Have I counted myths too fast?

I am doing some local experiments on different database systems. I collect (summarize) processor information from /proc/statusbefore and after the request. The difference should tell me the number of jiffies or USER_HZ during the execution of the request. But the difference is zero when (according to clock_gettime()) the request has a runtime somewhere below 0.001 seconds. Is it fast to use processor information, or am I missing something else?