More on the growing recognition in the storage industry that the days of disk capacity constraints are behind us. Of course there will always be a goal to keep disk spending down, and therefore any way to reduce that is helpful...watch the de-duplication activity underway.
But we see an increasing awareness that performance counts in two ways. The first is what we'll call capacity-optimized systems. These represent a new class of storage systems that can scale to very large capacities without suffering significantly on the performance front. Put another way, maintaining and managing that huge library of files is important, and as long as the system scales easily and doesn't slow to a crawl, IT managers are happy.
The other category is performance-optimized systems that focus on I/O operations per second (IOPS) and response times as the main criteria with capacity as a supporting metric. This is an area that cannot be served by current disk-based systems and requires a re-balancing of data center infrastructure to accelerate existing storage. Centralized storage caching, such as solutions from Gear6, deliver the IOPS and response times required for demanding applications while seamlessly complementing existing storage capacity.
Byte and Switch ran a recent story on a new company helping customers measure and assess performance. It opens with the following quote:
"CIOs gauge storage [value] based on utilization," says Michael Passe. "We engineer for performance." In his view, utilization of a SAN can be high or low, but response times and I/O operations per second (IOPS) tell the real story."
This is certainly an area we'll watch closely in the future.