Re: My DB has has 5TB, many operations are very slow (on Google Cloud Compute)

From: Melvin Davidson <melvin6925(at)gmail(dot)com>
To: Francisco Olarte <folarte(at)peoplecall(dot)com>
Cc: Rakesh Kumar <rakeshkumar464(at)outlook(dot)com>, David A <david(at)scalaacademy(dot)com>, "pgsql-general(at)postgresql(dot)org" <pgsql-general(at)postgresql(dot)org>
Subject: Re: My DB has has 5TB, many operations are very slow (on Google Cloud Compute)
Date: 2016-10-11 19:50:29
Message-ID: CANu8Fix-9xNuRER1p3CfEdWAWNaGw16og_VByypi5xqm6U37WA@mail.gmail.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-general

On Tue, Oct 11, 2016 at 3:16 PM, Francisco Olarte <folarte(at)peoplecall(dot)com>
wrote:

> Rakesh:
>
> On Tue, Oct 11, 2016 at 9:00 PM, Rakesh Kumar
> <rakeshkumar464(at)outlook(dot)com> wrote:
> >>Cores do not help, postgres is single-threaded. RAM MAY help, but I
> > I hope this is no longer true from 9.6 for those queries where PG can
> use parallelism.
>
> It does, AFAIK, but for queries, not AFAIK for this kind of data
> moving ops ( and I doubt it will, as presently you can easily saturate
> the channels with a single core for that kind of simple ops, and
> normally if you want to optimize this kind of op is better to target
> concurrency ( table can be used while moving ) than pure speed .
>
> Francisco Olarte.
>

>Requiring and exclusive table lock does not imply slownes. Just try
>'lock table x in exclusive mode' on an idle system. Pretty fast.

Sure on an idle system, you will get a table lock right away, but OP's
statements imply a large busy system.
And if there are transactions occurring against that table, there is no
telling how long it will take. Since we
do not have enough specific info, I stand by my statement.

--
*Melvin Davidson*
I reserve the right to fantasize. Whether or not you
wish to share my fantasy is entirely up to you.

In response to

Responses

Browse pgsql-general by date

  From Date Subject
Next Message Chris Richards 2016-10-11 19:52:53 Re: LOG: munmap(0x7fff80000000) failed: Invalid argument
Previous Message Jason Dusek 2016-10-11 19:29:28 SERIALIZABLE and INSERTs with multiple VALUES