Re: Slow queries on big table

"Tyrrill, Ed" <tyrrill_ed(at)emc(dot)com> writes:
> Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> writes:
>> This combination of indexes:
>>
>>> Indexes:
>>> "backup_location_pkey" PRIMARY KEY, btree (record_id, backup_id)
>>> "backup_location_rid" btree (record_id)
>>
>> is really just silly. You should have the pkey and then an index on
>> backup_id alone.

> Thanks for the help guys! That was my problem. I actually need the
> backup_location_rid index for a different query so I am going to keep
> it.

Well, you don't really *need* it; the two-column index on (record_id,
backup_id) will serve perfectly well for queries on its leading column
alone. It'll be physically bigger and hence slightly slower to scan
than a single-column index; but unless the table is almost completely
read-only, the update overhead of maintaining all three indexes is
probably going to cost more than you can save with it. Try that other
query with and without backup_location_rid and see how much you're
really saving.

> Index Scan using backup_location_bid on backup_location
> (cost=0.00..9573.07 rows=415897 width=8) (actual time=0.106..3.486
> rows=2752 loops=1)
> Index Cond: (backup_id = 1070)
> Total runtime: 4.951 ms

That's more like it ;-)

regards, tom lane