Clustered Storage for the Virtual Data Center

Add clustered storage to the list of upcoming technologies rising to meet the challenge of the virtual data center. As increasingly busy servers demand more data through high-speed networking devices, clustering disk drives so they behave as a single entity can be the quickest and easiest way to leverage existing storage pools for virtual environments.

Isilon Systems, which made a name for itself in clustered-storage systems, has released its latest-generation product, the IQ X-Series, which scales up to 1.6 petabytes with 10 Gbps throughput. That's unprecedented for a single-file, single-volume system. The X-Series uses the Xeon 5130 and the company's OneFS operating system found in earlier-generation IQ clusters.

The importance of clustering isn't lost on larger industry vendors, either. IBM recently shelled out an estimated $350 million for Israeli start-up XIV, maker of the block-based Nextra storage system, and folded it into its Systems Storage unit. Late last year, HP also bought its way into the market with the acquisition of PolyServe.

Clustered storage could be an attractive alternative to some of the more esoteric enterprise solutions gaining attention. Tech blogger Robin Harris wonders whether some of the benefits of cloud-computing proposals from Google and Microsoft might not be more easily attained with an active clustered system. And now that networking solutions are rediscovering the benefits of distributed architectures, there's some question about whether clouds have a future at all.

Of course, when it comes to the future, it always pays to see what the high-power computing industry is up to. And apparently, that crowd is already moving past clusters and heading into parallel storage. It seems that with multi-core processors working in parallel on extremely large data sets, going parallel offers the best way to meet storage demands, particularly in modeling and simulation applications.

At the moment, however, most typical enterprises probably will do well with a clustered solution. With the rest of the network kicking into high gear but with drive performance maxed out, it seems banding together provides the best way to keep up.