Resynchronizing Data among Replication Servers

If replication has failed for some reason and data is not synchronized, there are different ways to correct data mismatches between replicated tables.

The following table compares each of the methods. All methods except manual table unloading and reloading can be performed while replication is active.

Table 1. Resynchronization methods

Method	Description
Direct synchronization	Replicates all rows from the specified reference server to all specified target servers for a replicate or replicate set. Runs as a foreground process by default, but can run as a background process. Populates tables in a new participant. Quickly synchronizes significantly inconsistent tables when used with the TRUNCATE statement.
Checking consistency and then repairing inconsistent rows	Compares all rows from the specified target servers with the rows on the reference server, prepares a consistency report, and optionally repairs inconsistent rows. Runs as a foreground process by default, but can run as a background process.
ATS or RIS file repairs	Used to repair rows that other synchronization methods could not repair. Repairs a single transaction at a time. Replicates or replication server must have been configured with the ATS or RIS option.
Manual table unloading and reloading	Manual process of unloading the target table, copying the reference table, and then loading the reference table into the target database. Requires that replication be suspended.