Due to service-level agreements, industry regulations, or other business needs, companies today demand the highest possible system availability. In fact, incidents that were previously viewed as minor unplanned outages can now impact business operations severely. To address this requirement, Storage Area Networks (SANs) are designed to facilitate a high-availability environment that can help prevent (or better tolerate) system outages.

Some of the key availability benefits of SANs include built-in redundancy, dynamic failover protection, and automatic traffic rerouting capabilities. For instance, flexible connectivity options enable the development of SANs that have no single points of failure. In addition, Brocade Fabric OS software can automatically detect network problems and route traffic around any failed links to help ensure a continuous reliable path for data. Finally, SANs provide hot-plugging capabilities that enable organizations to install, configure, and bring storage online without experiencing server downtime. By combining multiple networked switches with a Fabric OS, organizations can build scalable, strategic SANs that provide extremely high availability.

With their resilient network design, SANs are much better than traditional direct-attached storage at tolerating failure scenarios to prevent system outages. SANs offer built-in redundancy, dynamic failover protection, automatic traffic rerouting capabilities, clustering, and other vital functions. These features meet high-availability demands imposed by service-level agreements, industry regulations, and other business requirements.

• Eliminate single points of failure through the network
• Prevent system outages with dynamic failover protection and automatic traffic rerouting
• Support "five-nines" availability with real-time installation and configuration
• Support high-availability operations by enhancing clustering implementations

SANs can also support high-availability operations by enhancing clustering implementations. Clustering is typically used to ensure that applications continue to run in the event of a host server failure. Traditional, non-SAN clustered environments typically include two servers sharing disk storage. If one server fails, the other server assumes the failed server's workload and continues running the application. The failover server accesses the data through the shared disk. This represents a relatively inflexible design because it is usually limited to two servers sharing storage with the failover server often remaining idle until pressed into duty. In addition, the servers and storage devices are usually located near each other, which provides only limited protection against disasters.

Figure 6. A high-availability clustering model with switch failover capabilities

With SAN clustering, many more servers can share SAN-attached storage. In some implementations, any server can handle the additional workload when one server is unavailable, thus ensuring that there are no idle server resources (see Figure 1). Longer distances between devices also facilitate a more effective disaster recovery plan.

A large telecommunications company employed a SAN infrastructure with a VERITAS Cluster Server and Brocade switches to create an environment with extremely high data and application availability. Before the SAN was deployed, the implementation included three clusters, each with two Sun servers. Resources were being wasted because there were three idle failover servers and neither the application load nor data was shared between clusters. The SAN implementation required just half the storage resources and only three servers (instead of six), any of which can assume a failing server's workload. Today, the company can efficiently manage its SAN by dynamically allocating storage and applications, and by performing non-disruptive server and storage maintenance.