GGF NM WG Notes 05/03/05 Call began at 11:00 a.m. EDT. Attendees: Martin Swany Eric Boyd Richard Hughes-Jones Jason Zurawski Jim Ferguson Jeff Boote John Estabrook Dan Gunter Susan Evett (scribe) AGENDA Work with new schemas (Martin et al) GGF14 (Chicago) discussion The most recent versions of the schema are available on the URLs that Jason sent out recently. They feel they’ve arrived at something their willing to make public. No more hand waving about things to be fixed in the next ‘time period’ – still needs lots of documentation but it is free to be reviewed and tested. Richard asked if it was the stuff that’s on the web page – references to client/server are on the web page. Martin noted that it was on the UDel page; Susan will link to this on the GGF NMWG page (nmwg.internet2.edu/). As soon as it is linked, Eric will send out an email pointing folks to them. Martin asked that the link from the main page be a link to the UDel pages instead of a copy onto the Internet2 pages. Eric asked if we should walk through the new materials or discuss the GGF14 plans – Jeff requested getting GGF14 agenda out of the way first. Group experienced some difficulties getting through to some of the pages. Eric noted that the links within the UDel directory are broken. NM_BASE – Martin described it – hasn’t changed much, pretty straightforward. Schema says nothing about the scope of identifiers which is to provide flexibility. In metadata section, pretty much the same thing. Anticipate those to be expanded; top level definitions. Discussions re: Numb_Bytes parameter – how many bytes should each parameter have? New: Common_Time – an optimization that allows you to group data elements that have the same time stamp (which appears in each data element) so you can find all those with a common start time. We’ve debated a lot about time so there are 2 ways we can see it being represented: inline (timestamp, utc, unix-encoded utc, iso-time) time type and value or not. It is optional – you don’t need to have timestamp anywhere. Traceroute values are going to return the time of kickoff but not necessarily the end time. Richard noted that you could specify time in a simple way but you could make it very complex if necessary/desired. Dan suggested that the group look at the Time_Value material: NM_TIME.RNC. Martin reviewed the examples at the top that show ‘common time’. Anticipating that there are equivalent representations (necessary evil because, to define a base schema that allows flexibility, this is necessary). Within a week or so, we’ll have examples of best practices; how an implementation would reference a particular encoding. Jeff asked: You’re referring to the different types of code – Unix is ‘ping’ but he suggested noting that a counter is one, as well. Martin: we’re counting on you to provide NTP and counter examples for us. John said he would provide some examples, as well. Richard commented that there needs to be a way to translate from some random clock back into Unix time. Martin agreed that there needs to be an option for ‘reference times’ and ‘time offsets’ in requests. Richard said he was relieved that they had thought about it – it hadn’t been discussed in the past, so it could easily have been missed. Dan said that they don’t ‘explicitly’ support translations but it is possible with the flexibility provided. Richard noted that the ‘home-made counter’ creator would have to provide some reference. Jeff said that there would be functions like Skew_Rate that would need this information to be created! Martin said there is no prohibition from using any particular time reference but enclose a datum element that incorporates some universal reference time so that you can start with whatever time reference you like but there is something off which others can calculate an offset. Jeff felt that it makes a lot of sense to factor out the time elements and use the ‘common time’ whenever possible. Time_Element has start_time and end_time parameters; they are not constrained in any way but there is a Boolean in there for flexibility. The Developer’s guide will include references to the fact that there is a lot of possible duplication from some parameters if set in certain ways. Richard agreed that having a ‘suggested’ set to start with would be sensible to get folks going. Martin agreed; having the flexibility makes it possible for folks to make changes later. Eric jumped in to mention that, with only 15 minutes left, the group should talk about the meeting in Chicago. Martin asked for folks to continue looking at it and get him comments. Eric thanked Martin for his work; Martin thanked Jason and Dan for their efforts. Everyone agreed that this is serious progress. Next phone call – scheduled for next week – Eric suggested pushing it back 2 weeks to May 17, same time. Richard suggested, based on discussion from Seoul, that the group move 1 hour earlier so that the Asian participants can participate. Jeff noted that the US/Europe is moving to DST and they aren’t; Richard noted that this still 12 Midnight there; there will be a discussion on the list as to making it an hour earlier – who would add, who would drop. GGF14 – call for subsections will happen soon. Mark felt that the suggested Monday time slot would be conflicting with a lot of other presentations. He suggested holding the 8-hour meeting on Tuesday and to ask early to avoid the conflicts with other groups. Eric asked, if it was held Tuesday, would people on the call be able to make it; Richard thought he would and suggested that we indicate which items folks would be interested in not conflicting (high-perf networks, for example). Eric will send out an email asking for folks to identify what side sessions they are interested in; Eric said that he and Mark would plan to stay for Wednesday to report out on the WG meeting. He asked what the presentations should be; he asked Martin and Dan to identify what presentations they would be interested in presenting and Eric will fill in around it. Richard asked if a discussion of the prototype schema should be presented in the open session and then again in the 8-hour group. Jeff suggested getting this out early so it could be discussed at a call before the GGF14 meeting. Richard asked about the tutorial that was offered on Web Services – should we pickup on it? It might be on Sunday – Eric felt it could be fit into the 8 hours on Tuesday, depending on how long it is. He asked Richard to find out and get back to us. Richard asked if, for the next GGF, the network measurements research group, there would be interest in getting folks to think/discuss how to measure the available bandwidth on a 10 gig link. Richard asked if the preparation of demos were progressing okay for the v1 schemas; Dan said yes, the SONAR project will use something similar to this. Martin agreed it would be nice to show two types of languages talking to each other; he said he had agreement with Internet2 to do that but the work was currently on the back burner. He invited volunteers to use the UDel host; Dan has the ability to get folks on, if interested. ACTION ITEMS: 1. Susan to link the UDel URLs to the newest schema information on the NMWG web page. When they are linked, Susan will send out an email letting folks know it is accessible via the WG web page. 2. Jeff and John to provide examples of NTP and counter examples to Martin for NM_TIME.RNC. 3. Everyone to review Martin/Dan/Jason’s new material and get them comments. 4. Richard to begin discussion on the mailing list re: starting calls 1 hour earlier to determine who would add/drop the calls. 5. Eric will send out email requesting topics members of the WG don’t want to have conflict with the f2f meeting and inform the group. 6. Martin/Dan to identify what presentations they’d like to make at the f2f meeting in Chicago and send that to Eric. He’ll fill out the rest of the agenda. 7. Richard to check with the folks who offered to provide a tutorial on web services to see how long they would need for the presentation; he’ll provide that to Eric to determine whether it should be part of the full-day meeting or held on Sunday. Next call is on 17 May 2005 at the same time. Call ended at 12:03 EDT.