By Evan Unrue, Chief Technologist, IoT, Analytics & Cognitive EMEA
With new innovations in the silicon (with technologies such as Triple Level Cell and 3D NAND Flash), achievable capacities will be increasing and associated cost are going to be driven down further to the point where it is widely believed that 2017 will be the year where price parity is achieved on cost per gigabyte with SSD, depending on the type of SSD.
When we consider the variety of SSD types in the market from high performance SSD PCI-E cards to less performing higher capacity SSD drives, expect to see a tiered approach to Flash storage get adopted in the all flash datacentre. Flash is not one size fits all, but we are not limited to one size.
Companies that adopt flash for tactical solutions to performance hungry applications will find that the headroom in their systems allows them to move from the focussed approach to flash and they will expand its footprint into general workloads over time and not necessarily a drawn out period of time. When the gains of flash are staring a company in the face, they will move quickly.
It is also important to note, that when discussing these gains, we are not just relating this to large enterprise. The small to medium business end will no longer see the barriers to entry on cost. Even today, this barrier is getting smaller and smaller.
We have also seen that Flash has been inhibited by aging protocols and computer architecture, and there is a visible evolutionary shift underway.
The next phase is starting to happen now, which is replacing traditional storage protocols with NVMe on the network. This allows an organisation to get the full force of SSD performance in a shared storage environment over a fabric delivered over technology we already know such as a fibre channel network. This is unlocking SSD performance in a way that will change the way companies deal with storage at scale forever, especially in the world of big data and the mass of data needing to be analysed in real time, driven by technologies such as IoT or Financial Trading Systems.
As exciting as NVMe technology is, it is really a necessity to ensure we’re not curtailing the innovation of flash technologies.
The revolutionary piece, is what’s next with the SSD drive itself. As SSD gets faster, the line between what is memory and what is disk gets thinner, to the point where all we are left with is storage persistence as the key difference. Enter Intel and Micron, who between them over the last 10 years have been developing 3D Xpoint technology, now packaged as Intel Optane, changing the way that bits of data or stored with SSD completely and in essence SSD becomes RAM with persistence of data or non-volatile memory (exactly as the term NVM in NVMe implies).
With vastly improved storage capacity and early quotes stating it will be 1000x faster than NAND Flash and 1000x more reliable, but Gen 1 quoted at being 5x faster and 3x the reliability (still very impressive), the future is moving to a computing architecture where there may not a requirement to have a separation of storage and memory which will advance our ability to compute order of magnitudes greater than even imaginable today.
The implications of this in the finance, life sciences, the datacentre, and in our homes represent at true milestone in what many refer to as the latest industrial revolution through digital transformation.
The bottom line is Flash is here. It is not a flash in the pan technology it’s getting bigger and will be the standard for enterprise storage.