Date: Fri, 2 Nov 2007 12:03:04 EDT
Reply-To: CICS List <CICS-L@LISTSERV.UGA.EDU>
Sender: CICS List <CICS-L@LISTSERV.UGA.EDU>
From: Robert Zenuk <Robzenuk@AOL.COM>
Subject: Re: problem investigation - historical information
Omegamon also has historical reporting, but it has to be set up and is
usually implemented as a wrap around file that limits the amount of history
available. If you end up using it, the Omegamon/Classic History is much more
detailed than the CUA history. Unfortunately, there is not a batch reporting
component, only online. This is basically SMF 110 transaction detail.
The CPSM HTASK history table will work well in small shops. In large shops
(hundreds of regions), I would think you would want to be EXTREMELY careful
about HTASK queries. After seeing the impact of CPSM queries against large
resource tables across a large set of regions (1M+ rows returned), you can have
Paging/AUX storage problems in a storage constrained environment. Our
production environment (300+ regions) has lugubrious amounts of storage and can sus
tain a large query, but Omegamon still squawks from all the extra
ICE's/AID's generated to satisfy the queries (can safely be ignored for the CMAS's).
In our development environment (2-way Sysplex and 100+ regions), we are not so
lucky and the CPSM QUExxxx dataspace will consume all available real
storage. Fortunately, the CPSM dataspaces will grow and expand to meet the
requirements, but unfortuately will not shrink when the need subsides. The HTASK
table is a clone of the TASK table and is basically SMF 110 detail.
TMON seems to have the nicest history tool I have had the opporuntity to
work with. The batch reporting is also very nice. This also looks like SMF 110
transaction detail. If I remember right TMON also had some other summary
history data available by hour and day.
I have not had any "hands on" with Mainview/CICS history or CICS/PA, so
can't really comment there.
If it is transaction detail history you are using to do the problem
diagnosis with, your SMF 110's can still be used (SAS, MICS, MXG or RYO).
Wish list. It would be nice if the stats interval could trigger the capture
and recording of interval data equivalent to DFHSTUP reporting (both basic
reports and the sample DFH0STXR Extract program's events). Being able to
reference this "after the fact" during a research session would provide
visibility into other conditions not apparent from the transaction detail. This would
be similar in concept to RMF Monitor III data for reviewing delay events.
Then again, if you have this set up, you can manually run DFHSTUP for the same
period. I guess I'm just lazy and want it done for me...
In a message dated 11/2/2007 7:29:22 AM US Mountain Standard Time,
I’d be surprised if Omegamon doesn’t have historical reporting. I know
Tmon does, as well as CPSM. My personal favorite is MAINVIEW for CICS from BMC
which has tons of information in history and is used regularly by our
customers for just this thing.
Regards, John Barnard
Dev director for MAINVIEW
From: CICS List [mailto:CICS-L@LISTSERV.UGA.EDU] On Behalf Of john clean
Sent: Friday, November 02, 2007 2:07 AM
Subject: problem investigation - historical information
Sometimes there is a problem occured on cics, and then it is solved some
time later automatically. Is there any method or tools for investigating what
happened at the problem time. What is the causer of the problem. I have
omegamon monitoring product. But it seems it is online monitoring tool. After the
problem is solved there is nothing found in the omegamon about the problem. I
think Performance Analyzer product may helpful in this case. Is there anyone
having experience about this subject?
************************************** See what's new at http://www.aol.com