I am running Linux ES3 on an IBM Bladecenter, which is connected into a Clariion CX500 via 2 Fabric switches (4 paths). Linux is running a large progress database with about 75-100 users connecteda t any given time. From time to time we see an enormous amount of IOwait on Linux, and the DBA's keep pointing back to the disk's being too slow. Can anyone give me a direction on how I could find out exactly which process is causing the High IOWait? I've user SAR and iostat, but nothing really details the IOWait portion. My ultimate goal is to show that the Disks aren't the bottleneck. I feel this all spawns from poor performance inside the database itself, but I can't find a way to strongly prove it.
