Re: There's random access and then there's random access

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Tom Lane wrote:
> Gregory Stark <stark(at)enterprisedb(dot)com> writes:
>> Recently there was a post on -performance about a particular case where
>> Postgres doesn't make very good use of the I/O system. This is when you try to
>> fetch many records spread throughout a table in random order.
>
>> http://archives.postgresql.org/pgsql-performance/2007-12/msg00005.php
>
> Since the OP in that thread has still supplied zero information
> (no EXPLAIN, let alone ANALYZE, and no version info), it's pure
> guesswork as to what his problem is.
Nonetheless, asynchronous IO will reap performance improvements.
Wether a specific case would indeed benefit from it is imho irrelevant,
if other cases can indeed be found, where performance would be
improved significantly.

I experimented with a raid of 8 solid state devices, and found that
the blocks/second for random access improved signifacantly with the
number of processes doing the access. I actually wanted to use said
raid as a tablespace for postgresql, and alas, the speedup did not
depend on the number of drives in the raid, which is very unfortunate.

I still got the lower solid-state latency, but the raid did not help.

Regards,
Jens-Wolfhard Schicke
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)

iD8DBQFHUwM7zhchXT4RR5ARAsziAJ9qm/c8NuaJ+HqoJo9Ritb2t92fRwCgnF9J
r5YU/Fa0mNYG7YXed4QW7K4=
=Mvyj
-----END PGP SIGNATURE-----