Re: Question about Hardware & Configuration for Massive

From: "Jim C(dot) Nasby" <jnasby(at)pervasive(dot)com>
To: Dan Sugalski <dan(at)sidhe(dot)org>
Cc: Kevin & Jessica Hermansen <kjhermansen(at)gmail(dot)com>, pgsql-general(at)postgresql(dot)org
Subject: Re: Question about Hardware & Configuration for Massive
Date: 2006-01-19 22:05:54
Message-ID: 20060119220554.GH78403@pervasive.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-general

On Thu, Jan 19, 2006 at 02:26:03PM -0500, Dan Sugalski wrote:
> At 11:39 AM -0700 1/19/06, Kevin & Jessica Hermansen wrote:
> >I'm looking to set up an informational database which will be
> >accessed by the general public on our website (there will be no
> >internal users). The database will likely have about 10 million
> >records with a total database size of 200-300 GB. All 10 million
> >records will be very similar (the same data fields), so I anticipate
> >that the entire database will be just 1 table. This is the main
> >application that will be running on our server. Users will be
> >querying the database by searching either a particular field (or
> >multiple fields) or by doing a full text search.
>
> Is the public access going to be entirely read-only? You may find it
> advantageous to have multiple database machines kept synchronized
> with slony, and distribute the queries from the website across the
> read-only systems. If a reasonably priced machine can manage, say,
> five simultaneous nasty queries in your timeframe, it's usually a
> *lot* cheaper to buy ten of those systems to handle fifty queries
> than it is to beef up the system enough so that one machine can
> handle fifty queries.
>
> It also makes it a lot easier to add on more capacity as you need it
> (or as the budget allows) this way -- throw an extra machine or three
> into the cluster as time and cash permit and need demands.

Actually, it doesn't need to be read-only for that to work, though it
does simplify things. But if you make your application smart enough to
know that read queries go to one set of hosts while updates have to go
to one specific host you can use Slony to scale. You'd need to be able
to tolerate the replication delay as well.
--
Jim C. Nasby, Sr. Engineering Consultant jnasby(at)pervasive(dot)com
Pervasive Software http://pervasive.com work: 512-231-6117
vcard: http://jim.nasby.net/pervasive.vcf cell: 512-569-9461

In response to

Browse pgsql-general by date

  From Date Subject
Next Message Doug McNaught 2006-01-19 22:16:48 Re: Upgrade Problem: 7.4.3 -> 8.1.2
Previous Message Rich Shepard 2006-01-19 22:04:34 Re: Upgrade Problem: 7.4.3 -> 8.1.2