Adoption Of Clustered Storage Ramping Up

In the past, the primary driver behind the adoption of a clustered storage infrastructure was the ability to improve operating performance. As it becomes increasingly necessary for mission-critical data to remain up and accessible at all times, I've advised my customers that clustering's ability to enable non-disruptive operation (NDO) is more important than ever before.

While many people still operate under the assumption that unplanned downtime is their biggest worry, today's business technology leaders increasingly understand that planned downtime -- for upgrades, routine maintenance, or to scale -- is not only more common but also far more disruptive.

As IT-as-a-service becomes a central business driver, organizations find they must contend with the challenges of providing consistent and constant service to clients and customers while operating with a shared infrastructure. This has become an increasingly difficult challenge with the large amount of virtualization used to consolidate services on fewer and fewer devices. The result is an exponential increase in the difficulty of scheduling downtime to upgrade or modify those devices.

For some customers, scheduling downtime is not just difficult, it is nearly impossible. They are often forced to postpone service upgrades until maintenance becomes absolutely critical. It often requires far more effort -- and resources -- to coordinate the downtime than to execute the actual event. In the interim, they are stuck operating with reduced capacity and/or hampered performance levels.

Many of these customers don't realize NDO is even possible. Until they are shown how a clustered storage infrastructure will allow them to move and redistribute data without limiting access, many assume that downtime is just an unpleasant and unavoidable business reality.

Some clients assume that the cost of clustered storage outweighs the benefits it offers. When I do the math with them though, we often find that clustered storage offers significant savings. For example, rather than deploying a single high-end controller to operate on a workload, it can sometimes be far more efficient to add multiple mid-range nodes that are less costly and cluster those nodes together. With this approach, I have seen savings of as much as 60 percent on capital expenses while also improving performance and enabling NDO.

That said, although a clustered environment is sometimes compelling, it is not always the most appropriate data storage solution for an organization. I always advise businesses to assess their actual business needs to see if clustering is right for them.

Here are a couple of -- highly simplified -- rules of thumb for IT managers weighing the adoption of clustered storage. First, for organizations that employ workloads that generate a high ratio of CPU computations relative to the amount of data, scaling out with a clustered storage environment is often more efficient and more cost effective, because it drives down the cost of the CPU resources. If an organization has the inverse ratio -- a large amount of data, but few CPU requirements -- it could be more likely that scaling up will work better.

Second, for organizations that have highly shared infrastructure where availability is critical, the NDO features of clustered storage can also be particularly advantageous. Communicating pain points and business needs to a trusted solution provider partner will allow their engineering teams to design a system based on which workloads function better on a horizontally scalable solution as opposed to a vertical one.

As explosive growth of data now impacts more organizations across more business segments than ever before, the benefits of clustered storage have both dramatically increased and become increasingly affordable. It's not always the most appropriate platform for all businesses -- no technology ever is. But, as more companies learn about clustering, and how it could work for them, I've seen that many are pleasantly surprised to discover it is an affordable and extremely powerful solution to their data management challenges.

Rob Cotter is chief technology officer for Insight Investments. He has more than 20 years of experience in storage hardware and software, networking, systems architecture and engineering.

PUBLISHED JUNE 20, 2013