Posted by G.Fink.Nottle on 06/22/06 02:32

Hello,

Being a bit of a SQL Server novice, need some advice with the following
situation.

Server A and Server B have SQLServer 2000 based databases. The vendor
of the application/system has implemented their own replication process
to ensure the 2 databases are in sync. However, there is no clustering
with virtual IP addresses implemented. So to an external client/db, it
is 2 identical databases with the same name on 2 distinct servers.

We need to develop an application that will reside on a networked
server C and with SQLServer 2000 as well. While most of the tables in
this database are self contained, around 10 tables will have to be
mirror copies of the same tables from either Server A or Server B.

Question, how do we implement subscription based replication on top of
a redundant database, when no clustering is implemented? So, in
essence, when Server A is alive, the database on Server C will
periodically (or on change) replicate the 10 tables from Server A. When
Server A is not alive, it needs to do same from server B. (When both
server A and B are alive, it is acceptable to get data from either,
since they are synchronized internally).

Any alternate suggestions on achieving this functionality are welcome
too. If SQL Server 2005 has some capabilities that may address this
problem, that is a consideration as well.

Thanks

[Back to original message]