Re: What kind of index to use for many rows with few unique values?

On 2 Dec 2002 at 17:10, David F. Skoll wrote:

> I've heard that a btree index performs badly in this situation.

As another poster has shown, it should be OK.  I recently dealt with
distributions simlar to your example.

> Is a hash index appropriate?  I've heard bad things about hash
> indexes in PostgreSQL.

No, don't use them.  I didn't get any performance increase out of
them.  Does your experience show poor btree results?
--
Dan Langille : http://www.langille.org/