SLAC logo

Routing Change Test

Warren Matthews   Page created: May 31, 2002, last update May 31, 2002.
Central Computer Access | Computer Networking | Network Group | ICFA-NTF Monitoring

Problem description

Teleglobe is a major network provider in major trouble. The UK particle physics community decided to test a change of routing between the US and the UK in case the existing connectivty via Telehouse in New York was lost. More information is also available from Ukerna.

Initial Response

At around midnight pacific time on May 30, 2002, the route from SLAC to UCL changed (changed hops are highlighted in red).

Before:

RTR-CORE1A.SLAC.Stanford.EDU
RTR-DMZ1-GER.SLAC.Stanford.EDU
192.68.191.146
snv-pos-slac.es.net
chi-s-snv.es.net
nyc-s-chi.es.net
193.62.157.213
146.97.37.89
po15-0.lond-scr.ja.net

146.97.35.2
146.97.40.34
ucl.lmn.net.uk
128.40.20.189
128.40.20.150
pc35.hep.ucl.ac.uk

After:

RTR-CORE1A.SLAC.Stanford.EDU
RTR-DMZ1-GER.SLAC.Stanford.EDU
192.68.191.146
snv-pos-slac.es.net
chi-s-snv.es.net
nyc-s-chi.es.net
62.40.126.5
62.40.126.14
janet-gw.uk1.uk.geant.net
146.97.37.81
po6-0.lond-scr.ja.net

146.97.35.2
146.97.40.34
ucl.lmn.net.uk
128.40.20.189
128.40.20.150
pc35.hep.ucl.ac.uk

Initial inspection suggest the route before and after the change-over was very stable.

PingER results show a clear decrease in minimum, average and maximum RTT. These are quick and dirty graphs that I will tidy up in due course (click to see full size).

   

Note the long-term stable minimum of 150 ms jumped up to around 158 ms at about 7.30am (Pacific) on May 29 before dropping to around 140 ms at the time of the route change-over. The reverse step was seen after the change back.

Further measurements