Skip Links

Getting the most out of flash storage

By Mohit Bhatnagar, senior director, flash product strategy, NetApp, special to Network World
May 08, 2013 10:53 AM ET

Network World - This vendor-written tech primer has been edited by Network World to eliminate product promotion, but readers should note it will likely favor the submitter's approach.

As powerful and beneficial as flash storage can be, many organizations are overspending on capacity they don't need, deploying flash in the wrong places, and inefficiently spending on flash applications that are not appropriate for their real business objectives.

There is little argument that flash represents the most disruptive technology to emerge for enterprise networks in recent years. Offering higher IOPS and lower latency, flash-based SSDs reduce the need for storage overprovisioning and require much less power, cooling and physical space than HDDs. And not only are they more operationally cost effective, the actual costs of flash have been consistently coming down, creating a very attractive offer of greater performance at a lowering cost.

[ BACKGROUND: Flash storage in post-PC devices advances ]

Notice that's lowering cost -- not lower-cost. Flash is still significantly more expensive than hard disk drives in terms of capacity. As workload capacities demand higher IOPS, though, many enterprises are finding their workloads are perfectly positioned for flash's cost-to-performance ratio. Still, without a fundamental knowledge of how, where and why flash should be used, these organizations will be experiencing more budget than benefit.

Aside from the direct-attached storage approach, all enterprise flash implementations can be boiled down to essentially one of three type solutions: flash as a cache on the server; hybrid storage, which combines HDDs for capacity and flash for performance; or an all-flash array. These implementations have different strengths and weaknesses, depending on what applications and workloads are running on the network.

There is no "one-size-fits-all" solution for implementing flash. Enterprises need to adopt a portfolio approach that offers support for flash at all levels in the network with the right mix of performance, capacity and economics.

Applications that require a combination of high performance with low-cost capacity storage (for example, file exchange services and OLTP workloads) benefit from using hybrid storage. For accelerating specific applications, such as database workloads, server flash enables low latency and high throughput. Finally, for mission-critical applications that require consistent low latency, high performance and a small physical/thermal footprint, all-flash arrays are emerging as an ideal choice.

The often-touted benefits that flash provides at the hardware level are only part of the story, though. The real differentiator between enterprise flash solutions might well lie at the software level. Companies implementing flash in their networks must be sure that whatever flash hardware solution they use is backed by software that delivers enterprise-class scalability, nondisruptive operation and intelligent data management.

Intelligent data management -- whether to prevent downtime or improve performance -- is particularly important when implementing flash in the network, since having the fastest flash-based solution available is of no use if it is not holding the hot data most in demand by the network. Determining hot data is a real-time operation that needs to be handled automatically at the storage and server levels, without any guidance from the network administrator.

Our Commenting Policies
Latest News
rssRss Feed
View more Latest News