From: | Craig Ringer <ringerc(at)ringerc(dot)id(dot)au> |
---|---|
To: | Ondrej Ivanič <ondrej(dot)ivanic(at)gmail(dot)com> |
Cc: | pgsql-general(at)postgresql(dot)org |
Subject: | Re: Postgres vs other Postgres based MPP implementations |
Date: | 2011-11-09 00:10:38 |
Message-ID: | 4EB9C4FE.4020104@ringerc.id.au |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-general |
On 11/08/2011 06:36 PM, Ondrej Ivanič wrote:
>> Are you saying there's too much random I/O? Is it possible it'd benefit from
>> a column store?
>> When you're using Greenplum are you using "Polymorphic Data Storage" column
>> storage "WITH (orientation=column)" ?
>
> yes, exactly. Column store and compression speed up queries even more
> (sometimes beyond 100x times) comparing to postgres.
Yup, that'll happen. Right tool, meet right job.
PostgreSQL isn't a column-oriented DB optimised for OLAP workloads. One
that is will outperform it in OLAP workloads. This is not surprising. If
a free compressed column store implementation becomes available for
PostgreSQL I'm sure that'll change, but it's a lot of work and right now
the only people who've done that work aren't giving it away for free -
or not in any form that can be integrated into PostgreSQL without
removing other capabilities other users need.
A non-ACID-compliant database will outperform PostgreSQL in transaction
processing (OLTP) workloads, too; if you don't need ACID compliance then
PostgreSQL won't be the best choice there either. You won't beat a
document database for fetching and retrieving free-form records for
similar reasons. Right tool, right job.
That's not to say Pg can't improve. It can, and not just by adding
column store or index-structured table support. Improved parallelism
capabilities are needed in Pg, but like column store support are a LOT
of HARD work and not something anyone can whip up overnight. Especially
someone doing it for fun in their spare time who we have no right to
make any demands of, or someone working on the project for a company
that has its customers' priorities to think of first.
> Yeah, I know about those.. I like iotop but enterprise distributions
> do not ship fresh kernels... I need something which can I "safely"
> (slightly worse performance is acceptable but machine must survie) run
> in production for several hours and then cross reference it with
> postgres and other system logs
I don't do "enterprise" distributions, so I can't help there. Ask their
support maybe, presumably you pay for it for a reason.
--
Craig Ringer
From | Date | Subject | |
---|---|---|---|
Next Message | Craig Ringer | 2011-11-09 00:14:50 | Re: Getting Error On pg_dump |
Previous Message | Prashant Bharucha | 2011-11-08 21:49:18 | Getting Error On pg_dump |