Re: Index performance

From: Gordon E. Hooker <gordonh_at_acslink.net.au>
Date: 1996/08/20
Message-ID: <321a2b6d.1161993_at_news.thehub.com.au>#1/1


njs3_at_doc.ic.ac.uk (Niall Smart) wrote:

>Naren Chintala <naren_at_mink.att.com> wrote:
>
>
>>> We are having performance trouble with a batch process and we've
>>> narrowed it down to index performance.
>>>
>>> We have created a brand new tablespace on a new drive. Our
>>> testing indicates a 1000 record insert that takes 6 second
>>> non-indexed, takes 55 seconds or more when indexed in any of a
>>> variety of ways.
>
>
>>Insert into an indexed table will take more time than an
>>non-indexed table. This is obvious because each insert
>>has to update the index also.
>
>Isn't the drop from 6 seconds to 55 seconds a bit excessive though? I
>didn't think indexes slowed updates *that* much?

That depends on how many and the complexity of the index!

>
>Niall
>
>

_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/ Gordon E. Hooker MACS PCP
25 Clarke Street Ripley Queensland Australia Phone 61-7-32940555
Email gordonh_at_acslink.net.au
_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/ We are just two lost souls living in a fish bowl Year after year

Pink Floyd Received on Tue Aug 20 1996 - 00:00:00 CEST

Original text of this message