Joint Engineering Team meeting November 25, 1997 Noon to 5pm NASA Goddard DRAFT Notes [from Suzanne Burgess] I. Participants Javad Boroumand [javad@isi.edu], ISI/NASA Scott Bradner [sob@harvard.edu], Internet2 (phone) Suzanne Burgess [sburgess@snap.org], DynCorp/LSN Dick desJardins [desjardi@eos.nasa.gov], NASA Phil Dykstra [phil@arl.mil], ARL (phone) Ken Freeman [kfreeman@mail.arc.nasa.gov], NASA (phone) Doug Gatchell [dgatchell@nsf.gov], NSF Andy Germain [andy.germain@gsfc.nasa.gov], NASA Sally Howe [howe@ccic.gov], NCO John Jamison [jjamison@vbns.net], MCI/vBNS (phone) Mark Luker [mluker@nsf.gov], NSF Paul Love [epl@internet2.edu], Internet2 Grant Miller [grmiller@nsipo.nasa.gov], NASA George Seweryniak [seweryni@er.doe.gov], DOE Dharmesh Shah [dharmesh.shah@gsfc.nasa.gov], CSC/NASA Tom Shaver [shaver@romulus.ncsc.mil], DOD/NSA John Steedman [john.steedman@gsfc.nasa.gov], NASA II. Actions - Continuation of efforts to coordinate multicast connections between FedNets (K. Freeman, FedNets) - Draft a discription of principles to match the principles (e.g., what is transit) (Bradner, Seweryniak, Jamison, Luker) III. Proceedings A. Review of Actions from 10/14 meeting The notes from the October 14 JET meeting were approved and any comments/changes should be sent to the JET mail list. The actions from the last JET meeting were reviewed and discussed. The action on information collection was defered until the next JET meeting. A working group may be formed to identify which information (e.g., interconnection points) should be collected and how it should be presented. On other action items, agencies will report on applications at the next JET meeting, and Aiken and Dykstra will report on FTS 2001 at the next meeting. With Dick DesJardin's retirement from NASA, the leadership of the Joint Engineering Team will change. The JET will become bi-coastal with Ken Freeman co-chairing the JET from the west coast. B. The ATM Addressing White Paper The cover letter and the paper were changed as discussed at the last JET meeting , making it clear and recommended approach and draft. Fed agencies were not required to use the plan, other comments small, the letter reflects what the white paper says, it is a source of addressing for the future, no comments on the paper LSN agenda will include submit the letter on the atm cover letter, add either a statement or a coordination line to the bottom of the letter, send comments to paper list tom will forward the broad questions on the paper will to the jet paper will also go back fed agencies will be mailed out by dec 15 action items interconnect points , id them and what kind of information to collect, what about monitoring, sould not rule out operational information. Arch infor now, who is the contact if there is a problem, phil will report at the LSN, further refine the information and operational information (lsn agenda - what is the jet doing) monitoring issue - privacy issues on what should be monitoring. Was discussed at EOWG. Phil will bring them mou info from steve - whole document was in limbo??? What is the scope on the international section, should be only STARTAP what about political issues, other peering points steve schultz and kevin tompson talk Delay to chicargo exchange, native multicast after that point, routers will be set up, dates coming for the ongoing activity and working group, more details at the next JET meeting, presentation at the next meeting, detail who what when where, stumbling blocks. JET meeting jan 14 DOE E243, mbone workstation with monitor and demonstrate PictureTalk allows screen capture and web browser. Convert it the mbone, last CIS meeting, transit questions, DOE policies on that, kevin talk to Joe at ESnet, set up access list, need official word, regular AUP and scenario applies, problem could keep going but has to have esnet site, one end point of any tunnel, no tunneling through esnet, ietf guidelines include no mbone tunnel should traverse someones structure. Native multicast, we don't have the technology to avoid this call it a working group with ken the chair, with report. Work on guidelines, restrictions, transit policies, Kevin thompson - preliminary agreement not talking for esnet. Joe e-mail policy donot want to allow multicast traffic accross esnet, non esnet be delivered into the vbns. Not sure if delivered to a esnet site, george will talk to joe MOU we would set the mou on the side, instead work on the issue that arose and then report those to the LSN. issues, and international section will fall under these below. HOW - what are they agreeing on?? Scenario with the lastup - would the separate peering arrangements and AUP take care of it?? What is the policy. How do they fit together? Could be as simple as the agencies agree to honor the agencies agreements. Was in version .03 - full mesh of aup understanding. Transit traffic was not permitted so we don't need a special section, no transit without explicit approval. No specific international issue. Startap another fednet, yet it is not the same. No aup for startap, mou will either say that there is a principle or need a mini-paper policy/political - what are the two organization, what does this mean, yadda yadda. Sally Howe. Relationships and what an agreement would mean - sally has a process, 99 blue book should say it. Who are coordinating, day-to-day operations of the connections - ngi engineering organization, operational responsibility for the fednets. who is there - start as a discussion thread . which networks - javad lead, with sally [Javad - talk what is the jet - info or engineering, qos example writing a white paper, how is the process vis-a-vis the NGI implementation team. We are not the implemenation team. How do you organize - LSN must fix the implemenation team. Jet is the coordination. Maybe the implemenation team can task the JET. What is the relationship with the LSN. maybe should go back to the eowg/fepg relationship maybe need the implementation team should. Look at the implementation plan. ] AUP and principles of traffic exchange - Bradner what is now and where want to be. applications support - this was completely unclear, why doing this. Should we enumerate them? Was a placeholder. Steedman. How does the jet think about applications, info, what needs to be discussed. Want a non-restictive policy. Maybe a list of talking points that we want to showcase. Could say what was just stated. John start a thread on the e-mail. Want to support the applications team, but hard to do the applications support function, what are their prerfomance requirements, proud 100 . what should the relationships be - easy the use the attributes were discussed at the last eowg, make all those transparent to the user. NGI is not bandwidth. One network must recognise QOS,maybve the jet is an applciations support function. Make the infrasture as transparent as possible for the applications possible - need this to develop a common infrasture to help the applications developers. difficulty serve two different goals - mesh of aups and the other how we wish things were - universal traffic. Separate them by writing a vision statement for the next round. What does I2 means when it network? Message on aup is changing, scot will have to track it down, regional networks, gigapopos, vbns, won't work on basis of applications source of packet carried - what is not allowed and everything eles is ok Not the case with the vbns. Blue book distributed I2 and NGI language in the executive summary, LSN section pages LSN meeting at SC 97 sally talk about, applications on highway one, relationship with NGI, recap of talk - ngi connectivity to orlando where do you want what is the JET's interest in this - should it be the implementation team, highway 1 is a vbns site. Implementation team - for I2 it was difficult to get to highway 1. Back to LSN what is the objective of this demonstation I2 meeting the week of april 13 different public relationship so maybe keep the demo separate from the I2, how do we tackle this problem, maintaining the campus is more than money from NSF, difference in money, etc. agency only NGI demo NGI goals - impl JET specific goals of agency interactions program difference between NGI and I2. Raise it at the LSN that we are in a partnership I2 wants to be seen as not feeding at federal trough. We will support the LSN as they direct demo picture tell live lan, h323 using nren, bandwidth half mega bit, pc max out so no full motion video, pc ibus card with all the extra stuff. Requires a highend machine. does not work will with 32 memory, pleased with results, shared white board, and text chat tool, take any application, share them, sends the whole bitmap. Share any application, live gateway, NASA had some isdn lines problems, system cost - $7000 total for 5 kits. Point-to-point but does not rule out multicast in the future. Limitation can't replace the camera, the cable is proprietary. It will monitor the bandwidth. Not sure why it drops out. First release of software. Works on both 95 and NT, not know for sure. Mbone has push to talk, and does no share application. Word document can be edit, stays on the local machine, has control, the one that orginated the application. Interoperate with other, like netmeeting, is based on livelan. Encryption - don't know. Ames is direct, you must register with a gateway ISDN, only live IP, good audio. NO one-to many demo on the isdn gateway. Doe is the next demo - NASA promote begin support the JET meeting with web based. Primary is audio, then pull the slides up, or link, third would be the video based. Interchanging them. NASA will have the one after DOE. ISI meeting Tom Kalil with MIT, use CAIRN. Setup with three machines at each end. Sparc ultra, one to send, receive video, and one to do audio. 22 frames per second. Used 96% on the spar, total dedication. 17 meg per second, 300 megahertz. Compression algorythnm was hacked - not released, public code that is released. Audio was 16 kilohertz Support high performance. Multicast from day one. Very expensive, show can be done. Tom Kalil very happy, public domain technology, NGI is encouraged to look into more public standards-based solutions. Take one of the meetings to ISI demo at NASA. After Jan 14 meeting, feb 11 meeting at NASA with a west coast site, S241 recommend that you can set aside the second Wednesday, but we are flexible fednet collection sites - nasa has proposed the spreadsheets, interoperability among the agencies. How work this problem, get POC, phil did already, subgroup to work it by the next meeting on site interoperable spreadsheet, maintence problem so must use the data already provided. Is the information dynamic? I2 is dynamic. Link to something that is linked to setup team lookin to how the information is kept, who will keep - team look at the NLANR stuff, nren maintain accessible sites? How many connections from university X? need commaniality of how the sites are named, compatable framework, automatically updated, example of boston uni. Nasa has crsa, in its list of sites, says crsa while the I2 list says boston uni. , javad say that boston uni. Should have that information update. Data mining on current information. The JET put out a guideline maybe??? Maybe there already is, who is going to do it with anyone to access, common data format , data is distributed, the jet recommends what data format phil has chair the subgroup, with dick will call him caida web site demoed at the next meeting with NASA to bring the projector alaska/hawaii agenda item for the next time, see phil e-mail and he will report on this at the LSN I2 gigapops? Paul will report fednets connected to the gigapops. Vbns and what are the others - mostly future connections, would put in a link for better performance and what are they using the the gigapop for - transit or gigapop service, maybe need some guidance, Scott's traffic exchange principles. difference between gigapop and NGI exchange, question of performance, purposes? Based on sites, all traffic - production traffic supporting reasearch. I2 needs to get this on a web page, on the agenda, paul chairs the topology working group - not clear yet. Vbns award is over in two years, program will continued or changed. JET and NGI implemenation john will do the white paper on applications jet to NGI we are not implementing the goal 2.1 implementation plan revised - the vBNS is 2.1 underway, 60 by end of calendar year, another round of awards, rural connections in the next round and they have applied, federal interconnections are in placed, quite a ways down the milestones in goal 2.1 applications are being done , applications support function, integration function is not there enhanced services are not funded by DARPA, they are funding the generation after next, while companies like cisco and others are doing these enhanced services. New cicso routers will support two classes of service, using RSVP hooked up. Need to take cisco and fore are working on, the Jet can focus with I2. CARIN not a goal 1 network Jet issue - nothing for goal 1, after getting the fednet connections, then the services will be asked and it will come vendors. Next meeting - jet role from goal 1. Invite bob aiken to speak. He is the jet chair on this. goal 1 is not being done, qos done on vbns. Ngi exchange points as far as ames, nren and vbns connection late fed. Nren get into chicago and wdc in december hayward and mountain view , who do we peer with? Nren have internal meeting who peer with, now with vbns as soon as turn up. Other peering could be the canadaians. Now do it on an as needed basis, peering at IP level, will not peer all routes, vbns is not know if they will peer all routes - within NASA how can , need to talk with esnet and dren, need as full peering as possible. Vbns will take all research sites. How do you separate routes, as opposed to full exchange. Gloss over in fednet, but won't in the international networks because have different policy. With bgp can do some of this specify sites within that community. Id which sites when the fednets peer - implementation to take this - can we nudged along, john jamison presentation on the bgp communities and what can be done. Under ken's agenda topic, kick start what should jet do about this Agenda - discussion what is being worked out on the talks with esnet, dren, disa. How do meeting start at 11am more talk on the mail list