On Dec 6, 2010, at 9:41 AM, Jan Wieck wrote:

>>
>
> You would build two replication sets. Set one originating on server A
> and subscribed by server B, set two originating on server B and
> subscribed by server A. Like Vick said, as long as you don't need
> partial table replication, you'll be fine.
>
> Adding tables later works the same as if you had just one origin. You
> need to create a temporary set that you add those tables to, subscribe
> it to the same nodes that are subscribed to the existing set and when
> those subscriptions are complete, you merge the sets together.
>

Assuming each chunk of data is isolatable (say, everything related to  
branch X)

Couldn't we create a partitioned table where we replicate say  
partition1 from a->b and partition2 from b->a?

then we'll see a happy unified view of a table..

That could be interesting... Now I'm thinking about a couple potential  
ideas based on this.

--
Jeff Trout <jeff at jefftrout.com>
http://www.stuarthamm.net/
http://www.dellsmartexitin.com/