Today has been a busy day... ChrisWelti has been working on an interesting SWITCH-internal performance problem, where the original symptom was slow access to Web pages on a server hosted by SWITCH from one given university campus. When we finally had two packet traces from client and server to compare (a rare luxury), Chris found out that there must be a middlebox, because the TCP sequence numbers change between the two hosts. So we are now concentrating our efforts on finding out more about this middlebox. We assume that the middlebox breaks the stream of HTTP requests/responses inside HTTP/1.0, and introduces blocking this way.

SimonLeinen was mostly busy with today's PERT progress videoconference and other planning work.

To make matters worse, a widespread power outage took out equipment in two major cities in Western Switzerland, and made us worry that our backbone might fall apart, which fortunately didn't happen.

-- SimonLeinen - 18 Jan 2005