In a replication system, the primary copy of a data row is the definitive copy. An update committed in the primary database is authoritative and is distributed to all databases with subscriptions for the data.
Replication Server distributes transactions after they are committed in the primary database. Because changes made to replicate data are not distributed, make the data in replicate databases read-only for clients and route all client transactions to the primary database.
There are four ways to update primary data in a replication system based on Replication Server:
Primary data maintenance is centralized at the primary site. Clients cannot update primary data from remote sites.
Clients at remote sites update primary data through network connections.
Clients at remote sites update primary data using request functions.
Primary data maintenance is distributed at multiple primaries. Any resulting conflicts must be avoided or resolved.
