All-flash data centre: roadblocks to reality
Fadi Kanafani, regional director Middle East & Africa at NetApp, discusses all-flash data centres
Within our corner of the storage industry, the idea of a data centre entirely built on flash has been an idea long passed around, built upon and aspired to. Yet it has remained some distance from becoming a reality.
While flash is ideal for programmes and data where fast and regular access is key, it has been seen as an expensive approach to deeper storage and the operation of less mission-critical apps. Now, however, the roadblocks to the all-flash data centre may well be falling away both from a business and a technical point of view. In fact, research house Gartner predicts that while no data centres use only all-flash arrays for primary data today, 25% will by 2020.
These new possibilities for an all-flash data centre begin with the growth of the ‘expectation economy', the idea that the convenience created by technology, particularly smartphones, has led to an expectation by consumers that they are able to access services and information instantly and from anywhere.
This puts businesses under immense pressure to deliver experiences that are absolutely seamless, and creates huge opportunities for those that can deliver on that expectation. Businesses like Uber, AirBnB and Deliveroo are products of and are capitalising on the expectation economy. They've built their businesses around speed and agility, from the experience they provide to the consumer through to their data management operations and infrastructure.
Flash storage provides opportunities for these businesses across the board. It's a reliable and cost effective underpinning for extremely high and intense usage of their apps, but it's also a weightless approach. It's less power and space hungry meaning they can move into a new venue and build out a flash-based data centre with far fewer infrastructure demands, and consequently in line with their rapid expansion.
In terms of basic economics within the data centre, flash is becoming increasingly more inexpensive to buy, eliminating the long standing barrier that the low comparative cost of hard disk drives (HDD) has created. Flash has always represented savings in OPEX - lower power, space and maintenance needs. The two combined increasingly make flash the obvious choice of primary storage.
Technologically, flash, once designed for extreme performance only has seen the continuous expansion and deepening of its capabilities with dedupe, compression, greater scalability, performance acceleration and encryption introduced in recent years. These developments have made flash storage a viable solution for an ever growing number of applications and needs. This area is, admittedly, the one which may hold the all-flash data centre back. There is still a significant number of applications that flash is not considered a valuable option for due to their non-mission critical nature.
And so it seems it is the moment for the all-flash data centre to come to life. In addition to Gartner's predictions of the rapid growth of the all-flash data centre, IDC is optimistic. It recently announced that the worldwide flash-based array market grew to $113bn in 2014. Flash, meanwhile, grew 101% year over year in the second quarter of 2015 in EMEA.
And why wouldn't the all-flash data centre be coming close to reality? I urge businesses to examine its adoption. In any business, flash can bring those technology and business advantages.
For the business, a flash array across the data centre will make every application and piece of data, mission-critical or not, available instantly - no wait times, wasted time or frustration. This, as the expectation economy leads to more applications within the business being viewed as mission-critical - from web hosting to CRM systems, makes reliable performance a must, and only deliverable via flash storage. Add to that the increasing use of virtualisation within the enterprises and each of those mission-critical applications may be running multiple instances, simultaneously - ever more instant access requirement.
At the data centre level, an all-flash implementation will change not just what's in the data centre, but the very face of it. Flash's smaller size and lower heating and cooling requirements could reduce the cooling requirements of data centres and allow enterprises to run their services from far smaller, less specially constructed rooms. They can hugely reduce property costs or simply reallocate once sparsely used but vital space to more productive ends.
Finally, it's green. Flash storage uses significantly less power, consequently it runs at a lower temperature than a typical HDD and needs less cooling technology in the data centre. These factors add up to a significantly reduced environmental impact across the data centre (and of course lower the costs).
With these benefits across the board, the question becomes, why wouldn't businesses migrate to an all-flash data centre? It's smaller, it's cheaper in terms of both OPEX and, increasingly, CAPEX, it's greener, it's ready for your applications of it, and thanks to its reliability and read/write speed across all of the data it holds, it's the ideal platform for businesses looking to embrace the expectation economy.
The only thing holding us all back is the fear of being first to jump in, but therein lies the rub. The expectation economy values speed and reliability, and the all-flash data centre can deliver both far more effectively than more traditional data centre approaches. We may be trepidacious, but the first to adopt the all-flash data centre will reap huge competitive advantage.