Minutes from MB-NG Task 2 meeting. Wednesday 22nd May 2002 10:30 am. Location: University of Manchester. Present: S. Bhatti Y. Li A. Di Donato R. Hughes-Jones F. Saka G. Fairey S. Dallison Actions and issues from the last meeting Friday 03 May 2002. ------------------------------------------------------------ Required measurements suggestion by Saleem Bhatti: Make a table of measurements against traffic patterns. Each measurement with a specified traffic pattern must be justified otherwise we do not perform it. ACTION 1.0: Richard Hughes-Jones will incorporate this method more formally. PROGRESS: A template document has been produced which lists definitions of tests, how to analyse the measurements and what we expect to learn from them. ACTION 1.1: Manchester List of measurements wanted. PROGRESS: This is ongoing. See preceding action point. ACTION 1.2: Ian Bridge to look into GPS cards for PCs to enable one-way delay measurements to be performed. PROGRESS: See minutes from MB-NG operational meeting on 24th May 2002. ACTION 1.3: UCL continue work on metrics. PROGRESS: A draft document has been produced. Discussions during this meeting led us to the conclusion that IPPM definitions are inconsistent in places so the GGF definitions (which also derive from IPPM) will be incorporated into the metrics document. ACTION 1.4: S. Bhatti will setup a tech-MBNG mailing list. PROGRESS: This has been set up and is being used. List members are: Saleem Bhatti Yee-Ting Li Frank Saka Andrea Di Donato Peter Clarke Ian Bridge Richard Hughes-Jones Gareth Fairey Robin Takser Chris Cooper Jon Crowcroft Tim Chown ACTION 1.5: S. Bhatti to pass around the url for a paper on Network entropy. Network temperature measurement. ACTION 1.6: Web-site: mbng.net There is a possible clash with UCL policy which states that all web site hosted at UCL must have UCL in the url. If this is true, R.Hughes-Jones has offered to host it at Manchester. PROGRESS: UCL proposed policy didn't go through. The web-site will be set up in two weeks. Current Status ------------------------------------------------------------ A. Di Donato gave a brief summary on the UCL draft metrics document. R. Hughes-Jones said, from GGF work, there are a few some inconsistencies over IPPM. He suggested GGF terminology be used instead. - A discussion on clocks (host time and wire time) took place. R. Hughes-Jones presented some plots illustrating that the error of the host clock reduces with increasing sample size. It was noted that this was not true with jitter measurements. ISSUES: - We need to calibrate the nodes and routers. We need to study the calibration error. If the errors are too large, then we need to study the nature of the errors. - Real-time Linux should also be looked at ACTION: R. Hughes-Jones to circulate GGF work. ACTION: A. Di Donato to use better defined GGF terminology in metrics document and circulate the document for comments. ACTION: A. Di Donato to distribute Surveyor and RIPE documents. AGREED: Metrics and background document will be incorporated into a single document. G. Fairey did some test on packet loss. He found that packets were being discarded at the IP level (by examining ip-discards in the /proc directory). He used 1400 bytes UDP packets, Intel Gigabit Ethernet cards with flow control enabled and connected back-to-back, Linux kernel version 2.4.14. No packet loss was observed when a switch (CISCO 3508) was present. ISSUES: Packet loss at the IP level must be investigated. R. Hughes-Jones presented his draft document "Measurements and methodology for MB-NG". This will go towards the completion of Task2. ISSUES:- The accuracy of the clock we require depends on the minimum inter-packet time. It also depends on the type of measurement we want to make. - In each of the measurements we define, we must justify the chosen clock resolution. In this way, if we have to borrow specialist equipment, measurements can be scheduled around the availability of the equipment. - Is the PC system clock good enough? i.e. stable enough? ACTION: R. Hughes-Jones to talk to RIPE guys and investigate clocks Summary of Outstanding actions, questions, issues and progress. ------------------------------------------------------------ 1) QUESTION: How do we get data from the core routers. Do we need PCs at POP? PROGRESS: S. Bhatti says we should be able to get PC at the POPs, but this should be confirmed with UKERNA ACTION 2.1 Richard to confirm this with UKERNA. 2) ISSUE: SNMP-Agent may not be able to keep up with the card (not all frames and errors are reported). Thus we may not be able to completely trust it's reports. PROGRESS: S. Bhatti says vendor implementations of SNMP is at fault. A student of his working with Sprint has confirmed that the SNMP statistics can lie under certain circumstances. Test must be done to assess SNMP limits. The students work is under non-disclosure agreement, but part of it will be made available in about two months. ACTION 2.2: S. Bhatti to distribute the work once it is available. 3) What is the capture time we can playback? Ian Bridge says we can capture up to 15 seconds of traffic with an analyser. QUESTION: Is this enough? ACTION 2.3: UCL to look at RUDE and CRUDE tools and media playback tools to be included in the paper on background traffic. 4) QUESTION: How do we know when background has stabilized. Need to assess this by measuring the transmission through the network for various network states and decide when the steady state is reached. PROGRESS:The work mention above (ACTION 2.2) also deals with measuring the "latent heat" of the network i.e. when the network is locally stable. The details will be made available in the same paper. 5) QUESTION: Where is the background traffic directed? PROGRESS: This should fall out from Task 5. 6) ISSUES: We must agree on: What tools we need. Who will work on what. PROGRESS: Agreements have been made for UCL to write a note on background traffic and Manchester to write a note on the measurements. The required tools should fall out of this work. 7) ISSUES: The ideal equipment seems to be GPS, however it may require an aerial to be mounted on top of the buildings. PROGRESS: It has been agreed that the Rugby clock will be too coarse. ACTION 2.4: Ian Bridge will continue looking at the GPS solution. 8) ISSUES: Simulation? This aspect of the project has been overlooked. It requires at least one person full time. PROGRESS: No progress on this. Summary of Actions from current meeting. ------------------------------------------------------------ ACTION 2.5: R. Hughes-Jones to circulate GGF work. ACTION 2.6: A. Di Donato to use better defined GGF terminology in metrics document and circulate the document for comments. ACTION 2.7: A. Di Donato to distribute Surveyor and RIPE documents. AGREED: Metrics and background document will be incorporated into a single document. ACTION 2.8: R. Hughes-Jones to talk to RIPE guys and investigate clocks ------------------------------------------------------------ Next meeting: 28th June 10:30 am at UCL.