You don't specify OS level so I'll assume V7R1. You can use the Job
Watcher portion of the performance tools to zoom in on the job specifics
and find the wait states etc. That is an extra cost item ( dirt cheap if
you ask me ) to the performance tools. One note, you can collect the
data on one machine that does not have the tools and analyze he data on
a machine that does have them.
Depending on the ratio of faults to pages, you may not have to worry
about them. Use a 5 - 8 minute interval and have a look, of the ratio is
high, meaning 100 pages to 5 faults, no worries. If it's 50 faults for
100 pages then the pool needs more memory. You might also consider
looking at the QZDASOSINIT and QSQSRVR jobs. Those are most likely
what's driving the Database activity in the *BASE pool. Move those jobs
to their own memory and manage them there.
There have been many threads on this forum in the last year discussing
how to manage those two jobs. You can use the search of the archives for
more detailed discussion points.
Chief Technical Architect
Agile Technology Architects
On 10/24/13 11:18 AM, Lance Gillespie wrote:
Is there a way to view faults by job?
That is, I see from WrkSysSts that there are
a bunch of DB faults in the *BASE pool.
I would like to see which jobs are responsible
for the faults with an eye to possibly changing
the code if the reason is bad design.
I can see which jobs are using all the CPU with
WrkActJob and sorting on the CPU % column,
but I have never found an equivalent for faults.