> On Tuesday 01 September 2009 03:26:08 Pierre FrĂŠdĂŠric Caillaud wrote:
>> > We have a table that's > 2billion rows big and growing fast. We've
>> setup
>> > monthly partitions for it. Upon running the first of many select *
>> from
>> > bigTable insert into partition statements (330million rows per month)
>> the
>> > entire box eventually goes out to lunch.
>> >
>> > Any thoughts/suggestions?
>> >
>> > Thanks in advance
>>
>> Did you create the indexes on the partition before or after inserting
>> the
>> 330M rows into it ?
>> What is your hardware config, where is xlog ?
>
>
> Indexes are on the partitions, my bad. Also HW is a Dell server with 2
> quad
> cores and 32G of ram
>
> we have a DELL MD3000 disk array with an MD1000 expansion bay, 2
> controllers,
> 2 hbs's/mount points runing RAID 10
>
> The explain plan looks like this:
> explain SELECT * from bigTable
> where
> "time" >= extract ('epoch' from timestamp '2009-08-31 00:00:00')::int4
> and "time" <= extract ('epoch' from timestamp '2009-08-31 23:59:59')::int
> ;
>
> QUERY PLAN
> ------------------------------------------------------------------------------------------------
> Index Scan using bigTable_time_index on bigTable (cost=0.00..184.04
> rows=1
> width=129)
> Index Cond: (("time" >= 1251676800) AND ("time" <= 1251763199))
> (2 rows)
This looks like a single row matches your conditions. Have you run ANALYZE
on the table recently? Try to run "ANALYZE BigTable" and then the explain
again.
BTW what version of PostgreSQL are you running?
Tomas
--
Sent via pgsql-performance mailing list (pgsql-performance@xxxxxxxxxxxxxx)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-performance
