Re: Rein in QZDASOINIT

Charles Wilt wrote:

Maybe.

It probably won't hurt unless the box is tight on memory already.

I agree with 'Maybe', without knowing to any degree of certainty. I bring this up every once in a while, hoping that an expert will either confirm, contradict or modify.

It seems to me that that big initial advantage in a separate memory pool is simply that it becomes far easier to _measure_ performance values.

With a bunch of different kinds of jobs in the same pool, it's difficult to separate the effects of one kind of job from another. Faulting and paging rates get reported by system pool numbers. If a given pool shows higher numbers than it should, how can you tell what's causing it?

Also, if these things (and others) are left to run in *BASE, AFAIK, you might as well give up on the performance adjuster. With a bunch of active jobs always using *BASE, the adjuster has a much harder time trying to move memory to pools that request it. (That's a pretty long stretch of my understanding, but it _seems_ to make sense.)

Tom Liotta

On Fri, Jan 23, 2009 at 5:32 PM, <elehti> wrote:

Will a separate memory pool for QZDASOINIT jobs help Brian's situation?
Anyone? EricL
-----Original Message-----

moving the jobs to their own subsystem and changing the memory pool on

them