Photios Loupis (4T) wrote:

>I am using a composite descending index on the 3 fields,
>ie "LogDate","Field1","Field2" and I am
>not getting the performance that I need. There will never be more
>than 5 records that share the
>same "LogDate" value, but there could be a lot of duplicate data in
>the other 2 fields
>
>Are composite indexes the best way to go in this case?
>
>
>
>

From my experience, no. Probably you will have better results with
indexes on single fields. Luckily create or drop indexes are cheap
operations. You don't need to stop or break anything to perform it.
So, my sugestion is: go try it, create one index at a time and check
your query on every step, watch the results and learn from them.

Ivan.