Has Flash in Server Peaked?

Here is is an abstract of a report (80 pages) by Arun Taneja, founder and consulting analyst, Taneja Group, on the topic of Storage Acceleration and Performance Technologies.

Has Flash in the Server Peaked?

The flash storage market is booming, and one approach in particular-server-based flash-has led the pack in user adoption. In a market pioneered and largely developed by a single enterprising vendor, server side flash has achieved significant mindshare among IT practitioners in the space of a few short years. PCIe SSDs have become commonplace in servers, and are being used to accelerate a whole host of performance-sensitive applications, from databases to virtualization to cloud.

But another type of flash deployment – the hybrid storage array-may soon knock server – based flash off of its lofty perch. The end user research, which included 280 online respondents and a number of supporting, phone interviews, reveals that hybrid arrays are now being deployed at nearly the same rate as server side flash solutions.

Looking ahead over the next couple of years, more than one-half of survey participants are planning to deploy solid state drives in the array, far higher than the plans indicated for pure, server-side flash storage.

Does this mean that server-based flash has peaked?
Actually, far from it. Server side flash remains the most valued type of deployment, edging out array-based approaches in this category. And the growing number of server side caching solutions are attracting considerable user interest. Given their reasonable cost, ease of deployment, and significant boost to performance, server-based flash cache solutions are achieving rapid adoption across a diverse set of industries and application areas, promising to keep this market hot for many years to come.

What’s driving users to purchase and deploy
flash technologies in greater numbers than ever before?
According to survey participants, virtualization and cloud initiatives are creating pressure points on storage performance, increasing application response times, limiting infrastructure consolidation efficiencies, and making it difficult for IT managers to meet SLA commitments.

Nearly 40% of respondents said that constraints on storage performance are limiting their ability to scale server virtualization environments, and 25% pointed to storage performance issues in private cloud initiatives.

With these growing pressures on storage performance,
many of which are now affecting critical applications,
what are IT practitioners willing
to give up to address these issues?
Surprisingly, very little. Nearly 30% of respondents are unwilling to give up any storage features or capabilities, even if they are faced with an urgent performance problem. Only 17% of users are willing to compromise ease of installation or use of their storage solutions, and less than 25% are OK with giving up a single point of storage management.

Far fewer numbers are willing to compromise performance or capacity scalability. This suggests that vendors must bring to market storage acceleration solutions that fully complement and enhance users’ existing storage infrastructure, rather than disrupting or detracting from the capabilities already in place.