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

Home -> Community -> Usenet -> c.d.o.server -> Re: multiblock read count question

Re: multiblock read count question

From: Daniel A. Morgan <Daniel.Morgan_at_attws.com>
Date: Wed, 20 Jun 2001 11:27:56 -0700
Message-ID: <3B30EB2C.47CBB4F9@attws.com>

Rob K wrote:

> Why would I be getting the same response time on full scans if the
> multiblock read count is set to 2 or 32?
>
> System:
> Oracle 8.1.7 block size 8k
>
> Sun 4500
> 2-5200 drives
> Veritas
> stripe = 4 drives with 256k interleave size = 1M stripe size.
> max i/o = 1M
>
> I change the user sessions multiblock read count to 2 and do a full scan on
> a table and sum up a column.
> It takes aprox 9 min to scan 60,043,128 rows 430080 blocks.
>
> I change the user sessions multiblock read count to 32 run the same query
> and it takes about the same time.

Did you bounce the database in between your tests? No doubt you have block buffers caching much of the data. Or ... this isn't your bottleneck and you need to look elsewhere.

Daniel A. Morgan Received on Wed Jun 20 2001 - 13:27:56 CDT

Original text of this message

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