Re: Is It Too Big ? - Customer Data Warehouse Growth Projection - Mailing list pgsql-general

From Tom Lane
Subject Re: Is It Too Big ? - Customer Data Warehouse Growth Projection
Date
Msg-id 23361.1011277998@sss.pgh.pa.us
Whole thread Raw
In response to Is It Too Big ? - Customer Data Warehouse Growth Projection  (Mark kirkwood <markir@slingshot.co.nz>)
Responses Re: Is It Too Big ? - Customer Data Warehouse Growth  (Neil Conway <nconway@klamath.dyndns.org>)
Re: Is It Too Big ? - Customer Data Warehouse Growth  (Mark kirkwood <markir@slingshot.co.nz>)
List pgsql-general
Mark kirkwood <markir@slingshot.co.nz> writes:
> I would appreciate any suggestions on how to plan for this growth.

> - Clearly hardware will need to be looked at (can we cope continuing to
> use Intel based platform(s)? )
> - Also software, are we okay using Postgresql for a 300G database (is
> 7.2 aimed at this sized undertaking?), How about the os ( can we keep
> using Linux or Freebsd ?).

7.2 is aimed more at 24x7 operation; as far as size of database goes,
I wouldn't think it would be much better than 7.1.  Probably the major
issue for you is that PG doesn't have any provision for spreading
a database across multiple filesystems --- so you will have to put the
whole DB on one honkin' big RAID array, or use an LVM layer to spread
the filesystem across multiple drives in software.

I seem to recall having heard that Linux has some ~100GB restriction on
the size of a single filesystem, which could be a problem.  That might
be obsolete information, but be sure to check max filesystem size for
whichever kernel you select.

> I am informed that the expected number of users is low, so that major
> challenges are big queries and the weekly loads. Most of the space used
> by the database will be in 1 very big and 1 big table.

PG's hard limit on the size of a single table is 2 billion pages, which
is 16TB with the default page size and 64TB if you compile with BLCKSZ
set to 32K.  The nonstandard choice might be a good option for this DB.

The prospect of having to dump this database for backup seems a tad
daunting.  What are you using for a backup device?

            regards, tom lane

pgsql-general by date:

Previous
From: Emmanuel SARACCO
Date:
Subject: how to limit postgresql process and memory time
Next
From: Steve Wampler
Date:
Subject: Re: Is It Too Big ? - Customer Data Warehouse Growth