Flash: New Wine in Old Skins?

By | October 28, 2013

new_mid-range_systems_Flash_Runner_HighResIs Flash the disruptive technology that the plethora of startups claim it to be? or is it merely another (albeit more performant) form factor for storing data?

There’s much to be said on both sides.

The argument that I heard at VMWorld from an ex-colleague who left NetApp eighteen months ago to join one of the new Flash startups but then to leave within 9 months, is that the window of opportunity for these new breed of players – especially the all-flash array vendors (AFA) – is very small. The established storage vendors like NetApp, IBM and EMC are getting their act together and announcing or shipping Flash systems which compete with these new guys. In fact NetApp and indeed EMC have been shipping Flash-based technology for the last five years, way before these startups were even “started up”.  Indeed, NetApp has shipped more than 44 PB of Flash (which is way more than these startups) and delivers it at the three layers of the stack – server, controller and disk. Along with our EF540 AFA, NetApp has the richest set of Flash storage technologies out there. Today more than 50% of our shipments contain some Flash, and there is no doubt that it is becoming a “must-have” specification – at least at some level of an IT infrastructure. For my post on the NetApp Flash strategy – see here.

But, is it truly ‘disruptive’ as is claimed? Well it certainly changes the performance dynamic in serving up data. No question that the difference is akin to that seen when we moved from tape to spinning disk for our tier one and two applications 20 years ago.  It is an order of magnitude faster and for applications that require lightening speed-response to data delivery such as say a trading application or a credit card authorization application, the gains are certainly impressive. Certainly there are customers who are benefiting enormously from this technology.

So why hasn’t Flash become the defacto and why isn’t disk going the way of tape any time soon? After all, it has been around in some commercialized form for 5 years or more from storage vendors.

Firstly it is a matter of cost. Although the price/performance is getting better, it is still 10 times more expensive than spinning disk. Yes there will be plenty of use cases where an AFA makes sense. However these will remain in the minority whilst the price/performance dynamic is not weighted in its favour. Secondly, delivering an array full of SSDs is one thing, but managing it, writing to it, handling the aging of the SSDs etc is actually pretty complex and is not necessarily the same as for spinning disk. That means clever storage software is required – it isn’t as simple as just pointing to a SSD LUN and hoping for the best.

We’ve seen lots and lots of startups in this space – from the hybrid array vendors like Nimble Storage to the acceleration vendors like Fusion-IO through to the AFA vendors like Pure Storage and Violin Memory. It has been said that everyday in Silicon Valley a new storage startup is born. Clearly not all these vendors will survive. As the time of writing Fusion-io is struggling with the demands of being a public company and with executive turnover and slow growth. Violin Memory carried out its IPO last month and is currently trading below its hoped-for offer price of $9, which can’t make their employees, customers or their shareholders particularly happy. Nimble Storage has filed for its IPO and we wait to see how that will fare. Whiptail, another acceleration vendor has been snapped up by Cisco, avoiding stock market subjugation and quarterly pressures. Other larger vendors have purchased other Flash startups.

As the market settles down over the next few years, we will start to see consolidation and aggregation. It might feel like a Flash bubble is forming were it not for the relatively low valuations that the Street is placing on these startups.

There is no doubt that the world of storage has gotten a lot more entertaining and more vibrant in the last couple of years. Along with Software Defined Storage and Hybrid Clouds, Flash certainly has the capacity to be truly disruptive.

Whether the proliferation of startups vendors is the indicator of a disruptive technology only time will tell. What really matters is how customers choose to adopt flash and how they report on the benefits they are achieving. Are they able to do something truly different as a result or is it just a little bit faster for a little bit more money – a sort of new wine in old skins?




Leave a Reply

Your email address will not be published. Required fields are marked *

+ 3 = ten