Re: code question: storing INTO relation

From: Simon Riggs <simon(at)2ndquadrant(dot)com>
To: Michael Paesold <mpaesold(at)gmx(dot)at>
Cc: pgsql-hackers(at)postgresql(dot)org, Greg Stark <gsstark(at)mit(dot)edu>
Subject: Re: code question: storing INTO relation
Date: 2004-10-22 19:13:16
Message-ID: 1098472396.20926.68.camel@localhost.localdomain
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

On Fri, 2004-10-22 at 19:20, Michael Paesold wrote:
> Greg Stark wrote:
> > In Postgres CREATE TABLE AS is currently being treated as a synonym for
> > SELECT
> > ... INTO ... So I think this may be an awkward feature to add. Also, like
> > reindex the logging would still be necessary for online backups. So this
> > may
> > be a dead-end direction in the long term.
>
> Putting aside the CREATE TABLE AS/SELECT ... INTO vs INSERT ... SELECT
> confusion, online backups seem to be an issue.

...*the* issue...

>
> Is there any command currently, that circumvents WAL and the result is still
> consistent with PITR?

None, nor could there be.

> How is this done in other databases?

As I have said in previous posts, the user-level feature to turn off
logging for certain large operations is a feature offered by DB2, Oracle
and Teradata and has been in production use for around 15 years in many
large datawarehouses.

> Implementing features that don't work with PITR does not feel right at all,
> of course, since eventually many if not most of the people will use
> continues logging.
>

I agree, hence why this should be a user option. The usage of this is
restricted to particular classes of database usage: data warehousing or
very large database applications. This isn't intended for use in OLTP or
web-site databases.

"feel right" is a subjective experience. I argue for the option only,
not to force everybody to use it if it is not sensible for your
application. Not all databases have the same purpose....

On this, and other issues, I will argue for further Data Warehousing
features, as options, in PostgreSQL. My reading of the user base is that
around 50% of users (Scientific + Data Warehousing users) would benefit
from improvements in this area.

--
Best Regards, Simon Riggs

In response to

Responses

Browse pgsql-hackers by date

  From Date Subject
Next Message Jan Wieck 2004-10-22 19:35:49 Re: ARC Memory Usage analysis
Previous Message Tom Lane 2004-10-22 18:53:43 Re: Nice vacuums