How analytics can work to make data storage smarter than ever

Storage systems are designed to provide optimized solutions for specific application workloads and use cases, say experts, so it’s important to design big data solutions appropriately for real-time analytics access patterns.

Jeff Rowe | Dec 19, 2017 12:00 am

In the past, data storage was kind of dumb.

Perhaps not the most flattering description, but Drew Robb, writer at the EnterpriseStorageForum, floats it as the beginning of his case that the combination of new storage technologies, including flash, is teaming up with ever-expanding analytics capacity to give healthcare and other data users new opportunities for greater insights.

But how do go about it?  That is, how should IT managers approach their storage needs in order to maximize analytics opportunities as well.

First of all, says Robb, define your needs. Better to figure out your storage needs and lay out the most applicable platforms before the selection process begins. Or, as Anoop Dawar, vice president of product marketing and management at MapR Data Technologies, summed it up for Robb, “Define your must-haves and desirables clearly before evaluating solutions.”

As for some technical specifics, Robb points to NVMe as one key step toward helping managers get the most out of the evolving relationship between storage and analytics.

“NVMe brings the necessary speed to data transport mechanisms closer to the velocity of modern processors and flash architectures,” he explains, pointing out that products already on the market that incorporate it “run up to six times faster when it comes random and sequential read/write performance. Although we are still in the early stages of its evolution in storage, it is fast becoming an industry standard, and it is being added to switched fabric. Developers are figuring out ways to utilize it throughout the entire storage array stack.”

According to Clodoaldo Barrera, chief technical strategist, IBM Storage, “analytic applications are becoming mission critical, and there is a need for the business continuity functions that storage provides for database environments, including snapshots and synchronous and asynchronous data replication.”

In general, Barrera said, as they move forward “astute IT teams will get out in front of application demand by identifying the analytics tools desired, and providing a managed storage environment that complies with enterprise needs but still allows data scientists the experience they want.” 

Or, as Anoop Dawar put it, “Make sure the solution was built ground up with analytics in mind and not added into the solution as an afterthought.”