Oracle FAQ | Your Portal to the Oracle Knowledge Grid |
![]() |
![]() |
Home -> Community -> Usenet -> c.d.o.server -> Re: CPU Upgrade Advice
This depends heavily on wheather or not your application(s)/DBMS(s) are multi-threaded ( SMP aware ).
Keep in mind that many, perhaps most, operations that take place on a daily
basis are single threaded. If you upgrade to an SMP where each chip is
slower
than the chip in your uni., your users will experience quite a performance
hit on
a system that is not heavily loaded. Doing this will however buy much
better
multiprogramming performance under load.
My experience with an upgrade from RS/6000 R24 to an 8processor R30 was
generally slower execution times for single-threaded transactions (
including
all the standard UNIX stuff; awk, sed, home-grown-reports, etc. ) on a
light to
no load system. However, under mid to heavy load, system throughput was
light-years ahead of the old uniprocessor system.
As for query performance, using the Parallel Query Option can significantly
improve performance on an SMP system. Be carefull though, the tuning gets sticky in an environment with a large number of users. It becomes much easier to reach a resource contention bottleneck as the number of active parallel queries increases ( especially with only 2 CPUs )!
I would suggest you take a look at the most recent TPC benchmark results
and/or
borrow/lease a test system for analysis.
Larry Monack <lmonack_at_sprynet.com> wrote in article
<01bbe371$0a67b940$a0d4aec7_at_larrym>...
> Hypothetical question: In a data warehousing environment, is it better
to
> have 2 cpus running in parallel or 1 cpu that is twice as fast?
>
> I have a 50+GB Oracle data warehouse running on an HP G30 that has 12
disk
> drives connected to 2 SCSI controllers. I'm about to do a CPU upgrade
and
> would like to get as much bang-for-the-buck as I can. I'm concerned
mostly
> about query performance.
>
> Any advice would be greatly appreciated.
>
> Larry Monack
> lmonack_at_fsco.com
> lmonack_at_sprynet.com
>
Received on Fri Dec 06 1996 - 00:00:00 CST
![]() |
![]() |