One method is transparent failover in which a secondary standby connection comes online of the primary connection fails. Because the new connection has the same address as the original failed connection, failover is transparent to the server connection and application performance is not affected.
Another method is to use dual or multiple active connections with each connection dedicated for certain logical volumes within a given storage system. If one connection fails the other active connections automatically assume its logical volume workload until it comes back online. During this time the alternative connections support all logical volumes so there might be a slight performance impact depending on workload and traffic patterns.
A third method used for storage path failover also utilizes dual or multiple active connections. In this case however both connections can simultaneously access the logical volumes. This design can improve performance through load balancing but typically requires host based software. During a storage connection failure the alternate active connection continues to access the logical volumes. After the failed connection is repaired the other path becomes active and load balancing resumes.
All of these failover methods are designed to ensure the availabilty of the enterprise applications that use them. In addition failover generally is co ordinated with server software to ensure an active path to data, transparent to the application.
Another effective way to achieve high availability in a SAN environment is by mirroring storage subsystems. SAN enable the efficient mirroring of data on a peer to peer basis across the fabric.
These mirroring functions contribute trmendous fault tolerance and availability characteristics to SAN based data. Combining the mirroring functions with switch based routing algorithms creates a resilent, self healing enironment to support the most demanding enterprise storage requirements. The mirrored subsystems can provide an alternate access point to data regardless of path conditions.
A common use of mirroring involves the deployment of remote sites within the enterprise. Implementing SANs thorugh fibre channel switches enables the distribution of storage and servers throughout a campus, metropolitan and beyond. Fibre channel overcomes many of the distance limitations of traditional SCSI connections enabling devices to be extended over much longer distances for remote mirroring , tape backup and disaster recovery operations.