Status Report for SLAC/DOE IEPM Project
For July & August 2000
Work completed or in progress since the last report.
JASThe JAS histogramming servlet was released into production (select a monitoring site from the first column in http://www-iepm.slac.stanford.edu/cgi-wrap/pingtable.pl. Further development is planned.
Passive MonitoringWarren worked with the OC3MON passive monitoring tool and enabled netflow on the SLAC border router. The data is written to a restricted area because of its potentially sensitive nature.
Warren installed the RRDTOOL package, and is modifying the database for jnflows to track the changes over a longer time period. Les used the JNflow netflow information to help with analyzing the impact of bulk data transfer on RTT. based on this several suggestions for improvements were made to the JNflow authors. Warren installed the RRDTOOL package, and is modifying the database for JNflows to track the changes over a longer time period.
Warren talked with the netflow people from campus about the analysis they do.
We talked with Bjorn Frogner of NetPredict and learnt about new dRMON2 from 3COM that provides enterprise MIBS for NIC cards with various statistics including RTT. Later we visited 3COM and learnt more about the tool. Currently it is Windows only and so does not easily fit in with our efforts.
Met with NCAR person & Richard Mount about alpha testing a new Linux instrumented TCP stack. This would enable finding about flows by interrogating a TCP/IP driver MIB. We hope to get a release of the code in September.
SACKEvagore was upgraded to Solaris 2.7. Warren has been reviewing documentation on performance tuning to make tests.
QoSWarren attended a Video Conference with UK to discuss their proposal for QoS testing with SLAC.
Rate LimitingLes did some more work on ICMP rate limiting in particular wrt Waikato U in NZ. Corresponded with Tony MacGregor of Waikato/NLANR to try and understand the results. Les also corresponded with SLAC security on using syn/ack to make measures of network performance and compare with ping.
SC2000Warren and Les will prepare information on the network performance monitoring to be included in the SLAC/Fermilab booth at Super Computing 2000. Les is also making bulk data thruput performance measurements to be shown at SC2000. These measurements include SLAC to the following sites: CERN, IN2P3, INFN, Caltech and Colorado. This required getting accounts at these sites and installing and running iperf. Also Les put togethera script to automate the measurements.
Missing DataThis month there have been problems with data collection at DESY, DoE/MICS and the University of Wisconsin. Some of it because machines were down. Warren has recovered some manually and worked with the local administrator to resolve other issues.
IPv6Warren exchanged emails with ORNL about IPv6 applications and testing.
TalksWarren wrote and presented a talk to the HEPCCC group about WAN performance.
Les attended the ESnet International meeting in Tokyo. He gave 2 talks, one an update on SLAC networking (see http://www.slac.stanford.edu/grp/scs/net/talk/slac-esi-jul00/), the other on Network Measurements (see http://www.slac.stanford.edu/grp/scs/net/talk/mon-esi-jul00/).
SupportWarren helped someone from Wayne University with the setup of the OC3MON optical splitter.
Warren replied to a query from IBM about out of order packets
Other organizationsLes discussed HEPNRC demise and future FNAL PingER efforts with Bill Lidinsky, Warren Matthews & Richard Mount.
Les reviewed IUPAP (International Union of Pure and Applied Physics) country Internet performance in response to a request from Burt Richter and later met and made some recommendations for Burt to pass onto IUPAP.
Les sent information to K. Claffy and friends on our comparisons between Surveyor, RIPE, AMP and PingER. Also sent information on the PingER project to people at Agilent.
Back to Top
Revised 25 July, 2000.
Comments to email@example.com