Improving System Availability With SANs
For small and large enterprises alike, business continuance plans have become a necessity. The risks of not having a plan in place to ensure 24/7 availability can be more expensive than putting one in place. The best possible plan must be created and implemented, because you cannot afford to do a bad job. If a disaster recovery plan is needed, then it had better work right the first time.
The stakes are higher today than ever before. According to the Fibre Channel Industry Association, system downtime costs can run into the millions of dollars per hours from $14,500/hour in lost automatic teller system (ATM) fees for a banking institution, to as much as $6.45 million/hour from disrupted operations in a stock brokerage firm.
To improve business continuance, Storage Area Networks (SANs) are being incorporated into enterprise systems utilizing a combination of redundant components, connections, software, and configurations, to minimize or eliminate single points of failure.
By reducing or eliminating single points of failure in enterprise environments, SANs help improve the overall availability of business applications. This high availability is achieved not through a single product, but through a comprehensive, fault-tolerant system design that includes all the components in the SAN and supports 24/7 uptime requirements. Delivering a high availability environment through a SAN requires establishing availability objectives, creating fault tolerance, and implementing an intelligent SAN infrastructure and fabric management.
With the increasing importance of the Internet and global e-business applications, more and more companies are implementing computing infrastructures specifically designed for at least 99.999 percent (the "five nines") availability, or the equivalent of less than 5.3 minutes of downtime a year.
Availability is a function of the frequency of outages (from unplanned failures or scheduled maintenance and upgrades) and the time to recover from those outages. Companies must identify specific availability requirements and predict potential failures in order to create a high availability solution that meets the needs of the organization. Objectives vary widely both among and within companies - some can tolerate no disruption, while others may be only minimally affected by short outages.
To address this uptime issue, many companies are now implementing networked fabrics of Fibre Channel devices to ensure a high-performance storage environment. These flexible SANs incorporate fault tolerance through redundancy, mirroring, hot-plugging capabilities, and no single points of failure. They also speed recovery through simplified fault monitoring, diagnostics, and non-disruptive server/storage maintenance and repair. The use of intelligent routing and rerouting, coupled with dynamic failover protection, minimizes human intervention during failover events.
Achieving True Fault Tolerance
One of the most effective ways to increase system availability is to implement fully redundant SANs consisting of alternate devices, data paths, and configurations. Particularly important is ensuring dual paths through separate components. This is especially true when physical location and disaster tolerance are a concern a single device cannot adequately address these issues.
For better availability, the focus shifts from servers to applications. Mission-critical applications should reside on highly available servers and storage devices so data can be accessed even during a failure. Sophisticated software enables application or host failover by moving workload to a secondary server, and clustering technology transfers workload to multiple active servers without disrupting data flow.
To further improve availability, servers should include redundant hardware components with dual power supplies, network connections and mirrored system disks. Servers should have multiple connections to alternate storage devices through Fibre Channel switches and a minimum of two independent connections to the SAN. In addition, these servers should feature dual-active or hot-standby configurations with automatic failover capabilities.
Another likely failure point for system availability is the path between the server and storage, including Host Bus Adapters (HBAs), cabling, fabrics or storage connections. Dual-redundant HBA configurations help ensure path availability and boost performance through the additional SAN connectivity.