Re: when to use hugemem
- From: brewman <breakfixit@xxxxxxxxx>
- Date: Fri, 07 Jul 2006 21:36:13 -0500
On Tue, 04 Jul 2006 07:31:15 +0200, Jan Gerrit Kootstra wrote:
brewman wrote:
Our environment currently is a mix of RHEL 3 and 4 on 32 bit Intel.Brewman,
We're planning to move everything to RHEL 4. Almost all of our servers
have over 16 GB of RAM and 4 dual core CPUs. We're primarily running
BEA WL and Oracle. I've done some reading from Puschitz about using
the hugemem kernel for Oracle, but I'm unsure about BEA WL. Is it
necessary to use hugemem on every server over 16 GB? All of the
servers will be HA/clustered. We're planning to test Oracle first with
hugemem. Does anyone have any feel for whether we will see improvement
with WL on hugemem as well? Any gotchas we should be looking out for
during the transformation to RHEL 4/hugemem.
Thanks.
RHEL AS or RHEL ES? Probably AS because of the number of physical cpu's,
am I right?
With the 'standard' smp kernel you should be able to address 16GB, over
the hugemem or large smp kernels are more appropriate.
Kind regards,
Jan Gerrit Kootstra
Thanks for the response Jan. Yes we are using RHEL AS, mostly 3 but likely
moving to 4. So I am hearing we should be using hugemem on every server
over 16 GB, which in my case is almost all of them...web servers would be
the only exception I think. Any gotchas to look out for?
.
- Follow-Ups:
- Re: when to use hugemem
- From: Jan Gerrit Kootstra
- Re: when to use hugemem
- References:
- when to use hugemem
- From: brewman
- Re: when to use hugemem
- From: Jan Gerrit Kootstra
- when to use hugemem
- Prev by Date: Re: Problems with yum!
- Next by Date: 'could not open session'
- Previous by thread: Re: when to use hugemem
- Next by thread: Re: when to use hugemem
- Index(es):
Relevant Pages
|
|
