All,

Have SQL Server 7.0 Transactional replication with central publisher,
publishing to multiple subscribers.
Developers have implemented 3 tables that only exist on the subscribers and
the data within them is subscriber specific. These 3 tables have FK
relationships to a PK on a published table.
This was causing a problem when snapshot occurred on the published table as
the RI prevented this data being deleted on the subscriber. Overcame this by
placing steps in Distribution Agent before and after distrib.exe ran, the
dropped and then re-created the FK constraints (this worked - in testing so
far anyway).

My question is however, the topology of transactional replication is that
all data updates occur at the publisher and all are replicated to
subscribers. Is there another way around the above ie. if follow
transactional replication correctly all 4 tables should be published, but how
can the data be subscriber specific.
I can only think that the tables have a key that identifies the specific
subscriber but even in that case, all subscribers will still have the other
subscribers data, it can just be filtered when DML statements are used.
Is it possible to publish the data to be subscriber specific though ? (I
guess its a reverse of the central subscriber topology - but trying to do the
same thing).

Appreciate any thoughts on this.