From: Greg Sohl To: All
I've seen and heard quite a bit about installing V2R1. Bug reports, installation nightmares, performance degradation problems and some folks saying, "Problems? What problems?" I've become fairly satisfied that V2R1 is safe to install, provided you have all the necessary PTFs, but I am still concerned about the possible performance degradation. We are about to install it on a B45. Can anyone give me the low-down on what the impact will be from a performance standpoint? Any and all assistance is appreciated.
From: Ed Smith To: Greg Sohl
We installed V2R1 on our B40 a few months ago. I don't recall experiencing any poor performance. We just upgraded it to a D45, which makes us very happy, so we may just be forgiving of any previous poor performance. I think those folks on smaller AS/400s may have something, if anything, to worry about.
From: Charles McLean To: Greg Sohl
We haven't noticed any performance changes in evening batch work, but sometimes during the day interactive work behaves differently--it seems to get balled up more frequently.
From: Ed Smith To: Charles McLean
Have you tried using the auto-tune feature? If you change the QPFRADJ system value to '2,' it'll size your shared pools and their activity levels to meet current demand. It really works!
From: Charles McLean To: Ed Smith
We tried shared pools for interactive work but were not impressed, so we have gone back to manually tuned pools. Our needs are to:
1) Give Customer Service the most consistent response time,
2) Limit the amount of resources taken by programming, and
3) Give everything else to all other interactive users.
Without more horsepower, our current limiting factor, this seems to work the best. We still have minimal page faults. What size is your machine?
From: Mike Guenther To: Greg Sohl
We installed V2R1 with no problems, however it is noticeably slower than V1R3. Our machine is a lightly used B35. Funny thing, however, I've noticed that every once in awhile for no apparent reason, the system utilization skyrockets and brings the system to its knees. This doesn't last very long (approximately five minutes, although I've never timed it) and then it fixes itself. During one of these incidents, I looked at WRKACTJOB and found that virtually nothing was running that could account for this. It's kind of scary to observe CPU utilization of 98.6 percent on an idle system. Hope it's not contagious. I've been told by another user that this same condition has occurred on her D45 machine. I hope this loop will be fixed in subsequent PTF releases. Yes, I'm at the latest PTF and have applied all PTFs that I could find which relate to my machine. Did I miss one?
From: Greg Sohl To: Mike Guenther
Thanks for the information. It seems, based on the messages I've read here and on NewsLink, that there is about a 50-50 split among those who have had problems, bugs or performance, and those who have had little or no problems. I guess you just do your homework, then load it and hope!
From: Ed Smith To: Charles McLean
Private pools are exempt from auto-tune. If you set up your critical work in private pools, and everything else in shared pools, and then maybe cross your fingers, your critical work should not suffer from changes in pool size or activity levels. Also, I just recently found out that interactive jobs run from the console are killers--especially backups. Any job run from the console (using default attributes) runs at priority 10. All other interactive jobs run at priority 20.
From: Charles McLean To: Ed Smith
Yep. Look at WRKACTJOB with F11 and see the RUNPTY for different jobs. First few looks really open your eyes. Good luck.
LATEST COMMENTS
MC Press Online