Josh Berkus wrote:
>
>
> The other major issue with the Bitmap index patch as it stood in 2007 was
> that performance just wasn't that much faster than a btree, except for
> specific corner cases. Otherwise, someone else would have been interested
> enough to pick it up and finish it.
>
> So performance testing of the patch is absolutely essential.
>
>
As Simon mentioned - index creation time and size was certainly improved
considerably.
There were certainly cases when row retrieval performance was not
improved much (as compared to using a comparable btree index) - my
analysis was that these were typically when heap fetch time dominated
index scan time i.e it didn't matter how good your index access was, you
were mired in heap seeks. ISTM that this situation will change
dramatically when index only access (via dead space map? or similar)
arrives.
Note that even if only for the on disk size savings, these are worth
having for data warehousing situations.
regards
Mark