Wondering if anyone is doing anything with data from SMF type 92, subtypes 50-59...
These subtypes are generated if IOEFSPRM specifies "smf_recording=On,<interval>", but there does not appear to be a way to synchronise the recording interval with rest of SMF data. If you know of a way to do this synchronisation, I'd appreciate hearing how to do it. At present, I've got collection of these subtypes switched on with "smf_recording=On,15", and for comparison purposes, I've also got a timed JES command issuing 'F ZFS,QUERY,ALL' (immediately followed by a 'F ZFS,RESET,ALL') every 15 minutes, on the 15-minute mark. The timestamps of the two sets of data just never match up. Sure, long-term totals are in *close* agreement, but even then I'm seeing differences that make accurate performance analysis a bit questionable. Also, there are differences in other data items. For instance, 'F ZFS,QUERY,KNPFS' generates a table on the SYSLOG showing PFS call types along with counts and average time per call. In the equivalent SMF record 92(51) data, the average time values are always zero. As yet it's early days in my efforts to make use of this data, but I'm expecting to see plenty more difficulties as I delve deeper in to it. Regards Sean ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN
