WebSphere and non-IP-based database failover

 

For non-IP database failover such as Oracle Parallel Server (OPS) and Real Application Cluster (RAC), the failover process is much faster, usually finishing in seconds. Therefore, very few failures can be observed. This type of HA data management is especially suitable for the WebSphere persistent session data. Some transparent failover techniques that database vendors use further reduce the failures clients may observe.

The failed database instance may cause data inconsistency. This data inconsistency is repaired by starting a new instance and performing crash checks in the IP-takeover situations discussed above. However, since the backup database instance is pre-existent, a new instance does not need to be started in the non-IP-takeover situation. The pre-existent healthy database instance must repair any data inconsistency problems left by the failed instance.

Compared with IP-based failover (within one to five minutes), non-IP-takeover failover is very fast (within seconds). Therefore, clients see very few failures.

  Prev | Home | Next

 

WebSphere is a trademark of the IBM Corporation in the United States, other countries, or both.

 

IBM is a trademark of the IBM Corporation in the United States, other countries, or both.