Understanding sysstat disk i/o data
I'm studying the scalability of an application. I've hit a bottleneck. The CPU utilization is no more than 65%. The load-average is well below the "no more than 3X the number of CPUs" rule-of-thumb. So some resource(s) other than computing horsepower must be the bottleneck. I suspect disk I/O.
However, I don't know if the disk I/O data reported by sysstat are "good" or "bad".
Where are some webpages that will yield sysstat disk I/O rules-of-thumb &/or provide some guidance to understanding "how big is 12"?
|