GZIP of pre-zipped output - Mailing list pgsql-performance

From Dave Crooke
Subject GZIP of pre-zipped output
Date
Msg-id ca24673e1003211004l68237f72r101eb04082f8e288@mail.gmail.com
Whole thread Raw
Responses Re: GZIP of pre-zipped output
List pgsql-performance

If you are really so desparate to save a couple of GB that you are resorting to -Z9 then I'd suggest using bzip2 instead.

bzip is designed for things like installer images where there will be massive amounts of downloads, so it uses a ton of cpu during compression, but usually less than -Z9 and makes a better result.

Cheers
Dave

On Mar 21, 2010 10:50 AM, "David Newall" <postgresql@davidnewall.com> wrote:

Tom Lane wrote:
>
> I would bet that the reason for the slow throughput is that gzip
> is fruitlessl...
Indeed, I didn't expect much reduction in size, but I also didn't expect a four-order of magnitude increase in run-time (i.e. output at 10MB/second going down to 500KB/second), particularly as my estimate was based on gzipping a previously gzipped file.  I think it's probably pathological data, as it were.  Might even be of interest to gzip's maintainers.

pgsql-performance by date:

Previous
From: David Newall
Date:
Subject: Re: pg_dump far too slow
Next
From: Bob Lunney
Date:
Subject: Re: pg_dump far too slow