Oracle7 Server Distributed Systems Volume II: Replicated Data

When to Use Conflict Resolution Methods

The purposes of conflict resolution are

to ensure data convergence

to avoid cascading errors

Convergence ensures that all sites in your replicated environment agree and have the same data. Avoiding cascading errors ensures that your system will run smoothly.

Attention: If one (or more) rows in a transaction causes a conflict that remains unresolved, the entire transaction is written to the error log. Subsequent transactions that depend on the original transaction can now conflict, and in turn, be written to the error log.

If one or more sites in your replicated environment propagate changes asynchronously, conflicts can occur if two or more sites update the same replicated data. Even if your environment is designed to avoid conflicts (for example, by partitioning data ownership), it is prudent to

monitor the occurrence of any unresolved conflicts

use a notification method to learn of any unexpected conflicts

Note: If you do not designate a conflict resolution method, Oracle logs any unresolved conflicts as error transactions in the DefError view of the receiving site.

If all of your sites propagate changes synchronously and you have no updatable snapshot sites, conflicting updates cannot occur, and you do not need to designate a conflict resolution method.