From: Mark Kirkwood <markir@paradise.net.nz>
To: Luke Lonergan <llonergan@greenplum.com>
Cc: Dan Gorman <dgorman@hi5.com>
Cc: Joshua D. Drake <jd@commandprompt.com>
Cc: Craig A. James <cjames@modgraph-usa.com>
Cc: pgsql-performance@postgresql.org
Subject: Re: Reliability recommendations
Date: Sat, 25 Feb 2006 19:10:55 +1300
Message-ID: <43FFF4EF.6060601@paradise.net.nz> (raw)
In-Reply-To: <C0252010.1DC7A%llonergan@greenplum.com>
References: <C0252010.1DC7A%llonergan@greenplum.com>
Luke Lonergan wrote:
>
> OK, how about some proof?
>
> In a synthetic test that writes 32GB of sequential 8k pages on a machine
> with 16GB of RAM:
> ========================= Write test results ==============================
> time bash -c "dd if=/dev/zero of=/dbfast1/llonergan/bigfile bs=8k
> count=2000000 && sync" &
> time bash -c "dd if=/dev/zero of=/dbfast3/llonergan/bigfile bs=8k
> count=2000000 && sync" &
>
> 2000000 records in
> 2000000 records out
> 2000000 records in
> 2000000 records out
>
> real 1m0.046s
> user 0m0.270s
> sys 0m30.008s
>
> real 1m0.047s
> user 0m0.287s
> sys 0m30.675s
>
> So that's 32,000 MB written in 60.05 seconds, which is 533MB/s sustained
> with two threads.
>
Well, since this is always fun (2G memory, 3Ware 7506, 4xPATA), writing:
$ dd if=/dev/zero of=/data0/dump/bigfile bs=8k count=500000
500000+0 records in
500000+0 records out
4096000000 bytes transferred in 32.619208 secs (125570185 bytes/sec)
> Now to read the same files in parallel:
> ========================= Read test results ==============================
> sync
> time dd of=/dev/null if=/dbfast1/llonergan/bigfile bs=8k &
> time dd of=/dev/null if=/dbfast3/llonergan/bigfile bs=8k &
>
> 2000000 records in
> 2000000 records out
>
> real 0m39.849s
> user 0m0.282s
> sys 0m22.294s
> 2000000 records in
> 2000000 records out
>
> real 0m40.410s
> user 0m0.251s
> sys 0m22.515s
>
> And that's 32,000MB in 40.4 seconds, or 792MB/s sustained from disk (not
> memory).
>
Reading:
$ dd of=/dev/null if=/data0/dump/bigfile bs=8k count=500000
500000+0 records in
500000+0 records out
4096000000 bytes transferred in 24.067298 secs (170189442 bytes/sec)
Ok - didn't quite get my quoted 175MB/s, (FWIW if bs=32k I get exactly
175MB/s).
Hmmm - a bit humbled by Luke's machinery :-), however, mine is probably
competitive on (MB/s)/$....
It would be interesting to see what Dan's system would do on a purely
sequential workload - as 40-50MB of purely random IO is high.
Cheers
Mark
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Reply to all the recipients using the --to and --cc options:
reply via email
To: pgsql-performance@postgresql.org
Cc: markir@paradise.net.nz, llonergan@greenplum.com, dgorman@hi5.com, jd@commandprompt.com, cjames@modgraph-usa.com
Subject: Re: Reliability recommendations
In-Reply-To: <43FFF4EF.6060601@paradise.net.nz>
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
This inbox is served by DDX for PostgreSQL; see mirroring instructions
for how to clone and mirror all data and code used for this inbox