I have the following scenario :
A cluster hosting a SQL 2005 CENTRAL Database ,6 servers residing in different sites are supposed to push data into 6 tables in the central database ,the tables structure (6 tables) in ALL servers is identical.
,data should be moved in TWO WAY ,from remote sites to the central database ,and vise versa.
We require to replicate 8 tables from the six sites to the central database , so that the central database has data from ALL sites ,the six sites have different data with uniquq flags (primary keys) in the tables.
1- do we use transactional or merge replication? and why?
2- de we need to apply any filters at the publications?
3- how do we solve the problem of a publication initializing the data at the time we create a new subscribtion? this is earasing all data that has been replicated by a previous subscription.