Status Report for SLAC/DOE IEPM Project
For March & April 2000
Work completed or in progress since the last report.
PingERWarren installed a new version of timeping at SLAC and ANL. Further testing will be done with installation at HEPNRC, BNL, and Stanford. This version implements the new send=n and size=m directives. We also modified the SLAC site list to take advantage of these directives to reduce the load on sites in Pakistan and Argentina.
Warren worked with HEPNRC to fix a graphing program.
Warren and Les worked with security to resolve some of their issues with traceroute.pl (the traceroute server code). Les has incorporated the NIKHEF traceroute so it can report the AS information. It also now reports the TTL (to assist in seeing asymmetric routes). For added security the new version of traceroute.pl limits the number of simultaneous traceroute processes running (to help avoid denial of service attacks) and sends email if the threshold is reached and email has not been sent in the last hour. It also hides internal hops if the request comes from offsite (to help guard against scanning the site router topology). Warren installed a whois server at SLAC to avoid excessive use (to find the AS) of the Merit/RA server. The traceroute server is now receiving over 1000 hits per day.
Warren began trialing application testing with CNRI and HP. Connie Logg is working on porting her LAN health watch to the WAN.
Warren worked on developing a Jitter Metric.
Warren received a complaint from a network in Pakistan regarding ping traffic. After explaining the project the Pakistan networkers have agreed to help with the project. Les is now working with them and with Stefan Savage (author of sting) and a Sun engineer, to try and understand the duplicate ACKs returned by the site.
Warren worked on cleaning up hung processes due to the feature in ping with Red Hat 6.1 such that the process remains until all pings have been answered. In the case of unreachable nodes this means the process waits until the machine crashes. The recommended solution is to use another implementation of ping, such as the NIKHEF version.
In response to some questions raised on the SCIC mailing list Warren put together a web page showing the correlation between PingER predictions and NetPerf TCP throughput measurements.
Les attended the Passive and Active Measurements workshop in Hamilton New Zealand and made a presentation on Internet Performance for the HENP Community.
Les provided Jim Leighton of ESnet with slides for an in-house Director-level ESnet Program Review next week.
IPv6Les presented a talk on IPv6 monitoring at the IETF inAdelaide. Warren was contacted by one of the IPv6 developers at Sun about using PingER-6 to evaluate the IPv6 implementation of Solaris.
Warren was provided with an account on an IPv6 machine at 6TAP.
Warren installed a machine with the FreeBSD + KAME implementation of IPv6.
Warren worked on porting IPtables, the successor to IPchains as the linux firewalling system and looked at IPfilter, the FreeBSD firewalling system.
Les, Warren, Bob Fink (of LBNL) and Richard Carlson met at the ESCC meeting and discussed possible interest in putting forward a proposal to fund IPv6 work.
VoIPWarren discussed QoS experiments with JAnet/UKERNA. Warren and Les met with ESnet engineers to propose extending the VoIP testbed. The ESnet folks are doubtful that we can do much at the moment.
New sitesWarren set up pingER at JLAB. Data is not yet available for analysis because the web server is not set up to serve the data.
Warren was contacted by CSC in Finland about their plans to set up PingER.
Missing DataMinor glitch with Riken and Stanford.
Rate LimitingMit continued to search for candidates for rate limiting in the PingER remote sites. Comparison between PingER packet loss and packet loss reported by a tool using TCP revealed some strong candidates. Further research is underway.
Back to Top
XIWT/IPERFWarren and Les discussed the CNRI/Darpa IPEX grant with Charles Brownstein from CNRI. Les also followed up on some paperwork required at SLAC.
IEEEWarren and Les proof read the pre-print for the paper for publication in IEEE magazine.
ESCCWarren and Les attended the ESCC meeting at Pleasanton California. Warren presented a review of IPv6 deployment at SLAC and Les presented an Update on the Internet Monitoring.
I2ICharles Granieri presented a paper to the I2 International group.
Warren has been invited to give a review of the PingER monitoring at the NLANR/I2 Joint Techs meeting in Minnesota.
SC2000Warren attended a meeting to co-ordinate with other members of SLAC to prepare a display for the Super Computing 2000 conference to be held in Dallas later this year.
Students and New Hires2 RA's finished. Another one left unexpectedly, and a fourth hasn't been in for sometime and his RA-ship will be terminated. Mit is the sole RA remaining. There have been two applicants for a summer student post, three enquiries for a RA one response to search for a full time position.
Revised 29 April, 2000.
Comments to firstname.lastname@example.org