> What else would a network/linux admin look at to help the testers and > developers locate performance bottlenecks? I'd like to see trending (cacti?) for: error codes (4xx and 5xx) bandwidth in/out service times worker/thread usage TCP states, maybe