|
Mike, Your mileage will vary, depending on your application mix. Generally you would calculate a base level for your machine and *BASE pools, then allocate your remaining memory to provide acceptable interactive response time and then provide the remainder to your base and spooling pools. Your machine pool is vital and the manual referenced below provides formulas for calculating a starting point. If you have not had the opportunity to read much in the Work Management manual, I would suggest looking at the section with the heading 'Setting Performance Values Manually'. This provides some of the guidelines for setting initial values and then adjusting them based upon paging and faulting levels. Tuning is somewhat of an iterative process - you give it your best shot, then adjust the values based upon the results you obtain You can do some of this using the values given by the automatic performance tuning. When your system reaches what you consider to be its peak interactive load, turn off performance adjustment and check your paging and faulting levels. Below are the maximum acceptable levels used by performance adjustment (from the manual above): Pool Minimum Faults Maximum Faults *MACHINE 10.00 10.00 *BASE 10.00 100 Interactive 5.00 200 Spool 5.00 100 Batch 10.00 100 With your configuration, your batch jobs should be running in *BASE but the same faulting levels apply. Gradually lower your interactive pool size until your paging/faulting levels exceed your maximum acceptable level, then make the pool slightly larger. That would be an appropriate pool size for your workload. You also need to deal with your activity levels and should be familiar with how adjusting the activity level of a pool affects performance. After dealing with pool size and activity levels, then determine whether you should set your pool paging to *FIXED or *CALC. The above is grossly oversimplified but could provide a starting point. You can feel free to play with pool sizing and activity levels because you can easily return to your status quo by setting you QPFRADJ value back to '3'. I would ask again whether your system is performing poorly. If it is working well under system adjustments, then you have more flexibility than if you manually set your values. The system will give highest priority to your interactive work and the operating system, but will make memory available for batch work if it is not required by those tasks. Regards, Andy Nolen-Parkhouse IBM Certified Expert, iSeries Technology > On Behalf Of Condon, Mike > Sent: Saturday, October 20, 2001 1:16 PM > > What pool sizing is recommended for mostly green screen apps, occasional > batch? > > -----Original Message----- > From: Andy Nolen-Parkhouse [mailto:aparkhouse@mediaone.net] > Sent: Saturday, October 20, 2001 8:44 AM > > Mike, > > I generally measure system performance by user complaints, not by > response times. If your users are content with the performance of your > system, then why change anything? > > From the looks of your system status displays, you really do not have > any pool allocations because all of your pools are managed by the > system. If you have some grizzled veterans among your counterparts in > the other divisions, they may feel that the performance adjustments > could never perform the level of tuning that they can obtain manually. > This has a historical base in the earlier releases of OS/400, which > really did not do an exemplary job of managing resources. It may also > have a base in arrogance and professional pride. > > As a grizzled veteran myself, I've found that for a fairly generic > system, performance under system control provides more than adequate > allocation of resources and responds to changing requirements promptly > enough. It is 'good enough'. If you start to require special batch > subsystems or something different, you may find some manual tuning > beneficial. > > What would your counterparts propose? > Are your users content? > > Andy > > Andy Nolen-Parkhouse > IBM Certified Expert, iSeries Technology > > > On Behalf Of Condon, Mike > > Sent: Friday, October 19, 2001 4:49 PM > > > > Some of our counterparts in other divisions managing AS400's have > > commented > > negatively > > on our pool size allocation, and also against using IBM's QPRFADJ set > to 3 > > (they prefer 0). > > Our system is primarily green screen rpg/db2 stuff, fairly > > straightforward. > > We have about > > 200 open interactive sessions on a typical day. Our interactive CPW is > 70. > > > > My question is - do any of you all have any recommendations for system > > tuning given our application? > > > > Here is a snapshot of our wrksyssts: > > > > Work with System Status > > S105DGAM > > 10/19/01 > > 15:29:26 > > % CPU used . . . . . . . : 7.0 Auxiliary storage: > > > > % DB capability . . . . : 3.0 System ASP . . . . . . : > > 77.30 G > > Elapsed time . . . . . . : 00:00:40 % system ASP used . . : > > 29.5261 > > Jobs in system . . . . . : 2233 Total . . . . . . . . : > > 77.30 G > > % perm addresses . . . . : .007 Current unprotect used : > > 2589 M > > % temp addresses . . . . : .053 Maximum unprotect . . : > > 54509 M > > > > > > Type changes (if allowed), press Enter. > > > > > > > > System Pool Reserved Max -----DB----- ---Non-DB--- > > > > Pool Size (M) Size (M) Active Fault Pages Fault Pages > > > > 1 111.03 67.31 +++++ .0 .0 15.3 25.2 > > > > 2 201.78 .26 43 .0 .0 3.7 13.9 > > > > 3 10.23 .00 8 .0 .0 4.0 8.0 > > > > 4 700.93 .00 36 1.5 6.0 10.8 25.0
As an Amazon Associate we earn from qualifying purchases.
This mailing list archive is Copyright 1997-2024 by midrange.com and David Gibbs as a compilation work. Use of the archive is restricted to research of a business or technical nature. Any other uses are prohibited. Full details are available on our policy page. If you have questions about this, please contact [javascript protected email address].
Operating expenses for this site are earned using the Amazon Associate program and Google Adsense.