Status Report for SLAC/DOE IEPM Project
For January & February 2000
Work completed or in progress since the last report.
PingER DevelpomentWarren and Brian Tierney of LBNL had discussions about using an XML format for the raw pinger data. Warren previously incorporated the ULM format in response to requests but it has seen litlle if any use. Faisal has ported his pinger analysis code to use the raw IPv6 data from the SQL database. In order to see if performance could be improved Warren talked to Paul Raines about performance issues.
SynackKishan has been looking at pthreads and modified the SYNACK code such that the SYN segments are sent at regular 1 second intervals (user defined) irrespective of the status of the previous connections, as opposed to sending the SYN after the previous connection is made/timed out. He started looking at sending pings at sub-second intervals partially so a Poisson distribution can be used. He made a start on a web page documenting the tool.
JAS (Java Analysis Studio)Kaushik is learning how to use SQL and is extending the histogramming code to use Java database connectivity and inserted data into an Oracle test database. At this point the JAS program is querying the database but the data has still to be "prepared" for display with the histogramming routines.
IPv6Warren has been confguring a Linux system to provide an IPv6 firewall (since Cisco routers do not provide sufficient ACL capability). Warren also discussed with Davide Salomoni and John Halperin how one might convert IPv4 ACLs to IPv6. Warren prepared and presented a talk on how to use the Linux system to secure IPv6 networks until ACLs in IPv6 capable routers can do a better job.
VoIPWarren talked to Joe Burrescia of ESnet about extending the VoIP testbed to Italy or the UK. Joe arranged to meet some of the Italians at the QoS conference in Houston. Les, Warren and Davide had some inconclusive discussions with people from CNAF/INFN/Italy at CHEP00, we agreed to have a video conference in March. Les & Warren had more fruitful discussions with Paul Jeffries of the Particle Physics Data Group/RAL/UK about QoS testing.
Comparisons of ProjectsLes and Warren extracted the Surveyor and RIPE data for January 2000 and put together a Comparison of Surveyor and RIPE results.
Evaluating new tools
RootWarren explored using the ROOT histogramming/analysis package from HEP/CERN and talked with a local expert. He created a macro for analysis of the RIPE RTT (RIPE uses ROOT for its analysis).
CichlidRajiv worked with the Cichlid data visualization software from CAIDA, and has successfully programmed a server to represent the table.pl reports. The next step is to look at an easy way to represent the graphs as animated gifs.
Rate LimitingThe pinger admin at Daresbury in the UK modified the timeping program at Warren's request to record details of all 10 pings, not just the min/avg/max. This data will be used to search for a correlation in loss versus ping sequence number (icmp_seq).
Warren & Mit resolved the discrepancies in their methods for determining packet loss as a function of icmp_seq.
New sitesData from the ANL monitoring site came online this month. Also pinger has been set up at the University of Wisconsin (a PPDG site) and data is being gatherd, but is not yet available for analysis. We now have 30 monitoring sites.
Les communicated with a site in Uganda and we have started monitoring it. Les put together a new draft of the Beacons list which has been extended from 56 to about 75 sites. It has been distributed for review to the pinger administrators. It is planned to install the new list at all monitor sites in March after the review is completed. It is available at http://www-iepm.slac.stanford.edu/pinger/beacons-mar00.txt.
Missing Data and Pathologies
Missing DataData was lost from Brazil & ITEP. The site from srael experienced problems with zombie processes and so shut down pinger for a few days for trouble shooting. There was a problem with data retrieval from NBI caused by an extra space in the raw data.
Reverse Traceroute ServersThe reverses traceroute server at BNL has been down for sometime, but was brought back online this month.
Back to Top
CHEPWarren and Les attended Computing in High Energy Physics (CHEP) 2000 in Padova, Italy and presented a poster, a parallel and plenary session talks (see Status Report on U.S. Networks at the turn of the Century.
As part of the preparation for this Les compared the performance between ESnet sites, .Edu sites, and XIWT sites (mainly .com). The results can be seen at Comparison of Performance between ESnet, .Edu, and XIWT sites. These were shared with ESnet people and there was an email discussion to understand the results.
Also a part of the preparation, Les prepared Internet NREN performance from U.S. to world at the turn of Century.
XIWT/IPERFSLAC hosted the Cross Industry Working Team's (XIWT) Internet Performance (IPERF) Working Group. There were about 17 attendees. Warren made a couple of presentation on PingER Futures and IPv6 monitoring, Mit made a presentation on detecting ICMP rate limiting, and Les made a presentation on Internet End-to-end Performance Monitoring for HENP.
IEEEWarren and Les revised and submitted the IEEE paper for publication later this year.
Paper for PAMWarren worked on the paper for the PAM conference.
Seminars etc.Rajiv talked about his work with Samrt Bit, SOAP and RRD.
AdministrationThe IEPM Field Work Proposal was pushed through SLAC business services etc. and sent into DoE.
Revised 27 February, 2000.
Comments to email@example.com