Users are complaining that their application gets frequent time outs. I tracked the database using perfmon for approx 2 hours (60 second increments) during peak activity and noticed that the %processor time showed steep peaks and valleys instead of consistent usage (of 118 samples 91 where either above 90% or below 10% for an overall average of 33%) and the processor queue length averaged over 5 (single CPU machine). My initial thought is a single (or few) nasty queries/stored procedures are periodically hogging the CPU, creating the timeouts, and driving up the queue length. I'm collecting transaction info. using Profiler now, but wanted to get some other opinions.