GLOBAL LAMBDA INTEGRATED FACILITY TECHNICAL ISSUES WORKING GROUP Minutes of the meeting of the GLIF Technical Issues Working Group held on the 13th of February 2005 at the University of Utah, Salt Lake City, United States. Kevin Meynell - Issue 1 PRESENT Name Organisation Country ---- ------------ ------- Erik-Jan Bos (Co-Chair) SURFnet The Netherlands Mike Collins ESnet United States Mike O'Connor ESnet United States Steve Corbato Internet2 United States Steve Cotter Internet2 United States Serge Figuerola i2CAT Spain John Graham UKLight United Kingdom René Hatem (Co-Chair) CANARIE Canada Jun Jian CANARIE Canada Mark Johnson MCNC United States Akira Kato WIDE Project Japan Otto Kreiter DANTE - Cees de Laat U. van Amsterdam The Netherlands Tim Lance NYSERNet United States Edoardo Martelli CERN - Joe Metzger ESnet United States Kevin Meynell (Sec) TERENA - Jörg Micheel NLANR United States Roland Nuijts SURFnet The Netherlands Kevin Oberman ESnet United States Bill Owens NYSERNet United States Donald Petravick Fermilab United States Mark Prior AARnet Australia Sylvain Radot Caltech United States Dave Reese CENIC United States David Richardson U.Washington United States David Sinn U.Washington United States Jerry Sobieski MAX United States Rick Summerhill Internet2 United States Alan Verlo UIC & StarLight United States Steven Wallace Indiana University United States Rod Wilson Nortel Tech Labs United States Linda Winkler StarLight United States Chu-Sing Yang NCHC United States 1. WELCOME Erik-Jan welcomed everyone to the meeting and said that he had invited René Hatem to share the chairing duties so that he could better participate in some of the discussions. He also thanked Rick Summerhill and Linda Charlesworth for organising the meeting room and refreshments. 2. CHANGE OF WORKING GROUP NAME Erik-Jan announced that he had received a request from TERENA to change the (short) name of the working group from 'TEC' to 'Tech'. This was because 'TEC' was well-known in Europe as an abbreviation for the TERENA Executive Committee and there was the potential for some confusion. It was unanimously agreed that name of the group should be changed to 'Tech' forthwith and that the mailing list address should also be changed to reflect this. ACTION 20050213-1: Kevin Meynell to change name of mailing list to 'tech@glif.is'. 3. THE ROLE OF TERENA IN GLIF Kevin gave a short presentation about TERENA's role in GLIF. It was agreed during the GLIF meeting in Nottingham that TERENA would assume responsibility for the secretariat functions, starting from 1 January 2005. TERENA was a not-for-profit association of European NRENs, currently with 15 staff located in Amsterdam. The secretariat functions entailed the hosting of the GLIF website that had recently been moved from KUN to TERENA, and was currently being updated and improved. The existing GLIF mailing lists were also now being hosted by TERENA and included a general list for all those who had attended previous meetings and those that wished to be subscribed, plus separate lists for each of the working groups: Governance & Growth , Research & Applications , Technical Issues , and Control Plane & Grid Integration . In addition to this, TERENA would administratively support the working groups. Liaison persons had been assigned to each of these as follows: Governance & Growth - Karel Vietsch, Technical Issues - Kevin Meynell, Control Plane & Grid Integration - Licia Florio, Research & Applications - to be determined. Starting from 2006, TERENA would be responsible for organising the annual GLIF meetings, probably in the format of a one-day open seminar followed by a day of parallel working group sessions. The 2005 GLIF Workshop was being held in conjunction with iGRID 2005 and would be organised by them, although TERENA had been asked to assist with the programme. A contact address had been established for the GLIF Secretariat which was 'secretariat@glif.is'. 4. BEST CURRENT PRACTICES IN GLIF Erik-Jan presented an overview of the proposed contents of the GLIF BCP, and the authors who had agreed to write each section. These were as follows: 1. Introduction (Erik-Jan Bos) 2. Management Summary (?) 3. Overview of the GLIF (Cees de Laat?) 3.1 Optical Exchanges 3.2 Links in the GLIF 3.3 The GLIF edge 4. Usage of the GLIF (Jerry Sobieski?) 4.1 Definition of a Light Path 4.2 Connecting to the GLIF 4.2 Using the GLIF 5. Technical Details of the GLIF (Linda Winkler) 5.1 Building Blocks 5.2 Ethernet Framing 5.3 Protocol Conversion 5.4 10GE LAN PHY and WAN PHY 6. Further Reading (?) Appendix A. Glossary and Abbreviations (René Hatem) Appendix B. GLIF Topology Map (?) References (?) John suggested that some deadlines needed to be set if the document was ever to be completed. Jerry suggested that he should present it at iGRID 2005 in September, which would mean that it would need to be finalised some time before that. It was therefore agreed that each author should produce their sections by 15 March, and a first draft be compiled from these contributions. Erik-Jan asked who was willing to put together the document and undertake any necessary editing. Kevin volunteered to do this, and he also agreed to chase-up the contributions as well. ACTION 20050213-2: Erik-Jan Bos, René Hatem, Cees de Laat, Jerry Sobieski & Linda Winkler to write their sections for the GLIF BCP by 15 March 2005. ACTION 20050213-3: Kevin Meynell to chase-up contributions and put together draft document. ACTION 20050213-4: Jerry Sobieski to present final document at iGRID 2005. There followed a discussion about how various elements such as lightpaths and service descriptions should be defined. This did not reach any consensus, so René suggested that he would prepare some draft definitions for discussion. Erik-Jan also said that he would work on standardising service descriptions. ACTION 20050213-5: René Hatem to prepare some draft definitions for discussion. ACTION 20050213-6: Erik-Jan Bos to work on standardising service descriptions. Cees suggested to redefine a lightpath as follows... "A Lambda (LightPath) is a pipe where you can inspect packets as they enter and when they exit, but principally not when in transit. In transit one only deals with the parameters of the pipe: number, color, bandwidth." 5. LESSONS LEARNT DURING SC'04 Linda stated that the experiences of SC'04 had shown that better organisation and understanding of requirements was needed in future. There had been insufficient information on both the number of circuits and booths required by the exhibitors, and this had led to scheduling problems. Many issues ended-up being resolved very late in the day, and whilst everyone's requests for circuits were eventually met, this meant scheduling some demonstrations at odd times. The planning for SC'05 (12-18 November 2005) had already begun and the requirements of equipment vendors and circuit providers would need to be known by June. Users also needed to be made aware of the steps they needed to take to get connectivity, because many assumed they simply needed to make a request to conference organisers, when in fact they also needed to make arrangements with NRENs and/or other institutions. René asked about the capacity and how users were matched with this. Linda replied that SC'04 had 20 x 10 Gbps circuits which was a five- fold increase on the previous year. Only a few users actually required the full 10 Gbps, but most needed at least 2.4 Gbps which ruled out the use of 1 Gbps circuits. Scheduling was on the basis on 'first come, first served' although there were still a few clashes that needed to be resolved and it made bandwidth sharing more difficult. Erik-Jan thought it desirable to make resource information available on the web, in a format similar to that shown by Cees (see http://staff.science.uva.nl/~delaat/sc2004/index.html). However, this was related to the next agenda item and should be discussed there. In the meantime, Linda and Jerry agreed to put the necessary procedural information for SC'05 on the web before June. ACTION 20050213-7: Linda Winkler and Jerry Sobieski to put procedural information for SC'05 on the web before June 2005. 6. GLIF SCHEDULABLE RESOURCES Erik-Jan said they had concluded during the Nottingham meeting that whilst static maps provided a nice overview of resources for PR purposes, they did not provide enough details, became outdated quickly, and sometimes made incorrect assumptions (e.g. ten 1 Gbps circuits were sometimes indicated as a single 10 Gbps circuit). It was therefore felt that a more standardised system of representing GLIF resources was needed which would enable users to automatically find paths, spare capacity, and the type of connections available. He had investigated how a dynamic system might be developed and followed-up a suggestion from Franco Travostino that the semantic web concept might be used. This is an attempt (led by Tim Berners- Lee) to create a universal mechanism for information exchange by associating content with machine-readable meanings. It aims to allow highly-structured webs of information to be generated from the existing World Wide Web, using the Resource Description Framework (RDF) metadata model based on XML. If RDF is used correctly, it should be possible to browse through the information and construct the necessary metadata. Erik-Jan proposed that the group investigate the possibilities of the semantic web in more detail, and build a small prototype by mid-April. This would require some clarification as to exactly what were GLIF resources, and to come-up with a suitable (RDF?) template. The group agreed this was worth further investigation and suggested that Erik-Jan pursue this. René also agreed to help clarify the GLIF resources. ACTION 20050213-8: Erik-Jan Bos to investigate the suitability of the semantic web for cataloguing GLIF resources by 15 April 2005. ACTION 20050213-9: René Hatem to clarify GLIF resources by 15 April 2005. 7. iGRID 2005 PREPARATIONS Linda announced that preparations were well underway for iGRID 2005 and the draft programme was now available online (at ftp://ftp.evl. uic.edu/pub/OUTgoing/maxine/igrid2005/). This meant that users requirements would need to be known by the end of March so that additional bandwidth could be arranged if necessary. There was currently 50 Gbps into UCSD, but they were hoping to increase this to 100 Gbps. In effect, this event would be a dress rehearsal for SC'05. The timelines and milestones for iGRID were due to be agreed at a meeting in a couple of weeks time, so these should be available shortly afterwards. Linda was asked to send these to Kevin to put on the GLIF website, whilst Kevin was also asked to establish links to the iGRID 2005 and SC'05 events. ACTION 20050213-10: Linda Winkler to send iGRID 2005 timelines and milestones to Kevin Meynell by 6 March 2005. ACTION 20050213-11: Kevin Meynell to put iGRID 2005 timelines and milestones on the GLIF website. ACTION 20050213-12: Kevin Meynell to establish links to iGRID 2005 and SC'05 from the GLIF website. All requests related to iGRID 2005 should be sent to Maxine Brown 8. OPEN OPTICAL EXCHANGES Cees gave a presentation about optical exchanges. Rick thought that the discussion over what was an open exchange or not was often misleading as an theoretically open exchange could still have restrictive policies (such as the requirement to have peerings with two or more other providers). In addition, the meaning of a neutral exchange point could also be defined in different ways. This led to a discussion on this issue, but due to time constraints, Cees was asked to continue it on the mailing list. As this also had implications for the Control Plane Working Group, it was suggested this issue needed to be resolved before that group met. ACTION 20050213-13: Cees de Laat to start discussion on definition of open and neutral optical exchanges on the mailing list. 9. PROTOTYPING LAMBDAMONS Jörg gave a presentation on the lambaMONs developed by NLANR which are used to passively measure DWDM optical networks. They enable the collection and real-time analysis of IP packet data from any active 10 Gbps wavelength carrier on a DWDM optical link. Cees mentioned that GlimmerGlass apparently also had an optical monitoring system. Jörg replied these were actually standalone devices that could not be built into racks. Erik-Jan asked why the lambaMONs were installed in rack-mountable cases as this increased costs and limited the number of expansion slots. Jörg replied a rack-mountable case was a relatively insignificant proportion of the overall cost of the equipment, but was necessary in order to install it in Internet Exchanges. OPEN ACTIONS 20050213-1 Kevin Meynell to change name of mailing list to 'tech@glif.is'. 20050213-2 Erik-Jan Bos, René Hatem, Cees de Laat, Jerry Sobieski & Linda Winkler to write their sections for the GLIF BCP by 15 March 2005. 20050213-3 Kevin Meynell to chase-up contributions and put together draft document. 20050213-4 Jerry Sobieski to present final document at iGRID 2005. 20050213-5 René Hatem to prepare some draft definitions for discussion. 20050213-6 Erik-Jan Bos to work on standardising service descriptions. 20050213-7 Linda Winkler and Jerry Sobieski to put procedural information for SC'05 on the web before June 2005. 20050213-8 Erik-Jan Bos to investigate the suitability of the semantic web for cataloguing GLIF resources by 15 April 2005. 20050213-9 René Hatem to clarify GLIF resources by 15 April 2005. 20050213-10 Linda Winkler to send iGRID 2005 timelines and milestones to Kevin Meynell by 6 March 2005. 20050213-11 Kevin Meynell to put iGRID 2005 timelines and milestones on the GLIF website. 20050213-12 Kevin Meynell to establish links to iGRID 2005 and SC'05 from the GLIF website. 20050213-13 Cees de Laat to start discussion on definition of open and neutral optical exchanges on the mailing list.