RE: SWAP Space Confusion Note 1354525.1

From: <Christopher.Taylor2_at_parallon.net>
Date: Tue, 29 Jan 2013 08:15:40 -0600
Message-ID: <F05D8DF1FB25F44085DB74CB916678E8856C530572_at_NADCWPMSGCMS10.hca.corpad.net>



[Resending for formatting problems]
I'm sure there is a engineering principle at work here in regard to how these machines operate but I have no idea what the technical aspects are.

I can only speculate that whatever RAM you have, will ALWAYS ALWAYS get written to (i.e. full) with currently active pages, and currently unused pages.

Eventually the dirty pages may need to be replaced with active pages. If the process that originally pinned the data into memory is still alive, then the OS is not free to totally remove that data from memory – the process may need it in the future.

Since the pages are taken out of real memory on a priority/least recently used algorithm, eventually these pages that aren't really in use, but cannot be destroyed need to go somewhere when they show up as "next in line" for removal. (I know I'm way over simplifying this and may be totally off the tracks at this point)

Any engineering guys want to chime in here?

Chris

From: AMIT VERMA [mailto:verma.labs_at_gmail.com] Sent: Tuesday, January 29, 2013 7:52 AM
To: Taylor Christopher - Nashville
Cc: odenysenko_at_gmail.com; oracle-l_at_freelists.org Subject: Re: SWAP Space Confusion Note 1354525.1

Christopher, exactly that is reason I want to understand why its require when you already have lots of physical ram.

On Tue, Jan 29, 2013 at 7:12 PM, <Christopher.Taylor2_at_parallon.net> wrote:

Keep in mind that even the OS makers recommend "some" swap space.

(I pulled the RHEL recommendations just for illustration)

https://access.redhat.com/knowledge/docs/en-US/Red_Hat_Enterprise_Linux/5/html/Deployment_Guide/ch-swapspace.html

--
http://www.freelists.org/webpage/oracle-l
Received on Tue Jan 29 2013 - 15:15:40 CET

Original text of this message