On Apr 14, 2:17 pm, david(dot)t(dot)wil(dot)(dot)(dot)(at)gmail(dot)com ("David Wilson") wrote:
> For instance, your count of distinct userids is probably not using the
> index you just created. If it still isn't using it after you ANALYZE
> the table, try rewriting the query using group by (select count(*)
> from (select userid from mytable group by userid) tmp). I recently had
> a similar performance issue on a 75m row table, and the above helped.
>
thanks. There's a lot of good info in your post. I'll give it a try.