Bidirectional replication versus peer-to-peer replication

If you want to replicate data between tables on two servers, you have two choices for multidirectional replication: bidirectional replication or peer-to-peer replication.

The following information will help you decide whether to choose bidirectional or peer-to-peer replication to better meet your business needs. If your configuration requires more than two servers, then peer-to-peer replication is the only choice offered for multidirectional replication.

Scenarios that work best with bidirectional replication

Consider choosing bidirectional replication for the following circumstances:
  • You do not expect conflicts to occur in your configuration, and you do not need to check if conflicts do occur. For minimal overhead and network traffic, specify for both servers to ignore conflicts.
  • You do not expect conflicts to occur in your configuration, you want to check if conflicts do occur as a safety measure, and it is acceptable to have one server win if an unexpected data collision occurs.
  • One server updates only certain columns of your data, and the other server updates the other columns. If you specify that the Q Apply program is to check both key and changed columns for conflicts, the Q Apply program merges updates that affect different columns in the same row.
  • One server updates only certain rows of your data, and the other server updates other rows.

Scenarios that work best with peer-to-peer replication

Consider choosing peer-to-peer replication if conflicts might occur in columns with LOB data types. Because versioning columns are used to detect conflicts in peer-to-peer replication, columns with LOB data types are handled the same as columns with other data types.