Re: Reducing Systempools

[ouuch@xxxxxxxxxxx]

Hi,

On 30 Jan 2003 at 15:55, Vern Hamberg wrote:

> 1. Keep similar work in a pool, separate from stuff that works
> differently. Jobs that read the same data in a sequential manner could
> well be separated from jobs that read by key.
> 
Basically, our current setup is several years old from times when our AS/400s
always were much too slow.
We are preparing for an HA-solution and I'm trying to streamline the system and
minimize setup issues.
We have a bunch of different applications and requirements, I could easily come 
up
with 10 more memory pools.
Our 820 is at (guesstimate) 70% CPU for batch and interactive should be in that
ballpark, too.

> 2. Change to shared pools and turn on Expert Cache (paging option =
> *CALC). Longer-running sequential data access can scream this way. But
> not if mixed with random, keyed access over the same period. BTW, you
> CAN change private pools to *CALC using the change pool tuning API.
> 
> 3. Just as multiple subsystems can have jobs in *BASE, multiple
> subsystems can run in one shared pool. Depends on your routing entries
> and your subsystem pool definitions.
> 
I guess Work Management Guide is still the place to start?
I'd like to get a setup like:

  *machine
  *base
  *spool
   all interactive
   all batch

Regards,

Oliver