Re: high through put system

From: Brian Peasland <dba_at_nospam.peasland.net>
Date: Wed, 23 Aug 2006 14:53:30 GMT
Message-ID: <J4GHD8.KBs@igsrsparc2.er.usgs.gov>

Bob Jones wrote:
> <vissuyk_at_yahoo.com> wrote in message
> news:1156311764.875771.198770_at_i3g2000cwc.googlegroups.com...

>> Hi All,
>>
>> We have a Servlet application that has to process 10K requests per
>> second.  In every request, we need to perform one lookup and one insert
>> operation.   The data size per select/insert is about 1K (makes up one
>> row).
>>

>
> That would depend on what kind of system you have.
>
> What do you run servlets on?
> Are they running on a separate machine from the database?
>

>> We are using Oracle 10.2. We are no where near the needed through put.
>> How can we scale this application?. Any suggestions are appreciated.
>> Any proven patterns will be very much appreciated. We are going to have
>> RAC too (soon), but this wont add to the tyhroughtput, I see it is a
>> increasing the availability.
>>

>
> You need to be more specific about what is not scaling up and how the
> throughput is measured. There are may factors that can affect your
> throughput.
>
>

I'm going to assume that you are 1) using bind variables after reading Sybrand's post and 2) have properly tuned your SQL statements.

If you are not achieving the throughput you desire, you'll need to determine where your bottlenecks lie in order to understand where to focus your efforts. It makes no sense to look at CPU or memory utilization if your bottleneck is I/O. So you'll have to go to your wait events to find the bottleneck. And once you know where to focus, take appropriate action. I know this sounds vague, but I do not have any specifics to go on in order to help you further.

HTH,
Brian

-- 
===================================================================

Brian Peasland
dba_at_nospam.peasland.net
http://www.peasland.net

Remove the "nospam." from the email address to email me.


"I can give it to you cheap, quick, and good.
Now pick two out of the three" - Unknown

Received on Wed Aug 23 2006 - 09:53:30 CDT