Flash Storage Architecture: What’s Available and Why It Matters
Thanks to the advent of faster NVMe host controller interfaces, the way IT systems are fundamentally designed is likely to change in the months and years ahead. One of the first examples of this shift came this week at the VMworld 2016 conference in the form of an IntelliFlash Cloud Platform launched by Tegile Systems.
Taking advantage of high-speed memory-to-memory transfer enabled by NVMe, the IntelliFlash Platform employs eight controllers that share a global name space, says Rob Commins, vice president of marketing for Tegile Systems. The result is an ability to connect tens of petabytes of solid-state disks to a system that eliminates all the network overhead associated with traditional scale-out storage architectures, says Commins.
Commins says that the Tegile Systems approach, scheduled to be available in the first quarter of 2017, will also serve to reduce the cost of Flash storage to about 30 cents per gigabyte.
Most IT organizations these days are adopting a Flash First mentality when it comes to primary storage. Not only is Flash storage orders of magnitude faster than traditional magnetic storage, the overhead associated with managing Flash storage is considerably less. Add in the space and energy savings and making the move to Flash storage at this point is a no-brainer, assuming the organization has the available budget to acquire these systems in the first place.
Less apparent at the moment is what downstream impact the widespread adoption of Flash memory technologies will have. As compute and storage become denser, the amount of physical space they require in the data center winds up being dramatically reduced. There may one day be applications that can invoke all that IT infrastructure horsepower. But for now it would appear that IT organizations are on the cusp of a major consolidation of data centers that, in terms of physical space at least, will soon be a shadow of their former selves.
