Hi, I am considering using replication server for a project, and was
hoping some of you could share your knowledge.  I want to use
replcation server to matain a central consolidation server and about 5
globally distibuted regional servers.  I need to replicate all data on
each regional server to the central server, and reference data from
the central server to each regional server.  I have a few questions
about this.

1.  If two regional server update the the same field with like this:
update t1 set c1= c1+500 where t1.id = 50

how will replication server handle this.  will the result in the
central server be that resulst of the last regionals servers update,
or will it be the sum of all the regional servers updates.

2.  How do identity columns behave under replication.  Can they be
replicated.  Can the max value be reset after replication?

If this is documented somewhere, can you point me there.

Thanks
-Ethan

Get yourself on a repserver course.

The way that replication works, is that the rep server takes the before
and/or the after image from the transaction log, then works out what
transaction it needs to make the change, according to the subscription.

When using n-way replication, you have to remember that the selection field
(ie. office_id) needs to be in each table to be replicated.

The repserver couse covers these ideas, and I remember discussing a similar
setup to the one you are contemplating.
You can't use identity, as it is unique per table/db/server, and will come
out different at the other end. (similar problem with  timestamp).
One golden rule of replication, especially when you get to the complicated
stuff. You have to design your system to work with replication (including
the database schemas). You can not apply replication easily to an existing
schema, and expect to have great success.