> -----Original Message-----
> From: IBM Mainframe Discussion List On Behalf Of Ted MacNEIL
> 
> >Given that performance is erratic, but cpu usage is
> consistently less than 5%, then it would seem it's more a 
> question of I/O waits, not I/O processing.  I would look at 
> resource contention, e.g. channel, device, etc.
> 
> There is no contention unless you consider 2000 I/O's 
> contentious. OMEGAMON shows this as the main bottleneck. Seek 
> analysis drops it right inside this dataset. It is doing 
> un-indexed reads (BROWSEs), even though the file is indexed.
> 
> I checked all those things (I've known how to do so for awhile).
> CICS is going in spurts because it's waiting for the I/O to 
> complete, then it starts up again, then it waits.

That's a pretty good definition of I/O contention....

> The problem is there is too much I/O for a single CICS region 
> to handle on a sustained basis.

No, the problem is that your CICS region is not properly tuned to handle
that many BROWSEs.

> The solution is to reduce the I/O.

Since you've dismissed all the suggestions offered so far, it appears
your only recourse is to reduce the number of users of the CICS region.

> There is only one dataset on the pack in use.
> 
> The vendor (niche product) refuses to change the application.
> 
> Reducing the I/O is a re-development effort.

The term "re-development" implies programming effort.  Programming
effort implies availability of the source code.

> Even a CMDT is not the complete answer.
> We have a similar (and canned) application that had its most 
> popular file moved to a CMDT, and the resource usage didn't 
> go down significantly.

Define "significantly".

> Because of that, my management wants a justification to re-do 
> this one.

Define "justification".

> I guess I'll have to go elsewhere.
> Nobody here wants to answer my question.

Somebody already posted a URI to the exact page in the CICS Performance
Guide that answered the exact question you restated earlier.  There
likely is no more accurate answer to the precise question you restated,
outside of VSAM Development.

Therefore I must conclude that you don't want the question answered.

    -jc-

----------------------------------------------------------------------
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