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

Jeremiah Jahn wrote:
| Slony seems to be running a little slow, and I was wondering if there is
| a way to speed things up that I missed in the documentation.
|
| It is taking more than 48 hours to replicate one 58 gig db to another
| machine over a gigabit network. The client and server have raided 15K
| scsi 360 drives. This initial dump is taking more time than a dump and
| restore. I thought this would move a little faster. Did I miss something
| in the documentation? Is there some setting that would vastly speed
| things up? I'm using sl-1.0.5 and pg-7.4.5.

Slony initilization always takes longer than a dump/restore. I recommend
removing non-critical indexes before starting load process and replacing
them after the load completes. By non-critical, I mean all the indexes
on replicated tables except the UNIQUE one on the primary key.

Once the initial copy completes, you can restart the slon with a nice
big grouping number like say -g 80 to facilitate the catch-up (assuming
the origin has been active durring the subscribe and your machines are
beefy enough).

Almost guaranteed, your bottleneck is the disk IO. Putting WAL on a
seperate RAID1 pair is critical. This is where it's actually worth
buying 15kRPM drives. I assume you're running RAID10 for PGDATA. The
only real win there is adding mirror pairs to the stripe. Going from
10kRPM to 15kRPM doesn't make much difference since seek times don't
change much.

- --
Andrew Hammond    416-673-4138    ahammond at ca.afilias.info
Database Administrator, Afilias Canada Corp.
CB83 2838 4B67 D40F D086 3568 81FC E7E5 27AF 4A9A
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.5 (GNU/Linux)

iD8DBQFB1X/Igfzn5SevSpoRAtdBAJ9gOoQHxtS3kYRRmb8+PpZw2DxGqQCfS+yt
/Ly6OPvDg1sxIq5x/mwowgI=
=piuq
-----END PGP SIGNATURE-----