Half way through January is always a good time to take stock of the flood of 2016 predictions that have
crossed my desk in the last 3 weeks.
With the laser focus on Flash Storage Technologies, there were two that resonated with me. The first being
the prediction that the All Flash Datacentre will become a reality.
Less than two years ago I was speaking about the impact of TLC Flash. At the time there was doubt that this technology
would become a mainstream enterprise capable storage technology. NetApp were one of the first to be quoted
in the IT Press regarding the endurance challenges that TLC Flash presented. At the time, minimising Flash writes was
and remains the key, while log-structured files systems in both the SSD Flash Translation Layer (FTL) Firmware
and the WAFL file system presented a stack which minimised writes from the host OS right through to the Flash Media.
The management of the write process throughout the stack brought the additional benefit that the overprovisioning
of Flash capacity in the SSD, to cope with endurance challenge, could be significantly reduced. The benefit being
higher density delivered at reduced costs. Today TLC Flash is here to stay with new ways being constantly proposed to
squeeze the last ounce/gram of endurance and performance out of the technology
Now, less than 18 months on, we are at the same point again. At the Flash Memory Summit, held every August in Santa Clara,
I noted several sessions on QLC, Quad Level Cell Flash Technology. This doubles the density over TLC from 3 bits per cell (8 numbers) to
4 Bits per cell (16 numbers). Once again there are doubts about endurance because discriminating between the 16 voltage levels
in the cell leaves no room for error. However we are some way off QLC becoming commercially (as we were with TLC 18 months ago !)
viable but the forecast is that this technology will apply well to archival workloads.
Which leads nicely to the second prediction – Flash deployments will expand far beyond performance applications.
Let’s face it, in ten years time Flash, or Flash like semiconductor storage technologies, will be the de facto
standard. Today, however, the sweet spot Flash workloads are tier 1 Database, Virtualisation/VDI and Analytics.
It makes perfect sense (too me) for Flash to meet the demands of archival workloads. Having worked in the
supercomputing world for many years where data volumes are huge I can think of plenty of WORM,
(Write Once, Read Many), applications where Flash could, and will, play a role. Other areas to consider are
Object Based Archiving and Big Data, CRM, Data Mining / Warehousing, Email and Block Storage
In short we will see Flash storage technology eat into tier 2 and 3 storage workloads in the
coming year while the rise of Development and Operations (DevOps) heralds a new era in rapid
application development culture.
My prediction is that, in the coming year, Flash becomes eligible for an ever increasing
number of workloads and applications. This will drive the adoption of the All Flash Datacentre but
lets be clear, fast is good, but the requirement for rock solid data management and efficiency do not change.
In fact, if anything they are of higher priority today than at any time in the past.
Yes the All Flash Data Centre is a reality and I can see it from here !
If you’d like to read more NetApp 2016 predictions please look here 2016-Year-of-Simplicity-in-IT .
Or you can link to this Blog Post from Tara Bal
Laurence James – On behalf of the EMEA Products, Alliances and Solutions Marketing Team
Follow me on Twitter – @lozdjames