Status Report for SLAC/DOE IEPM Project
For December 1999 & January 2000
Work completed or in progress since the last report.
PingER DevelpomentWarren and Mit continued to work on the new metric to detect rate limiting.
Warren added the IP addresses to all the nodes monitored from SLAC to avoid losing data when DNS fails and worked with the pinger administrator in the UK to amend the version of timeping running from Daresbury Lab.
Warren and Faisal put the data for the IPv6 monitoring into the oracle database, and Faisal worked on porting the analysis code to make reports directly from the database.
SynackKishan worked on Web pages to document his synack program.
JAS (Java Analysis Studio)Kaushik continued devloping the histogramming program. The initial version will go public when the web server is upgraded. Kaushik is looking at data base access for the next version of the histogramming program. Warren worked with Kaushik to set up a test oracle account.
IPv6There is a problem with connectivity from the IPv6 monitoring machine Pontus. Warren has been working to reinstall the machine to get data taking back on line. Warren installed a new machine Scylla and the IPv6 address has been added to DNS. Warren has been working with Marc Blanchet to arrange to set up an IPv6 monitoring machine at the 6-Tap. Also a site in China has expressed interest in becoming an IPv6 monitoring host. Kishan is working on porting TTCP to IPv6. Warren has designed a system to deploy the IPv6 network further and is working on configuring a firewall. Warren is preparing a presentation on security for the SLAC IPv6 network.
Smart Bit BoxRajiv downloaded the smart libraries from netcom, but there are compilation errors that must be fixed.
VoIPRajiv looked at tcpdump and sniffit for the VoIP testbed.
Evaluating new tools
Trunks_mapWarren talked with Jambi Ganbar from vBNS (ex-CAIDA) and obtained his trunks_map program. Warren contacted Chin Guok of ESnet to find out about extending PingER to get data from ESnet routers.
Rate LimitingMit looked at synack vs ping and discovered a bimodal RTT for routes passing through gin-nyy-bbl.teleglobe.net Warren identified 25 other sites where the route from SLAC passes through this router. Warren looked at historic PingER data for evidence of this bimodal effect. Warren worked with Mit on some coding techniques.
Les and Mit put together a web page on ICMP Rate Limiting (see http://www-iepm.slac.stanford.edu/monitoring/limit/limiting.html).
New sitesWe have a two new monitoring sites, at the STAR-TAP and at the Budker Institute of Nuclear Physics. Les has been communicating with the BINP since the want to become a SPIRES mirror site. Warren found the data retrieval from STAR-TAP was flakey and worked with ESnet to fix the routing. Details of the new sites have been added to the participants page and the administrators have been added to the pinger-admin list.
Warren worked with the pinger administrator from KEK when they discovered their machine was not Y2K compliant and the data gathering was smoothly transitioned from yumj to yumj2. The change was also applied to participants page.
Warren followed up with the Australian test monitoring site to find out if they will install the full beacons list and become a fully fledged monitoring site.
Warren contacted Nikolai N. Kostyukovich for details on networking in Belarus.
Les is working with SDSC & UWisc to explore installing PingER for the PPDG project.
Missing Data and Pathologies
Missing DataNo major data loss, just minor problem with data from Stanford, DoE and CERN1.
Compare resultsWarren worked on comparing pingER results from the two monitoring sites at CERN. The RTT results differed by large amounts in some case. This is believed to be due to loading caused by running Netperf.
Warren worked with the RIPE Project co-ordinators to get raw data from their database to compare it with performance measured by surveyor. Warren retrieved and processed the Surveyor data from December.
Duplicate packetsWarren changed the University of Witwatersrand node from www.wits.ac.za (Southern Africa's SunSite) to concave.cs.wits.ac.za (a name server) because pinger was recording a large number of occurences of duplicate packets.
Bimodal RTT distributionsMit, Warren & Les are trying to understand bimodal RTT distributions to England. It appears to start at a router in the US.
Anomalous hostsLes revised the requirements for Beacon hosts to include a robust TCP stack (following problems with duplicate acks identified by sting to Pakistan and Macedonia), and to have high availability (following problems in Pakistan and Singapore). Cleaned up host lists by removing some sites which were not responding for a long time (a month). Communicating with Pakistan & Macedonia to try and understand double acking problem.
Back to Top
LogoTerry Anderson designed a new PingER logo for use in posters and presentations. We are so impressed we are having polo shirts made at our own expense.
IEEEOur paper for the IEEE Com Mag on Network traffic has been accepted, subject to a few minor alterations.
CHEP 2000/HENPWarren worked on the talk, the poster, and the written papers for the CHEP'00 conference.
Les was invited to present a plenary talk on the Status of U.S. Networks at CHEP00. Les communicated with Jim Leighton of ESnet and people from Internet 2 to get updates on their networks and drafted a talk.
Warren and Les have arranged to visit CNAF in Bologna following CHEP00 to discuss collaborating with monitoring projects.
Warren and Les created a map of world HENP collaborators. Also Les cretaed a map of PingER deployment for the CHEP00 poster (see PingER Deployment.
Les set up a web page to track Network Measuremets at PPDG sites. Also coordinated getting Surveyors at all PPDG sites and sent email to the Surveyor folks, the PPDG folks and the contacts.
Paper for PAMWarren worked on the paper for the PAM conference.
Seminars etc.Mit and Kishan presented talks about their work.
XIWTLes reviewed and commented on the XIWT/IPERF white paper on "Internet Service Performance: Data Analysis and Visualization". Les and Warren provided input for the XIWT/SLAC proposal to DARPA.
Warren attended the WIWT/IPERF conference call. SLAC will host the next face-to-face meeting in February. Warren is working to arrange teas/lunch, network connectivity and a detector tour.
NATOMore communications with a small startup interested in setting up satellite links to to Central and East Europe and FSU republics where we would provide them with measurement capabilities.
AdministrationWarren and Les worked on putting together a Field Work Proposal to be sent into DoE/MICS later this months, to provide a continuation to the IEPM program.
Revised 23 January, 2000.
Comments to email@example.com