 reference is important according to Loki and Al incorporated this amongst other comments. Al requested audience to point any ambiguity in the text. Al then acknowledged Carsten for his comments and discussed the comments. There was just 1 TBD from Bob Holley's comments. Al asked "Do we rely on delay variation for path change ? " Al asked if someone disagrees with the comparison table since that could be discussed. Al summarized that IPDV and PDV have their strengths and weaknesses. Suggestion for additional tasks and circumstances - if any should be sent soon. Chairperson - Henk suggested that the list of unresolved comments should be sent to the list. Al agreed. ---------------------------------------- Reporting IP Performance Metrics to Users Draft by Martin Swany Swany mentioned he took over as the editor of the ID. Talked about the renamed Metrics. Also went over other changes to the recent version. There are a couple of open issues and Swany requested comments on these. Swany asked if there was more clarifications than the naming issues. No question from audience ? Chairperson then mentioned there is a related draft by Al who was requested to present. ---------------------------------------- Reporting Metrics : Different point of view By Al Morton. In a humorous note - Al mentioned Steve Konish commented on Thanksgiving day and that he responded same day. Al: Average Delay Variation: a lot of people report this, but it doesn't relate to something useful for application design. Better to report the range or Pseudo range. Matt and Henk said they read this ID. Al indicated that this has been out for a while and solicited comments. No comments from audience. Al and chairpersons decided to ask the list. Steve Pratt, Loki and Jay Karthik volunteered to review this. ---------------------------------------- Duplicate Drafts - Henk Asked for comments and if this can be moved to WGLC. Al mentioned that it is in good shape for a near future - WGLC. ---------------------------------------- ITU-T SG 12 Performance Metrics Liaison - Al. Al mentioned that IPv6 is considered and raised a point that IPv6 needs to be looked at work done at IPPM. Henk clarified that none of the drafts are specific to IP versions. Matt suggested Al to let him know if something was explicit. Al gave an example of how TTL is now Hop Count. Consent is the term that ITU-T uses to indicate that the text is about to begin their Accelerated Approval process (AAP). If there are comments, then the text may still change. Al asked for comments on the G.chirp draft text from Q13. No question from audience. ---------------------------------------- Future of the WG. Henk mentioned that we are nearing the end of our task list and solicited comments/questions on this. Any opinions ? Al asked if there were any new proposal ? Henk mentioned not much. Lars Eggert mentioned 1 remaining item missing was definition of statistical equivalence in order to verify that multiple implementations implement the same metric correctly. This is needed to progress metrics along standards track. A implementation report is needed on this as well including details on whether the results generated for the same scenario are statistically equivalent. Al further mentioned - Test Plans - How are we making comparison - what is the threshold - How do deal with _____ - need to decide what is equivalent. Lars asked if there was a metric that the group was interested in a specific metric to be used as a test case for this process. (The process he mentioned earlier) Henk mentioned - interop was done for delay metrics. Henk took an action item to do dig up the old work from 2000. Matt agreed and gave an instance of this being done in the past. Al - slide will not produce the color green with Henk's work Saverio - like to have extension of the traceroute measurement draft taking into account additional measurement like ping etc. Matt - reluctant to open a can of worms. Yet he did open. :) Wanted to have clean separation between loss and delay measurements. Maybe they need to ride solo with clear procedures. Random thought. Joe Kopena From Drexel university - That would be useful - worthwhile in collecting it in a central way - Joe mentioned that this is a serious problem for MANETs. On mobile adhoc networks packet loss is significant and it complicates measurement. - Al agreed. He started characterizing them in the applicability statement. Kind of implementers guide for those 2 metrics. - Lars - What he heard thus far - Suggested ideas that were thrown were all close to the scope of the current charter. If folks are interested in further work from the charter, they need to prepare for a BOF. Matt clarified that by asking if the Charter needs a change the idea should be thrown on the list. Henk suggested we discuss this on the list as well.