MS SQL Server Main Replication Models

Replication Publisher and Subscriber :

To understand how SQL replication works, we’ll have to define two replication terms – Publisher and Subscriber. In a SQL replication the Publisher is the part that offers the data for distribution. The Subscriber is the part that consumes (gets updates) the data from the Publisher.




MS SQL Server Replication

MS SQL Server has 3 main replication models – Merge Replication, Transactional Replication and Snapshot Replication.

The Merge replication is used when both Publisher and Subscriber need to make changes to their respective databases. In this case both databases might have been changed between runs of the Merge SQL replication, and the replication merges the changes in both locations. Of course when using merge replication you should be aware that there might be conflicts, for example duplicated primary keys. If there is a conflict, then the merge replication follows predetermined conflict resolution plan to correct the issue.

The Transactional replication is usually used with databases where data changes frequently and there’s need for constant refreshing of the data. The replication process watches the publisher’s database for any changes, if there are changes, it distributes them over to the replication subscribers.

The Snapshot replication is relatively rarely used, and as its name implies it does a snapshot of the publisher database at certain point of time and distributes it to the subscribers. The Snapshot replication is time and resource consuming compared to the other two SQL replication types and is used when you need to fully overwrite the database at the subscriber or when you do an initial database replication update, which will be refreshed using one of the other 2 SQL replication types after the initial copy.