we have a sql2000 in the headquarter, and 3 new remote branches opening in few weeks with no sql servers installed yet.
We plan to setup a merge replica. The headquarter will have the highest activity on the replicated tables.
Shall we use all SQL2000 in the remote branches? Or could we use SQL2005, using a 2005 as a publisher and distributor?
TIA
If possible, upgrade all your nodes to SQL 2005. Merge replication performs and scales much better with the new pre-computed partitions. If this is not possible, then the only supported configuration is that the version of the distributor >= version of the publisher >= version of the subscriber. You can find more info about this in topic Using Multiple Versions of SQL Server in a Replication Topology".
|||ok thanks.we're not going to upgrade the only sql2000 server we have, but i'd prefere to use sql2005 in the remote branches: in this scenario, i could set one of the 2005 as distributor and publisher.
This would be in contrast with the network topology, that centers everything in the main plant; BTW there will not be a big traffic of replicated data: a theoretical maximum of about 10 records/minute.
