Re: NUMA init.ora settings

I do not know the exact meaning of these parameters however I worked on Sequent's NUMA a while back. So I will try some educated guesses here.

Basically in a NUMA architecture you can bring up one instance on every QUAD (4 processor card). By "binding" an instance to a quad you make sure that the instnace processes do not migrate to another quads. This "binding" of an instance to a quad makes better utilization of hardware resources. I believe the first parameter called instance mapping might be related to this.

The second parameter called the pool size might be ralated to the size of the SGA for each instance on a quad.

Once again, I am trying to do an educated guess here based on my past experience. You may need to consult the platform specific installation document from Oracle. Your hardware vendor is also a good source for help and information. The hardware vendors usually have database specialists, database porting engineers, and consultants who are very familiar with the initialization parameters and optimal configuration of databases for a particular complex hardware platform.

    Regards,
    Sam  

noemail_at_no.no (bert) wrote in message news:<Xns9146D7E66AA22Mikenlddsnl_at_212.120.66.218>...
> I have also asked oracle support about these 3 parameters. I the moment I
> haven't got a reaction. Does anybody know the use of these params.
>
>
>
> dherri_at_acxiom.com (Dave) wrote in
> <d31a72dc.0109071301.1c0fa9c9_at_posting.google.com>:
>
> >Andrew Mobbs <andrewm_at_chiark.greenend.org.uk> wrote in message
> >news:<WPu*fEI5o_at_news.chiark.greenend.org.uk>...
> >> Dave <dherri_at_acxiom.com> wrote:
> >>> Folks, I'm running an Oracle 8.1.7.2.0 database on a GS160 16x16
> >>> (Tru64 5.1) and have noticed 3 hidden NUMA-related params:
> >>> _NUMA_instance_mapping, _NUMA_pool_size, and
> >>> _enable_NUMA_optimization. The latter is set by default to TRUE, but
> >>> the other 2 are currently "Not specified". Does anyone have
> >>> experience with these, such as suggested ranges of values and ways to
> >>> test their effectiveness?
> >>
Received on Sat Oct 27 2001 - 05:54:24 CDT