Re: BULK COLLECT - Can it be implemented for a few rows .

Aravindh wrote:
> Hi ,
>
> The original requirement was to REDUCE the amount of DISK READS. Since
> the OLD PROC was consuming a heavy disk reads and it was loading the
> database . Thats why we designed a new procedure using the BULK COLLECT
> approach .
>
> I read all over the net and all the m,aterials said that the BULK
> collect was the fastest and the efficient . But the net disk reads did
> not come down when it was run on the INCREMENTAL basis for about 850
> rows / 4500 rows.
>
> However the total disk reads when run for a FULL LOAD did come down for
> the new procedure .
>
> So can you please suggest ways to reduce the disk reads in the NEW PROC
> .
>
> Thanks a lot for your concern
>
>
> Bye
> KN Aravindh

Assuming what you are doing is grabbing related information ... I would look at turning the table into an IOT or a sorted hash cluster (if 10g).

-- 
Daniel A. Morgan
http://www.psoug.org
damorgan_at_x.washington.edu
(replace x with u to respond)