Re: pmap - anon pages
- From: "hopehope_123" <hopehope_123@xxxxxxxxx>
- Date: 28 Jul 2005 05:28:45 -0700
Hi Sybrand ,
Thanks for your message. I am trying to understand the memory
allocation and page cache usage of my system . Since i use ocfs , and
parallel query , in theory no page cache of redhat linux and oracle
buffer cache is in use. But when i monitor the system by using vmstat i
see lots of pagings. My first aim is to understand the real reason of
these pagings. Do these pagings occur because of file cache (oracle
caching ) or not ? (are they just oracle processes anon pages? - With
direct io , data that is read from the disk is put directly into the
process address space , so the process itself has the data directly ,
further requests to the same data again needs disk access since no page
cache is used.)
My second aim is to size an oracle process. Since this is datawarehouse
, and oracle parallel query is mostly used , pga memory is more
important than buffer cache. But although i use automatic pga
allocation , what i see is most of the oracle processes use 10 or 20
MB. private memory . I have 8gb. of memory , parallel query but only
use little amount of private ram but excessive paging (20-30mb. per
second - sar -B output) .
You are right that oracle has views to get the pga size for each
session and process , (although the values does not exactly match
comapring to pmap (no litle differences - great differences ) , but
first i want to understand how things happen.
Kind Regards,
hope
.
- References:
- pmap - anon pages
- From: hopehope_123
- Re: pmap - anon pages
- From: sybrandb@xxxxxxxxx
- pmap - anon pages
- Prev by Date: Re: RMAN question: How do you connect SCN# with real time
- Next by Date: Re: Sybrand Bakker: Gets humiliated in public?
- Previous by thread: Re: pmap - anon pages
- Next by thread: Irritating XMLDOM issue
- Index(es):
Relevant Pages
|