Re: SQL Delete Performance - Why so Horrible?

[William Washington III <w.washington3@xxxxxxxxxxxxx>]

Thanks for all who responded so far...

More info:

The file has no indexes.  (I just added one by company to test some things
out.)

The file will delete a high percentage of records (from 50% to 100%, based on
the number of companies currently being processed.)  I can only delete the
records of the company being processed, so that eliminates DROPping and
reCREATEing the file.

Will a non-unique index help with this high of a delete percentage?

Is there a way to "turn off" the overhead for this file, because I really use
it as a work file (but I must keep it until I run the process again.)

... a tricky thing, this SQL stuff...

William


> date: Thu, 14 Apr 2005 15:44:30 -0500
> from: "Elvis Budimlic" <ebudimlic@xxxxxxxxxxxxxxxxxxxxxxxxx>
> subject: RE: SQL Delete Performance - Why so Horrible?
> 
> I forgot to ask, how many indexes already exist over this physical file?  If
> there are many, they need to be maintained during the delete.
> 
> -----Original Message-----
> Subject: SQL Delete Performance - Why so Horrible?
> 
> Hi All,
> 
> Why is the SQL DELETE performace so terrible?
> 
> I have a file with 1,500,000 records.  When I do:
> 
> DELETE FROM filename WHERE company = pCompany
> 
> in a stored procedure running in batch, it takes nearly 30 minutes!  Of
> course,
> an SQL DROP or a CL "CLRPFM" works within seconds, but I really must check
> the
> company before deleting.
> 
> Would indexing by company help?
> 
> Thoughts?  Ideas?  Thanks.
> 
> William