Re: Getting even more insert performance (250m+rows/day) - Mailing list pgsql-performance

From Markus Schaber
Subject Re: Getting even more insert performance (250m+rows/day)
Date
Msg-id 447AD6D1.3070904@logix-tt.com
Whole thread Raw
In response to Re: Getting even more insert performance (250m+rows/day)  ("Daniel J. Luke" <dluke@geeklair.net>)
Responses Re: Getting even more insert performance (250m+rows/day)  ("Daniel J. Luke" <dluke@geeklair.net>)
List pgsql-performance
Hi, Daniel,

Daniel J. Luke wrote:

> One big transaction every 5 minutes using 'COPY FROM' (instead of
> inserts).

Are you using "COPY table FROM '/path/to/file'", having the file sitting
on the server, or "COPY table FROM STDIN" or psql "/copy", having the
file sitting on the client?

From our tests, having the file on the server can speed up the things by
 factor 2 or 3 in some cases.

Also, using BINARY copy may give great benefits due to lower parsing
overhead.

As you say you're I/O bound, spreading tables, indices, wal and input
file to different spindles won't help you much.


HTH
Markus



--
Markus Schaber | Logical Tracking&Tracing International AG
Dipl. Inf.     | Software Development GIS

Fight against software patents in EU! www.ffii.org www.nosoftwarepatents.org

pgsql-performance by date:

Previous
From: "chris smith"
Date:
Subject: Re: select with "like" from another table
Next
From: "Cstdenis"
Date:
Subject: Re: How can I make this query faster (resend)