Re: Load experimentation

From: Greg Smith <greg(at)2ndquadrant(dot)com>
To: Ben Brehmer <benbrehmer(at)gmail(dot)com>
Cc: pgsql-performance(at)postgresql(dot)org, Thom Brown <thombrown(at)gmail(dot)com>, Kevin Grittner <Kevin(dot)Grittner(at)wicourts(dot)gov>, craig_james(at)emolecules(dot)com, kbuckham(at)applocation(dot)net, scott(dot)lists(at)enterprisedb(dot)com
Subject: Re: Load experimentation
Date: 2009-12-08 07:35:12
Message-ID: 4B1E01B0.7070605@2ndquadrant.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-performance

Ben Brehmer wrote:
> Since my input data is split into about 200 files (3GB each), I could
> potentially spawn one load command for each file. What would be the
> maximum number of input connections Postgres can handle without
> bogging down?
You can expect to easily get one loader process per real CPU going.
Beyond that, it depends on how CPU intensive they all are and what the
resulting I/O rate out of the combination is. You're probably going to
run out of CPU on a loading job long before you hit any of the other
limits in this area, and potentially you could run out of disk
throughput on a cloud system before that. PostgreSQL isn't going to bog
down on a connection basis until you've reached several hundred of them,
your loader will be lucky to hit 10 active processes before it grinds to
a halt on some physical resources unrelated to general database scaling.

--
Greg Smith 2ndQuadrant Baltimore, MD
PostgreSQL Training, Services and Support
greg(at)2ndQuadrant(dot)com www.2ndQuadrant.com

In response to

Browse pgsql-performance by date

  From Date Subject
Next Message Scott Marlowe 2009-12-08 07:58:50 Re: Load experimentation
Previous Message Ben Brehmer 2009-12-08 07:22:10 Re: Load experimentation