From fs@hep.ucl.ac.uk Wed Mar 26 13:44:54 2003 Date: Fri, 7 Mar 2003 10:17:47 +0000 From: Frank Saka To: MB-NG@JISCMAIL.AC.UK Subject: Minutes of the MB-NG meeting 28 February 2003 Minutes of the MB-NG operational meeting Date: 28 February 2003 Time: 10:30 am Location:UCL Pearson media lab and via video-conferencing Tel: 020 7679 7029 Present: F. Saka, P. Clarke, A Di Donato Via Video-conf: S. Dallison, G. Fairy, R. Hughes-Jones Actions from 4th Collaboration meeting. ======================================= 4.9 S. Bhatti & UKERNA: Arrange to meet with J. Couzens and Experiment 8 people to discuss policies, classification and policing for QoS by 15th Nov 02. Progress: S. Bhatti visited UKERNA on the 21st November. 13/12/02 Ongoing. M. Allenby will help S. Bhatti on setting up a workshop on task 5. 10/01/03 F. Saka will push S. Bhatti on this issue. 31/01/03 A meeting is being arranged with S. Bhatti at his next available slot to discuss this task. 14/02/03 Meeting with S. Bhatti has been arranged for next week. 28/02/03 A draft report has been written by M. Rio. Efforts should be made to meet once a week with S. Bhatti until this task is complete. Gantt Chart: ============ Task 3: Traffic generation and measurement (equipment provision) -------- 08/11/02 ACTION: R. Hughes-Jones to arrange meetings to discuss options and what we loose if we do not have GPS. 10/01/03 No progress. This is a low priority. ACTION: UCL to investigate access to UCL based GPS time server (Speak to Lewis Grantham or Piers O'Hanlon). ACTION: S. Dallison to investigate the availability of a time server at Manchester. 31/01/03 Possibility of using NTS servers synchronised by the Rugby clock. The resolution must be checked. 14/02/03 No progress. 28/02/03 Manchester have a GPS system which they are investigating (works with Windows but not yet with Linux). Task 5: Classification/policies/policing for QoS -------- See Action 4.9 above. Task 9: Managed bandwidth service -------- 01/11/02 On hold until network is in place. 28/02/03 P. Clarke commented that UKERNA should get more involved with this. Task 10: Plan extension of demonstrations to sites in Europe and US -------- 01/11/02 On hold until network is in place. 31/01/03 A talk on MB-NG and our use of the Spirent equipment should be given at the Internet 2 members meeting. Task 11: The deployment and integration of the Middleware and APIs (GARA) -------- 13/12/02 ACTION: Javier will arrange a meeting with V. Sander (GARA) in February in London. Also with Leon Goomans and S. Bhatti. 31/01/03 Presentation days on GARA and AAA has been arranged for the 20 and 21 February 2003. 14/02/03 Ongoing. 28/02/03 The meeting went well. Volker Sander will be helping us to setup GARA. We will provide some development help. A CVS repository and a mailing list will be setup at UCL. Task 12: High throughput programme -------- A document has been produced and has been sent to the MB-NG mailing list. Members are asked to make comments and send them to S. Dallison. 31/01/03 After P. Clarke's recommendation S. Dallison is adding timescales and specific information on the experiments. 14/02/03 Ongoing. 28/02/03 Done. 10/01/03 S. Dallison has installed Globus and is intending to investigate the performance of disk-to-disk copy using Grid FTP. The ongoing work on TCP will be integrated at a later date. 14/02/03: We must agree on host-names and apply for the Globus certificates. 28/02/03 The host-name have been agreed and deployed. 17/01/03 It will be interesting to do GridFTP transfers between UCL and Manchester through the MB-NG network. Paul Mealor at UCL can help with this. 14/02/03 ACTION: S. Dallison to look into the possibility of setting up a central Grid authority. 28/02/03 This will be abandoned since it is too much work. Equipment ========= Status of testbeb ================= 13/12/02 Delivery of the 7600 OC48 line cards is delayed due to manufacturing problems. Delivery of the 7600 Gigabit Ethernet line card has slipped to Q1 2003. 14/02/03 The news from Harvey Lang is that the new OSM OC48 line cards and the Gigabit Ethernet WAN cards are orderable. They are currently on "product hold" until production rumps up (allocation on greatest urgency). M. Mckeown will give us the estimated delivery date later. 28/02/03 The delivery dates supplied by Mike Mckeown is 1st April for the OC48 line card and 1st May for the GE-WAN card. ACTION: P. Clarke to send a mail to Mike Mckeown to check on how firm those dates are ACTION: Alternative plan should be made that does not depend on the timely arrival of the Cisco cards. E.g. Control plane or MPLS. Running MPLS in the core network should be discussed with J. Couzins. 10/01/03 Status: TAC case has been raised. The Cisco IOS at CERN will be updated today (from 12.3-11-E1 to 12.3-11-E3). The DataTag link has been reserved and QoS will be tested today. 31/01/03 The outcome of the tests show that the Cisco 7600 router cards are not working properly. We require the enhanced version of the cards. This may be a problem for DataTag. 14/02/03 Huw Jones has forwarded the setup details to Cisco USA. He will let us know when results are available. Meanwhile we will test the Gigabit Ethernet WAN at CERN to see if it supports QoS. 28/02/03 Huw Jones: The Cisco engineers have reproduced the bug and are working on an IOS fix. QoS does not work with the version 1 GE-WAN. 13/12/02 I. Bridge: There is the possibility of a project TAC support and access to a CISCO techie (Huw Jones). Status: The issue is currently left with Mike Mckeown. It was agreed that we will push him on this at a later date. 17/01/03 ACTION: P. Clarke to contact M. Mckeown about TAC access. 14/02/03 Ongoing. 28/02/03 The current strategy is to report to Huw Jones. We will wait until the new line cards arrive and if there are problems with them we will pursue this more. Task 7: e2e Network equipment configuration -------- Experiments ============ Task 8: QoS Experimental measurements -------- 13/12/02 ACTION: A schedule of tests should be drawn up. Ongoing. 14/02/03 This action is postponed until the March report is finished. It also depends on the available equipment. Spirent would like us to test their QoS conformance test suite. We should look at it, but we should not spend too much time on it. Spirent would also like: Reports on our tests. A mention in publications and at network workshops. Any TCL scripts developed for tests. For MB-NG's use of Spirent equipment to be presented to Internet 2 Tests are being done on the GSR. Test on the 7600 will be started at the next week. 14/02/03 A report on the OSR and GSR tests are being written. 28/02/03 A copy of the report should be sent to Spirent and Cisco stating that we want to make it public and asking if they could comment on things they would like changed. AOB ===== Next meeting: 07th March 2003. ------------------------------------------------------------------ Tests for the March report. Booking Cisco's Bendfont lakes for standalone tests. This will be interesting after the completion of the project report. We need to discuss: - Equipment we require and confirmation of availability - Plan of test we need to carry out (We will only have two days). - Who will be going. Proposed test for MB-NG QoS 1. Latency(end-to-end): Back-to-back and through the network. For each packet size. Without congestion. 2. Three streams. Maximum throughput through the network. No policy. Packet size 1472. 3. 2LBE and 1BE flow with policy 98%BE 1%LBE 4. Policing and re-classification. 5. Rate limiting. Limiting the 2.5 Gig POS interface to 1.0 Gig. If this works then we can assess whether we can congest the GSR's output port. This may allow us to QoS. We must also check at what level MDRR works. Based on the results of tests 4 and 5, we will decide whether to use the 7200 in the multi-domain tests or not. 28/02/03 Experiments 1 to 3 have been completed. Issues: Throughput measurements for Manchester to London is only 400 Mbit/s per host compared with 957 Mbit/s in the reverse direction. ACTION: Check the tests again without the Manchester Firewall and Web server. 28/02/03 We removing all unnecessary processes running at both end-point (NFS, port map ect.) to try to pin-point the problem. 28/02/03 Manchester are looking to perform loss throughput and latency tests across the network. R. Hughes-Jones will investigate and buy some raid disk controllers to do high throughput tests from between the MB-NG sites. Estimated price is 325. It was proposed to have a face-to-face technical meeting on 21st March 2003 to discuss and bring together our results. -- Department of Physics & Astronomy, University College London, Gower Street, London WC1E 6BT Tel:+44 20 7679 3775