SQL Replication/Federated Databases Help

[anon]

Question about SQL Replication in a SQL 2000 Federated
Databases environment:

We're running Win2K3 on our production servers, SQL 2000
w/SP3a, we're currently looking to expand our database
operations and looking into Federated Databases. We
currently have replication going on between two of the
SQL servers (in different physical locations) for our
databases.

My question is - if we move to a Federated Database
solution, is replication still an option, and are there
any known issues with replication not working well with
FDB's?

My follow-up to that is - is replication even needed in a
FDB environment?

Help is much appreciated on this!

C

[Hilary Cotter]

Federated database solutions require each instance of SQL Server which is
part of the federation to have a sql server database which contains a
partition of that data.

Consider the authors table. You could have the first 5 states in the first
server in the federation, the next 5 in the second, the following 5 in the
third, etc.

You could use replication with filtering from an outside server to push the
data to the correct server for that parition in the federation.

You decision on using replication in a federated database solution will
depend on what you are trying to accomplish.

--
Hilary Cotter
Looking for a SQL Server replication book?
http://www.nwsu.com/0974973602.html

"anon" <anon...@discussions.microsoft.com> wrote in message
news:025601c496a6$afded390$a401...@phx.gbl...

[anon]

Ok...I suppose that is answering my question which was
basically - is replication possible using Federated
Databases? (and is there any known issues/problems with
it if so)

My DBA claims that replication is not possible in a
Federated database because of the way the partitioned
views are set up.

>.
>

[Paul Ibison]

There are no issues that I am aware of. Perhaps your DBA was talking about
replicating the view itself, which would replicate the view's definition and
not the data, whereas creating an indexed view would be precluded due to the
presence of the union operator and multiple databases.
If you have a distributed partitioned view on the publisher, the question is
do you want the same view on the subscriber, or for the data to be
amalgamated? The simplest way is to have separate articles for each table
involved in the view which replicate to separate subscriber tables and
replicate the view definition also to the subscriber. To amalgamate the data
into a central table is more difficult but also possible - just like a
central subscribers and multiple publishers topology.
Regards,
Paul Ibison