Status Report for SLAC/DOE IEPM Project
For July & August, 1999
Warren has been working with the RIPE and SLAC Ops to complete setting up their test-traffic box at SLAC. There has been a number of technical issues, most recently the cable to connect the machine to the GPS antenna on the roof was found to be broken. Warren processed the data from the NLANR Active Measurement Project (AMP) machine recently installed at SLAC, and Rebecca looked for correlations to the performance measured by PingER. The correlation was quite poor. We believe the different time sampling is the reason, and we will conduct further studies. Warren and Les had a voice meeting with HEPNRC to co-ordinate development. Les and Rebecca have compiled a large list of sites and their group affiliation. Les has worked on revising the PingER Beacon list. The new list is available. About half the monitoring sites have implemented the changes. Les talked to Walter Toki about ESnet University working group needs and communicated with Larry Price (chair of ESSC) on giving a talk.
Warren and Yvonne have been continuing to work on database connectivity. Yvonne is working on a GUI for the JDBC/Guthrie program. Warren set up a Postgresql server off-site and succesfully tested making a remote connection using just the postgresql jdbc driver at SLAC. Further development was to involve connecting directly to the PingER archive held on a SAS database at HEPNRC, but SAS will not sell the JDBC driver separately and will only licence it to us as part of a package. Warren and one of the new students starting in October will look at alternatives. Warren has been looking at the CAIDA CoralReef package and the OC3MON hardware to characterise the SLAC traffic and provide some passive measurements to correlate with the active measurements. Present SLAC LAN tools cannot provide this information. Boris made a study of traffic sniffed with TCPDUMP to do some initial test of Passive Measurements. These initial tests were quite succesful and Warren will work with Dave Millsom to set up sniffing on the SLAC DMZ. Kishan worked on using TTCP to generate traffic on the test VoIP link to LBNL, and looked at how ping round trip times varied with load and how setting the type of service (TOS) bits in the IP header (combined with WFQ code in the routers) could improve performance. The results showed little or no improvement setting the TOS. Warren is working with the network administrators to use SNMP to verify the load is being applied correctly. Warren amended the PingER reports to include a TCP thruput metric based on a theoretical formula linking the round trip time and the packet loss with the maximum bandwidth. Initial comparisons looked quite good, but presently it's not so good, perhaps there is a bug. New group for the NGI-testbed project. Les talked to Minh van Duong about how to apply chaos theory to Internet measurement, so far not very successful.
Warren wrote a summary of the pingER project for a forthcoming XIWT white paper. Rebecca continued to developed the project Web Pages. Les created some slides to be presented as part of an overview at the Lepton Photon 1999 conference. Les has prepared a talk on the IEPM work for the PPDG at the ESSC meeting next week. Warren has prepared some slides on the IEPM work on the VoIP link for Charley Granieri to present at the ESSC meeting next week.
Two summer students have left, but we have a new R.A. just started and 2 more starting in October.
Revised 23 August 1999
Comments to firstname.lastname@example.org