A SAN Volume Controller cluster has several features that can be used to deploy a high availability storage system with no single point of failure.
Each I/O group within a cluster consists of a pair of nodes. If a node fails within an I/O group, the other node in the I/O group assumes the I/O responsibilities of the failed node.
If a cluster of SAN Volume Controller nodes is split into two partitions (for example due to a SAN fabric fault), the partition with the majority of nodes continues to process I/O operations. If a cluster is split into two equal-sized partitions, a quorum disk is accessed to determine which half of the cluster continues to read and write data.
Each SAN Volume Controller node has four fibre-channel ports, which can be used to attach the node to multiple SAN fabrics. For high availability, attach the nodes in a cluster to at least two fabrics. SAN Volume Controller software incorporates multipathing software that is used for communication among SAN Volume Controller nodes and for I/O operations among SAN Volume Controller nodes and storage systems. If a SAN fabric fault disrupts communication or I/O operations, the multipathing software recovers and retries the operation through an alternative communication path. Also for high availability, configure your host systems to use multipathing software. Then if a SAN fabric fault or node failure occurs, I/O operations among host systems and SAN Volume Controller nodes are retried. Subsystem device driver (SDD), which is multipathing software, is available from IBM® at no additional charge for use with SAN Volume Controller. For additional information about subsystem device driver (SDD), go to the Support for IBM Systems Web site:
The SAN Volume Controller Virtual Disk Mirroring feature can be used to mirror data across storage systems. This feature provides protection against a storage system failure.
The SAN Volume Controller Metro Mirror and Global Mirror features can be used to mirror data between clusters at different physical locations for disaster recovery.