Status Report for SLAC/DOE IEPM Project
For February & March 2000
Work completed or in progress since the last report.
PingER DevelpomentWarren worked on a geographical map based on the maps from the CHEP poster which have attracted a lot of attention. Warren also worked on automating the process and making an animated gif using a program called gifsicle.
Following a request for further information from David Williams of CERN for one of the CERN director's Warren and Les created a map with Internet performance seen from CERN for February 2000. Followed up on questions from David Williams on the bad loss to Denmark, and the poor reachability to France.
Working with Michael Ernst & David Williams to understand DESY performance from N. America over last couple of years.
IPv6Warren talked with Mike Collins and Bob Fink about a tunnel to Abilene. Warren will make a further list of action points and look into IPv6 Aggregatable addressing.
Les communicated with someone in Poland about their IPv6 monitoring (see http://www.6bone.pl/ and http://6bone-gw.6bone.pl/mrtg/)
ComparisonsWarren obtained more RIPE-TT data and processed the Surveyor data for February. Les analyzed the data to quantify the agreement between RIPE and Surveyor data. Sent email to RIPE & Surveyor to try and understand subtle differences in their results. As a result chose a new set of RIPE/Surveyor data to analyze. The analysis revealed some interesting features of the agreement, in particular very good general agreement but little correlation at short (seconds and sub-second) time frames. See Comparison of Surveyor and Ripe Results for more details. We agreed to provide slides to RIPE for their talk at PAM-2000.
Les analyzed PingER and AMP data from December 1999 that Warren prepared to compare the results. This is to be presented at the PAM meeting in New Zealand next month.
Warren looked at the netperf graph showing throughput from CERN to Caltech and compared it with the throughput derived from the packet loss and RTT measured by PingER. There is quite good agreement. See Pinger vs netperf.
Warren talked with John Halperin about using his xchkaccess to make a WAN version of the new SLAC LAN health watch page (visible inside the SLAC firewall only).
VoIPWarren talked with Robin Tasker from the UK about some QoS and VoIP tests. Robin spoke to the UKERNA network operators and they will implement WFQ settings for the tests. Warren has arranged to meet with some ESnet people at the forthcoming ESCC to work out a plan. Arrange for meeting at ESCC
TracerouteGiven the increasing number of false positive alerts for port scans raised by traceroute servers, Les added new information on security aspects of Traceroute to the Traceroute Server page and to the application itself.
Following a request from the XIWT for more information on AS's involved in traceroutes, we discussed using the NIKHEF traceroute (that provides this information). The SLAC security folks want to review the code for possible buffer overflow security exposures before enabling this at SLAC. Les also communicated with John MacAllister of Oxford to ascertain progress with porting his traceping to Perl/Unix, and to get some old data relevant to DESY/N. America performance.
Apply an enhancement made by CNEA/Argentina to the SLAC traceroute.pl public domain traceroute server code.
New SitesThe University of Wisconsin has been added to the analysis. However There seems to be a problem with ping in Red Hat 6.1 such that the ping processes remains until all the packets are received, so in the event of packet loss the processes remain until they are manually cleared. Warren will work on clearing the processes from within the timeping program.
JLAB will install a machine and provide an account for Warren to install and administer data gathering. We are also working with SDSC and CalTech to get PingER installed there as part of the PPDG project.
We also worked with CNEA Argentina to assist them in bringing up a monitoring. Given their poor Internet connectivity it is doubtful that they will be able to monitor many sites.
Missing DataThe BNL site was mistakenly put behind the firewall for a few days, which prevent the analysis code picking up the data. However, no data was lost, and all the data was collected after the firewall was reconfigured. In all, This has been one of the best months for data collection.
HEPNRCWe had a voice conference with HEPNRC. We agreed upon a schedule for release of the new timeping.pl program and the new Beacons sites. We also agreed to move responsibility for timeping.pl to HEPNRC from SLAC.
CAIDA Metrics Work GroupLes and Warren joined the CAIDA Metrics Working Group (see the web-searchable archive). Warren attended a meeting of the CAIDA Metrics Working Group to represent the end-user (ie a site requiring high performance networking) and will contribute to some documentation. Such efforts are as they provide a way for HENP to communicate and work with router vendors and network providers to provide the low packet loss networks required for HENP.
Back to Top
XIWTSLAC hosted the 2 day XIWT/IPERF meeting. See Les' rough notes for more information.
IEEEWarren Submitted details for the IEEE paper to Com Soc.
Passive and Active MeasurementsLes prepared a talk for the PAM meeting in New Zealand in April.
IETFLes and Warren prepared a presentation on IPv6 monitoring for the IETF in Adelaide later this month.
Internet 2 InternationalWarren prepared a talk and an accompanying paper that Charley Granieri will present to the Internet2 International group later this month.
AdministrationLes prepared follow up paper work for a CRADA for the CNRI led DARPA sponsored measurement project.
Les submitted paperwork to DoE to travel to the IETF & PAM conferences. Started arranging travel and accomodation. Adelaide seemed pretty full, finally got a Hotel that DoE can afford.
Revised 22 March, 2000.
Comments to email@example.com