Cluster managers and high availability disaster recovery (HADR)

You can implement Db2® high availability disaster recovery (HADR) databases in non-pureScale environment on nodes of a cluster, and use a supported cluster manager to improve the availability of your database solution.

Note: HADR deployments in Db2 pureScale® environments use a cluster manager for High Availability within each cluster. There is no automation that is provided to perform a cluster-level takeover for HADR.
You can have both the primary database and the standby database that is managed by the same cluster manager, or you can have the primary database and the standby database that is managed by different cluster managers.

Set up an HADR pair where the primary and standby databases are serviced by the same cluster manager

This configuration is best suited to environments where the primary and standby databases are at the same site and where the fastest possible failover is required. These environments would benefit from using HADR to maintain DBMS availability, rather using crash recovery or another recovery method.

You can use the cluster manager to quickly detect a problem and to initiate a takeover operation. Because HADR requires separate storage for the DBMS, the cluster manager should be configured with separate volume control. This configuration prevents the cluster manager from waiting for failover to occur on the volume before using the DBMS on the standby system. You can use the automatic client reroute feature to redirect client applications to the new primary database.

Set up an HADR pair where the primary and standby databases are not serviced by the same cluster manager

This configuration is best suited to environments where each site has a cluster manager with one or more nodes. The automated HADR failover is provided by the cluster manager, on the site where the primary database is located. The cluster manager of the other site can still provide automation of bringing back the standby database after unexpected outage, as well as provide automated failover after the primary database has been moved manually to the other site. Below are the two possible failover scenarios.

  • Cluster manager can provide automated recovery when a partial site failure occurs. If failure affects the HADR primary database, the cluster manager can automatically failover the primary database to the remaining node on this site.
  • If a complete site failure occurs where the primary database is located, there is no automated failover. You can initiate a manual HADR failover to maintain DBMS availability. If a complete site failure occurs where the standby database is located, you can repair the site or move the standby database to another site.