Jump to Content
  Previous Next
Content starts here

Methods and Call Trace Information

Methods where JRockit spends most of its time are called hot. Once you have identified such a method, you might want to investigate it to see if it is a "bottleneck" for the application or not. The way that BEA JRockit collects method information is via a sampling thread that is called the hotspot detector. It uses statistical sampling to find Java methods that are candidates for optimization. The samples are collected by iterating through the Java threads in the virtual machine and suspending them one at a time. The current instruction pointer of the suspended thread is used to lookup in which Java method the thread is currently executing. The invocation count of the method is incremented and the method is added to a queue of methods to be optimized if the invocation count exceeds a certain threshold.

The JRA recording system makes use of the hotspot detector by setting it to a high sampling frequency during the recording and directing the samples to the .jra file.

This section is divided into the following topics:

  Previous Next