![]() |
Enron Mail |
Hi Vince, I (and possibly Stinson as well) will be attending this initial
meeting looks like kick-off type of meeting. I will try to attend to drill into what they can offer and what we have committed. make sure that we get from the arrangement n John Griebling & Jim Irvine's perspective and ours. I'll fire off additional information as I get them. Ravi. ----- Forwarded by Ravi Thuraisingham/Enron Communications on 02/23/00 10:51 AM ----- nevil@ipn.caida.org 02/22/00 12:16 PM To: members@caida.org cc: nevil@caida.org, (bcc: Ravi Thuraisingham/Enron Communications) Subject: CAIDA 'Metrics' WG meeting, 2 Mar 00 Hello CAIDA Members: Update on the CAIDA Working Groups .. A. 'metrics@caida.org' mailing list B. WG Charters, Meeting on 2 Mar 00 A. 'metrics@caida.oeg' mailing list I've set up a single mailing list with this name, for discussions on WG topics, passive measurements, etc. To start with it's a moderated list (i.e. you have to be a member of the list to post to it, you join by sending email to nevil@caida.org asking to be added to the 'metrics' list), with the following initial set of members: Sue Moon <sbmoon@sprintlabs.com<, Brett Watson <bwatson@vix.com<, Hans-Werner Braun <hwb@nlanr.net<, Matt Mathis <mathis@psc.edu<, Ian Graham <ian@cs.waikato.ac.nz<, Tony McGregor <tonym@nlanr.net<, John Cleary <jcleary@cs.waikato.ac.nz<, Joerg Micheel <joerg@cs.waikato.ac.nz<, Kevin Thompson <kthomp@mci.net<, Jambi Gambar <jambi@mci.net<, Daniel McRobb <dwm@caida.org<, David Moore <dmoore@caida.org<, Sean McCreary <mccreary@caida.oe,rg< Rene Hatem <rene@canarie.ca<, Shankar Rao <srao@qwest.net<, Cindy Bickerstaff <cbickers@td2cad.intel.com<, Jeff Sedayao <sedayao@orpheus.sc.intel.com<, Steve Feldman <feldman@twincreeks.net<, Bill Woodcock <woody@zocalo.net< Two questions for CAIDA Members: i. Who else would you suggest be invited to join the list? ii. Should the list continue to be moderated, or should it be changed into an open list? B. 'Working Group' developments Following the CAIDA Members' meeting on 8 Feb 00 I've attempted to define exactly what problem we could consider getting an IETF Working Group started on. My summary of the existing IETF WGs with interests in metrics is given below (Appendix B), but it seems unlikely that we could get a new IETF WG started. I believe that we should instead run a single CAIDA Working Group on 'Network Metrics,' rather than the two proposed earlier. My draft of its charter is appended below. It focuses on producing educational material about network measurement, and on developing new metrics - these were the two areas of greatest interest amongst the CAIDA members. The WG co-chairs are Sue Moon (SprintLabs) and Brett Watson (MFN/Abovenet) You are invited to attend the first WG meeting. The agenda is as follows .. Agenda for CAIDA WG meeting on: Thursday 2 Mar 00 ----------------- 10 am - 4 pm, AboveNet, Downtown SJC (see below for details) ------------ ------------ 1. Review WG Charter - Is it reasonable as set out in the draft? - What should be removed or added? 2. Work through revised charter in detail - Identify the work required for each part - Determine who's willing to work on it - Attempt to determine delivery times 3. Discussion of new metrics - First attempt at making a list of metrics to be considered 4. Anything else ? Location: AboveNet is located in the Knight-Ridder Building, attached to the Fairmont Hotel complex. The address is 50 W. San Fernando St. San Jose, CA 95113 RSVP: To help us with organising the meeting, please send email to nevil@caida.org telling us how many will attend from your organisation. Cheers, Nevil ------------------------------------------------------------- Nevil Brownlee Visiting Researcher Phone: (619) 822 0893 CAIDA, San Diego CAIDA Network Metrics Working Group: Draft Charter, Tue 23 Feb 00 Goals: 1 Education + FAQ on What does 'measuring the Internet actually mean?' - Why measure anyway? - What can be measured? How? Where? By whom? - Active vs passive, end-to-end vs provider network only, application vs transport layer - Rating schemes: provider 'net performance' pages, Internet 'Weather Map's, Keynote, etc. Publish as CAIDA web pages, or maybe as an Info RFC + Survey paper on metrics and Internet Measurement - Current measurement efforts (Surveyor, RIPE Test Traffic, AMP, IPERF, AT&T, Keynote,skitter, ...) - Current tools Publish as CAIDA web pages 2 Service Metrics + Define new metrics - Taxonomy of current metrics (IPPM, RTFM, ITU, ..) - Summary of metrics used for current services - Gather information/ideas about new/emerging services, especially DiffServ-based ones - Make list of new metrics, either to improve measurement of existing services or to support new ones [list of 'metrics' questions (Appendix A) goes here] + Organise experimental implementation/testing of tools for new metrics + Make recommendations on implementation - Define core set of 'really useful' metrics - recommend that CAIDA implement these as a 'Service Measurement Toolkit' + Publish new metric definitions through IPPM or RTFM + Produce document "measurement requirements for hardware/software vendors." Publish on CAIDA web pages ------------------------------------------------------------------------- Appendix A: Questions from the earlier draft CAIDA WG charters a. What types of network- and transport-layer metrics are being used by ISPs in engineering and operating their networks? By Customers for verifying service guarantees? b. What new services are being (or are likely to be) offered, e.g. DIFFSERV? Is there a need for higher-layer metrics to better monitor and manage these services? c. Will these new differentiated transport- and application-layer services need new metrics? d. How can the service metrics be measured in a multi-ISP environment? e. How can customers verify these measurements? f. What requirements would service measurement introduce for equipment vendors? g. How relevant are specific techniques (e.g. which flow) and points of measurement to specific users (ISP, customer, etc.) requirements? h. How do these metrics relate to network behavior as perceived by users? How do they correlate with performance? ------------------------------------------------------------------------- Appendix B: Background on the IETF Working Groups * RTFM WG: Realtime traffic Flow Measurement RTFM is concerned with passive measurements of two-way traffic flows, specified in terms of their end-point attributes. Its primary goal was To produce an improved Traffic Flow Measurement Model considering at least the following needs: a. Wider range of measurable quantities, e.g. those relating to IPv6, and to class of service b. Simpler ways to specify flows of interest c. Better ways to control access to measured flow data d. Strong focus on data reduction capabilities e. Efficient hardware implementation * IPPM WG: IP Performance Measurement The IPPM WG charter is to develop a set of standard metrics that can be applied to the quality, performance, and reliability of Internet data delivery services. These metrics will be designed such that they can be performed by network operators, end users, or independent testing groups. It is important that the metrics not represent a value judgement (i.e. define "good" and "bad"), but rather provide unbiased quantitative measures of performance. RFCs Framework for IP Performance Metrics (RFC 2330) Metrics: Connectivity (RFC 2678), One-way Delay (RFC 2679), One-way Packet Loss (RFC 2680) Round-trip Delay (RFC 2681) I-Ds Bulk Transfer Capacity (2x) Instantaneous Packet Delay Variation One-way Loss Patterns * Other WGs The RMONMIB WG is thinking about 'Application Performance Measurement.' This is clearly a hard problem (e.g. does this just mean response-time measurement, can it be done by passive means, how should the measurements be presented, etc.). In short - RTFM provides a good distributed measuring system for traffic volumes - IPPM has concentrated on transport-layer behaviour of the current, best-effort Internet. - RMONMIB is beginning to consider application-layer measurement -------------------------------------------------------------------------
|