Scott Marlowe scott.marlowe at gmail.com
Mon May 9 13:09:42 PDT 2011
(Please stick to plain text on the lists.  Some clients can't display
html email well.)

On Mon, May 9, 2011 at 1:35 PM, Little, Douglas
<DOUGLAS.LITTLE at orbitz.com> wrote:
>
> Hi,
>
> We use greenplum (based on PG 8.2.13) for our data warehouse. We’re on a 10node system, with 1 master that coordinates the work.  We use before row triggers a lot to implement at rest encryption.

I'd ask Greenplum, as I don't know how well their system works with
slony.  Other than that, the issues you face when dealing with a large
number of objects seem lessened with slony 2.0 versus 1.2.  1.2 has
some pretty pessimistic locking and as you get into the thousands of
objects it can get a little slow at some things like creating a set.
We replicate ~900 objects.  With lots (45k) of other objects in the
master database, we take up to 5 hours to create set.  create set from
a master without those 45k extra objects set creation takes < 5
minutes.  That's with slony 1.2.  We haven't (and probably wont) try
slony 2.0 in production any time soon.


More information about the Slony1-general mailing list