HotSpots and ColdSpots

A HotSpot is triggered when the GiAPA performance data collection (job GIAPAPFCOL) finds a job that within an interval exceeds the CPU-% limit defined by the user when starting the data collection. Job GIAPAPFCOL passes the ID of the job exceeding the usage limits to job GIAPAHOTSP resulting in collection of HotSpot data which consists of call stack and file usage information for the job.

HotSpots are samples that only are captured whenever the CPU usage limit is exceeded. For jobs using that much resources repeatedly and therefore causing many HotSpots, this sampling gets very detailed, and will for any “heavy” job normally be a very reliable way of obtaining rather exact information about the issue at hand.

ColdSpots collects the same data as HotSpots, but are triggered by interactive jobs that use minimal CPU without terminating a transaction within two consecutive collection intervals, typically seen in case of e.g. record lock conflicts. The automatic collection of the job call stack ensures that the program and statement where the job was waiting can be seen indicating the culprit of the delay.