Apologies for the delay, I had to consult with my Db2 colleagues to get a response for you.
As you know, when the DPRSMF job runs, information from the last SMF statistics record encountered for each DB2 subsystem is written to the spin file and then used as the first record for the next DPRSMF run.
This implies that for each job that extracts DB2 statistic SMF records for some DB2s, the SPIN file should re-used for the next run. If the customer splits the extraction of DB2 statistic records into several jobs each of these jobs should use its own SPIN file. When re-starting one of these extraction jobs it needs the last records from the previous run to calculate the statistic interval deltas – and these ‘previous run last records’ are saved intermediately in the SPIN file.
Hope that helps.