"Gauthier, Dave" <dave.gauthier at intel.com> writes:
> What are the pros/cons of using Slony vs the nre streaming replication in v9?

- The built-in WAL-based replication is probably somewhat simpler to set
  up than Slony.

- WAL-based replication readily captures DDL changes, which are very
  much more troublesome with Slony.

- WAL-based replication carries over *ALL* changes, in binary form,
  going into the "master," and is therefore unable to cope with the
  following cases:

  - Replicating between different versions of PostgreSQL (e.g. - from
    8.3 to 9.0)

  - Replicating between different architectures (e.g. - from Linux/x86
    to AIX/PowerPC)

  - Replicating only part of a database cluster (e.g. - specific tables
    and sequences)

  - Having any kind of updates running against replicas (e.g. - running
    triggers on replicas that manage caches)

  - Having even the slightest schema differences between nodes (e.g.
    - different indexes)

If you need any of the 5 items listed (and this is taken off the top of
my head; not quite comprehensive), then the WAL-based replication built
into PostgreSQL will NOT be acceptable, as it will not satisfy those
sorts of needs.

There are replication "use cases" where the form of replication built
into PostgreSQL 9.0 makes it unnecessary to consider Slony-I (and
related systems like Londiste and Bucardo).  But there are cases where
the built-in stuff won't work out.
-- 
output = reverse("ofni.sailifa" "@" "enworbbc")
Christopher Browne
"Bother,"  said Pooh,  "Eeyore, ready  two photon  torpedoes  and lock
phasers on the Heffalump, Piglet, meet me in transporter room three"