Bruno Wolff III <bruno@wolff.to> writes:
> Rajesh Kumar Mallah <mallah.rajesh@gmail.com> wrote:
>> create unique index ab on t(a,b);
>> create index b on t(b);
> If the b data is large you might get some performance increase from adding
> an index on just a. You will be trading off increased work when doing updates
> to handle the extra index versus faster searches in a more compact index.
This is covered in some detail in TFM:
http://www.postgresql.org/docs/8.2/static/indexes-multicolumn.htmlhttp://www.postgresql.org/docs/8.2/static/indexes-bitmap-scans.html
regards, tom lane