Oracle FAQ Your Portal to the Oracle Knowledge Grid
HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US
 

Home -> Community -> Mailing Lists -> Oracle-L -> RE: Comment on 'Practical Oracle 8i'

RE: Comment on 'Practical Oracle 8i'

From: Guy Hammond <guy.hammond_at_avt.co.uk>
Date: Fri, 17 Aug 2001 04:39:15 -0700
Message-ID: <F001.0036E9AA.20010817042337@fatcity.com>

It's not actually too difficult to do this, if you have some hardware with lots of CPUs. With Solaris 8, you can create "processor sets", grouping a number of CPUs together, then running processes on those CPUs, they will not get any cycles scheduled on other CPUs in the system. This is different from Logical Partitions, because it only affects CPU, not memory, disks, etc. See psrset(1M) for more details. It would be trivial to script the creation of different sized groups, start Oracle on a group and run a benchmark, shut down and resize the group and so forth.

The concurrency of the application makes a huge difference. My multithreaded application, with a single open connection to Oracle performing transactions sequentially is capable of 7.8 (fairly involved) transactions/sec on a quad-SPARC. But, through experimentation, I have discovered that, in my specific case, having 33 connections and transactions running in parallel gives me optimal performance (over 26 transactions/sec, the bottleneck is now disk I/O). Transactions per thread per second went down from 7.8 to 0.79, of course, but overall throughput was massively improved.

The moral of this story is, performance is a subtle art, and can often be counter-intuitive unless you understand the underlying principles.

g

-----Original Message-----

Sent: Thursday, August 16, 2001 6:13 PM
To: Multiple recipients of list ORACLE-L

Of course, it would be rather nice to be able to set up an entire environment
with a high-stress application, and
then run a test which kept the total
available MHz constant but changed
the number of chips. But even then
you'd have to be very careful about what it was you were actually measuring.
--

Please see the official ORACLE-L FAQ: http://www.orafaq.com
--

Author: Guy Hammond
  INET: guy.hammond_at_avt.co.uk

Fat City Network Services    -- (858) 538-5051  FAX: (858) 538-5051
San Diego, California        -- Public Internet access / Mailing Lists

--------------------------------------------------------------------
To REMOVE yourself from this mailing list, send an E-Mail message
to: ListGuru_at_fatcity.com (note EXACT spelling of 'ListGuru') and in the message BODY, include a line containing: UNSUB ORACLE-L (or the name of mailing list you want to be removed from). You may also send the HELP command for other information (like subscribing). Received on Fri Aug 17 2001 - 06:39:15 CDT

Original text of this message

HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US