Jim Nasby wrote:
> On Jul 25, 2006, at 3:31 PM, Tom Lane wrote:
> >Hannu Krosing <hannu@skype.net> writes:
> >>What would be the use-case for hash indexes ? And what should be
> >>done to make them faster than btree ?
> >
> >If we knew, we'd do it ;-) But no one's put enough effort into it
> >to find out.
>
> Do they use the same hash algorithm as hash joins/aggregation? If so,
> wouldn't hash indexes be faster for those operations than regular
> indexes?
The main problem doesn't seem to be in the hash algorithm (which I
understand to mean the hashing function), but in the protocol for
concurrent access of index pages, and the distribution of keys in pages
of a single hash key.
This is described in a README file or a code comment somewhere in the
hash AM code. Someone needs to do some profiling to find out what the
bottleneck really is, and ideally find a way to fix it.
--
Alvaro Herrera http://www.CommandPrompt.com/
The PostgreSQL Company - Command Prompt, Inc.
