Status Report for SLAC/DOE IEPM Project
For March 2002
Work completed or in progress since the last report.
PingERWarren debugged the way the analysis deals with corrupted entries resulting in nonsense nodes.
BandwidthLes, Connie and Manish worked to try and understand the passive vs. active bandwidth measurements, see Correlation Table for Active and Passive Monitoring.
INCITESLAC hosted a visting PhD student from the Rice/INCITE project and worked with her to calrify the needs for tomography and topology. Warren worked with Yolanda Tsang from Rice to understand IP tunneling for the INCITE project. We had a phone meeting with the INCITE folks.
IPEXMore software installation with Allen from Telcordia. Warren & Les had a phone meeting with Telcordia folks to understand the privacy and security measures built into the Telcordia probe that we wish to deploy on the visitors subnet. We followed up with several emails and then wrote a request to the SLAC security folks and management formally requesting to install the probe and start measurements.
GlobusWarren is working with Adil Hasan and our collaborators to enable DOE SG certificates. BNL and ANL are intending to use them, but have not declared when.
VoIPWarren talked with the administrator at DESY on their plans for VoIP. Les helped someone from NetPredict with understanding the effect of delay on Mean opportunity Scores as a function of compression type.
Web100Warren re-installed web100 on antonia & hercules and explained to Les how to get at the data. Les added Web100 variable recording to his TCP windows and streams measurement script.
TracerouteLes worked with John Halperin to write a doument on what to do when traceroute intermittently fails to resolve a name.
Les followed up on 4 false port scan reports. He then modified traceroute.pl to try and ensure novices read the warning about traceroute looking like a port scan, see http://www.slac.stanford.edu/cgi-bin/nph-traceroute.pl He also blocked access to the traceroute server from 4 hosts which were sending many requests in a burst each hour, which causes the traceroute queue to overflow. Had previously tried to contact the security, postmaster, abuse, root addresses on these hosts.
Tool evaluationLes updated the network management tools page.
Les and Connie did a rough analysis of the effectiveness of pipechar compared to iperf, see Pipechar predictions of bandwidth.
Jiri and Les worked on evaluating a new network estimation program from LBNL, called netest.
Trouble shootingLes worked with IN2P3, ESnet and others to resolve some problems with the IN2P3 link via Renater. The link has now been upgraded to 100Mbps but the ESnet connection is limited to 30Mbps, see case study.
Les worked on iperf throughput from SLAC to DL and RAL and improved the latter quite dramatically by increasing window sizes at RAL, see case study.
Les responded to a report of poor performance between SLAC and INFN/Rome, see the case study
Missing DataStill no data from BNL nor INFN. A little trouble this month with gathering data from NBI and Wisc due to technical issues at those sites.
MiscThe usual paperwork, installations and administration.
IEEELes, Warren, & Connie submitted a paper to IEEE on the performance measurement results.
IPEXWarren created a few slides on the IPEX project for a DoE visit.
TrainingWarren attended a Project Management course and 'Building Cisco Remote Access Networks (BCRAN)'.
Travel & TalksLes visited Daresbury and CERN, see trip report trip report. He gave a talk at the CERN ICFA/SCIC meeting.
Les prepared and gave an invited talk on IEPM-BW at the Institute for Pure & Applied mathematics at UCLA.
DoEThomas Ndousse visited SLAC so we spent most of the day with him with presentations, and discussions including Connie, Jiri, Yo Yo and Richard.
MiscWe are working with Caltech to gear up for attempt on the Internet land speed record.
Created March 24, 2002, last update April 4 2002.
Comments to firstname.lastname@example.org