Release 10.1B: Progress Fathom Replication
User Guide


How the failover process works

Fathom Replication typically occurs with activity between the replication server on the source database on the primary machine and the replication agent on the target database on the secondary machine. If the replication agent loses communication contact with the replication server, you perform failover to move all database update activity from the source database to the target database. Once the primary machine becomes available again, you can move back, or fail back, all the database update activity to the primary database or machine. To minimize downtime to your application, you can schedule the failback process to run when you want.

The process shown in Figure 4–1 illustrates the primary failure on the primary machine and what ensues when you perform failover.

Figure 4–1: Fathom Replication initial failover process

As shown in Figure 4–1:

  1. Fathom Replication is running between the primary machine, which hosts the source database, and a secondary machine, which hosts the target database. Replication activity is occurring normally between the Replication server on the primary machine and the Replication agent on the secondary machine.
  2. If a primary failure occurs and communication between the Replication server and Replication agent ceases, the failover process to transition the database on the secondary machine from a target to a source database can begin.
  3. At this point, any source AI extents that were not processed can be applied to the target by using the following command:

    dsrutil db-name -C applyextent 
    

  4. Use the DSRUTIL transition command, as shown:
  5. dsrutil db-name -C transition 
    

The transition operation restarts the database on the secondary machine as a source database, which starts the Replication server (but not the database on the primary machine). All after-image activity continues to accumulate in the secondary after-image extents until replication can be established from the database on the secondary machine to the database on the primary machine. This process is known as secondary replication, and it begins the overall failback process.


Copyright © 2006 Progress Software Corporation
www.progress.com
Voice: (781) 280-4000
Fax: (781) 280-4095