Re: Process memory behaviour of Oracle

<pidmky_at_gmail.com> wrote in message
news:e4332e76-0d06-4531-8acf-cfbb4d384346_at_e23g2000prf.googlegroups.com...
> Hi,
>
> I have the following problem concerning memory of Oracle 9iR2 processs
> on a RHEL 4 Linux box: I have observed that the 24G RAM memory on the
> Oracle box is pretty full most of the time (c. 98%) and since the box
> is used to host Oracle processes only it comes as no surprise that
> after a ps/top analysis it is the c. 600 or so processes + handful of
> DB instances that appear to be consuming the memory. After taking a
> closer look (i.e., summing over V$SESSION PGA figures and cross
> checking with the OS using the V$PROCESS etc.) the numbers though do
> not add up, i.e., my UNIX/Oracle processes' memory consumption simply
> does not add up to 23.xG etc.
>
> Then I tried the following: I wrote a small C program that acquires
> and the frees a large chunk of memory (e.g., 10G) and ran it as the
> Oracle user under which my DB runs. Not only could it allocate memory
> (which if it is 98% shouldn't happen, i.e., it ought to fail or
> produce a seg fault etc.) but once it had freed it again the RAM on
> the box (as reported by the Linux free command) remained low (e.g.,
> 60%) for a long period of time (e.g., several days.) To rule out the
> effect of swapping I also disabled swap during this test and confirmed
> that my C program did in fact acquire the memory from RAM by checking
> top etc.
>
> Is there any explanation for this behaviour ?
>
> Any thoughts welcome,
> Tony.

The presentation and notes at the following URL may help: http://www.pythian.com/blogs/741/pythian-goodies-free-memory-swap-oracle-and-everything

-- 
Regards

Jonathan Lewis
http://jonathanlewis.wordpress.com

Author: Cost Based Oracle: Fundamentals
http://www.jlcomp.demon.co.uk/cbo_book/ind_book.html

The Co-operative Oracle Users' FAQ
http://www.jlcomp.demon.co.uk/faq/ind_faq.html