Re: count(*) very very slow with few records ?

I generally split entities with a CLOB column into two parts, with a one-to-one relationship between them. One part handles information about the CLOB (the parent table) and the other just has a PK and the CLOB. I've found that this improves performance and manageability. Technically you shouldn't have to do this, but it's a technique that will work on pretty much any DBMS. In this case your count(*) would operate on the small parent table, which would have a relatively small high water mark.

linux wrote:

> Hello,
>
> Any idea for that :
> count(*) very very slow with few records ?!
> The table contains one CLOB column and have frequently "insert/delete"
> operations
>
> Thanks
Received on Mon Jun 19 2000 - 00:00:00 CDT