CA spill files

CA spill files buffer the log input records when there are more CA groups to process than there are parallel address spaces that can be started.

The number of spill files generally is the total number of CA groups to be processed minus the number of parallel sorts assigned. Sort address spaces start when the first processed log is opened and end with the closing of the final input log.

If the distribution of log input data sets is not common to all the groups being processed, the number of spill files might be reduced. This happens when two or more groups process disconnected sets of logs. Because these groups do not share log inputs, they are not scheduled for processing in parallel, thus making the spill files unnecessary.