Chris,

I'll assume that by "stats job" you mean an execution of the DB2 RunStats 
utility, yes?  If so, did you specify SHRLEVEL CHANGE?  Also please 
clarify:  are these two independent DB2 subsystems or are they part of a 
DB2 data sharing group?

Assuming the subsystems are part of a data sharing group, then I'd suggest 
looking at the appropriate SMF information to determine how the 
long-running job spent its elapsed time. First check the virtual pool 
settings (buffer pool size, threshold settings, data space, etc.) to see 
if there are any differences between your two systems. Also look for 
activities on the systems that would generate Inter-DB2 Read/Write 
Interest.  Example: As your RunStats job is executing on System A, other 
jobs/transactions on System B are also accessing the same objects, leading 
to lots of global buffer pool use.

One possibility that comes to mind is that the hit ratios in the 
bufferpools across the two subsystems are quite different, possibly 
depending upon what other pagesets are being allocated into them during 
your RunStats executions.

Also, you might consider cross-posting this to the DB2-L ListServ.  Lots 
of good help over there as well.

Lock Lyon
Compuware Corp





Chris Sultis <[EMAIL PROTECTED]> 
Sent by: IBM Mainframe Discussion List <[email protected]>
09/20/2006 03:50 PM
Please respond to
IBM Mainframe Discussion List <[email protected]>


To
[email protected]
cc

Subject
Q: DB2 Help






We have DB2 V8 in compatibility mode on 2 separate LPARS on 2 separate
Z9 boxes. We can run a stats job on the same tables, same DASD, same XCF
with structures and ECS catalogs but one job finishes in 1/3 the time. 

No, the boxes are not equal and no the workload is not equal either but
our bigger box is the one running slower. This is affecting DB2 batch
JCL only. Non DB2 batch is working as designed.

It just started Monday PM but no maintenance was performed this
weekend. We feel like we have inadvertently flipped something on or off
but what that is... we are at a loss. I have poured over channels, DASD
and CPU RMF reports. I have reset cleared and deactivated CPC images.

Any suggestions as to were to look next?

Thanks
Chris M. Sultis




----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO
Search the archives at http://bama.ua.edu/archives/ibm-main.html

Reply via email to