I have a large-ish (30GB) SQL 2005 database that is being merge-replicated around our wan in a "hub and spoke" topology like this:
All of the subscriber servers replicate twice a night, so each day they all have each other's data. It works really well.
I have just set up a new server ("cor2") with SQL 2008 R2, and I want to introduce it into the replication schema with a view to phasing out the old subscribers and swapping in newer ones pointing to this new server.
So I'm wondering if I can back up the database on "cor" and restore it onto "cor2", then make "cor2" a publisher/distributor and have "cor" subscribe to it. Like this:
So I guess my questions are:
asked Nov 07 '10 at 04:03 PM in Default
Matt, I have never done what you are describing and I cannot see your reasons for keeping 'cor' active, maybe they lie outside the replication processes? Personally I would stop replication on each of the subscribers and then the publisher - cor. Once that was all closed down I would backup the database and restore it to cor2 and then rebuild replication by setting cor2 as a publisher and then subscribing to it from each of the other servers. You can generate script to drop and create replication but in all honesty I have never seen them run without problems - going through the wizards always seems more reliable. YMMV.
answered Nov 08 '10 at 01:25 AM