Re: FW: Hosted LPAR performance

CPW, rPerf on partitioned systems is a can of worms.
I attended a few sessions by Ear Jew on AIX performance tunning and it
really is a lot to take in (And IBM i DOESN'T have vmstat AFAIK). It
depends on the size of the LPAR (entitled vs virtual cpu), locality of
memory and whole bunch of other details on how the machine and LPAR are
setup.
Multi chip LPARs suffer the most from what I understood. And having a lot
of "uncapped" 0,05 LPARs instead of giving them a bit closer to their real
consumption also hurts performance (due to the forced
allocation/deallocation vs gracefully giving up unused cycles).
all in all, it's a very nice topic even if it seems a bit like being Neo
and watching the Matrix code...

On Fri, Feb 15, 2019 at 5:35 PM Nathan Andelin <nandelin@xxxxxxxxx> wrote:

On Fri, Feb 15, 2019 at 12:51 PM Bob Cagle <bcagle@xxxxxxxxxxx> wrote:

They said they temporarily uncapped us to allow us to flex above the 960
CPW we were allocated.

I question the validity of a CPW claim on any partitioned system. There's
overhead pertaining to partition swapping, which could be huge. I'm not
aware of any IBM guidelines concerning CPW measurements on partitioned
systems.
--
This is the Midrange Systems Technical Discussion (MIDRANGE-L) mailing list
To post a message email: MIDRANGE-L@xxxxxxxxxxxxxxxxxx
To subscribe, unsubscribe, or change list options,
visit: https://lists.midrange.com/mailman/listinfo/midrange-l
or email: MIDRANGE-L-request@xxxxxxxxxxxxxxxxxx
Before posting, please take a moment to review the archives
at https://archive.midrange.com/midrange-l.

Please contact support@xxxxxxxxxxxx for any subscription related
questions.

Help support midrange.com by shopping at amazon.com with our affiliate
link: https://amazon.midrange.com