Re: performance tuning question

"Stephan Bressler" <stephan.bressler_at_pdb.sbs.de> wrote in message news:aatfgl$7l6$1_at_news.mch.sbs.de...
> Vladimir,
>
> the index on id is a must in your case. It should give you more than 20%
on
> a table with several 1000 rows.

I suspect that my index is not used due to some reason. How do I check if it is used?

>
> Did you analyze your table while it was empty or contained only very few
> rows? This may make the optimizer believe that a full scan is cheaper than
> the index access.
>

How do I analyze the table?
Here's the steps that are done during update:

1. Initially the table has PK set on ID field that is dropped just before the data load, does not seem like I need to drop it but trust me, there are reasons for doing it
2. a temporary index is created on ID
3. data is loaded using the logic I described
4. temporary index is dropped
5. PK is recreated

Like I said it seems that the index created at step 2 is not used during selects and updates, would the above steps confuse the optimizer so full scans are used instead of the index??

Thanks,
Vlad.

> Regards
> Stephan
>
>
Received on Sat May 04 2002 - 02:24:28 CDT