Oracle RAC provides such a high availability (HA) environment with automatic failover and load balancing of transactions across cluster nodes (database instances). It allows for full read and write access to data across all nodes since they share a common storage. If there is a node failure, the cluster goes through a short period of reconfiguration to remove the failed node. It also recovers the transactions from that node to ensure the integrity of the data files. Once that is done the workload continues seamlessly. Oracle RAC also allows all the user sessions from the failed node to automatically reconnect to the surviving nodes.
Oracle Extended RAC increases database resiliency between two locations, while it benefits from the cluster HA. When Oracle Extended RAC is deployed, it allows both the source and target storage systems to be used simultaneously for both reads and writes. The cluster can have some nodes co-located with the source storage system and some with the target storage system. All database writes are mirrored between the storage systems. Reads are served to the node from the storage system closest to it. If there is a node failure, the user sessions automatically reconnect to the surviving nodes.
Oracle Extended RAC is not a new concept, however, it has not been deployed extensively due to concerns about performance, server overhead, and the complexity of the solution. Since writes are mirrored by a logical volume manager software on each cluster node (host-based mirroring) it consumes host resources. Write I/Os are doubled, and CPU cycles are consumed due to the increase in host writes activity. Distance between the two sites adds latency and impacts performance. The number of devices seen by each node in the cluster doubles to include the mirrored devices. Each cluster node has its own agent responsible for mirroring and health monitoring. In most cases, there is a requirement for a third “arbitrator” site to determine a surviving database site in the case of a disaster. Even more complexity is added as the DBAs need to carefully place cluster quorum files in each of the sites, including the arbitrator.
Due to these challenges, although Oracle Extended RAC improves HA, the deployment rate remained low. Dell EMC SRDF/Metro offers a deployment solution for Oracle Extended RAC without the complexities involved in host-based mirroring.