Re: Suggestions on how to efficiently search a large table

On Nov 14, 8:55 pm, chrism..._at_gmail.com wrote:
> I have a billion+ row table that I need to do queries on.
>
> The table looks something like this:
>
> item
> id number
> process1_id number
> process2_id number
> process3_id char(35)
>
> I need to perform multiple processes on each item. When I process an
> item, I then set the corresponding column with the result of the
> process.
>
> I need to do queries like the following to determine which processes I
> haven't completed yet:
>
> select * from item where process1_id is null or process2_id is null
> or process3_id is null
>
> Any recommendations on how to index or query such a table to get good
> performance? The cardinality of all the process*_id columns is very
> high.
>
> Thanks.

You've provided no information regarding:

• version of Oracle software
• edition of Oracle software
• how many client processes will be hammering the table
• how the table is being loaded
• hardware, OS, clustering, RAC, server color
• desired response time

That I think that Daniel Morgan will jump in on this as soon as he awakens to say that its a homework problem.

-bdbafh Received on Thu Nov 15 2007 - 06:42:54 CST