52
Curtis Chan,
Senior Editor,
Broadcast Beat Magazine
High availability storage has seen explosive growth over the years, as users become increasingly reliant on their digital assets and the speed at which those assets can be transferred from work stations to servers, and back again. In systems where a single file server is used for user-critical assets, a server failure can have from inconvenient to catastrophic consequences. As such, mission-critical storage systems have evolved; developers and manufacturers have taken various approaches to this growing need.
In a virtualized environment or a 24-hour business cycle where keeping data and file assets safe are critical, the need for high availability becomes evident. A high-availability cluster can be used to provide critical shared data on a network, and a popular approach is to mirror all data in real time between identical storage servers. Using RAID storage servers with sufficient capacity to meet current and future needs, along with specialized operating systems providing seamless and highly fault-tolerant mirroring operations, is a very good method for guaranteeing nearly 100% up-time without excessive cost.
With conventional NAS or SAN servers, even with RAID protected storage, there are still single points of failure that can take the system down. We know disk drives fail, and RAID operations can protect against a single or sometimes even multiple drive failures; however, that doesn’t protect against controller, CPU, memory or other mother board failures, or power supply combining circuit failures. Network port failover features and power backup systems are helpful, but if the single server crashes, you’re down until it’s repaired or replaced, and sometimes last transactions are lost.
Benefits of a High-Availability Cluster
With a high-availability cluster, if a hardware fault occurs, it is detected via intelligence features and the software seamlessly (and instantly) moves storage requests to the other mirrored server. Smart systems use a cluster of two mirrored servers with real-time continuous data replication and synchro-nization; both servers contain identical data to offer absolute redundancy.
As an example of such a system, the JMR BlueStorTM SHARETM system with HA Cluster powered by euroNAS® supports:
• SMB/CIFS/AFP
• Active directory authentication for CIFS and AFP
• NFS
• iSCSI, with persistent reservation iSCSI LUN and target services
Powerful cluster manage-ment functionality enables the user to define which resource should run on which server; this way, both servers are used in the most efficient way possible – providing the highest level of performance simultaneously. If one server fails, services automatically move to the other server; when the failed server comes back online, its services will automatically