Storage Insiders
Get the latest thought leadership and information about HP storage products and solutions at HP Blogs

The flash storage battlefield – spilling the beans and the blood

Flash – now we’re talking

I’ve resisted commenting on the flash storage topic for some time now, even as the hype in the market reaches a crescendo. But now the time is right for me to get some things off my chest and set the record straight. Not least because many vendors – start-ups and established players alike – have had their say for far too long. I’d like to provide a more pragmatic and balanced view of the current flash landscape, what the future is likely to hold and most importantly the pitfalls customers need to avoid.


What’s the big deal with flash?

Flash provides very high performance storage (lots more IOPS and significantly less latency), using substantially less power.  The technology is not new, and last time I counted there were 24 companies with an offering! The drawback however is that the cost of AFAs (All Flash Arrays) has, so far, limited deployments to organisations with demanding applications and deep pockets.


But the continued explosion in the volume and variety of information means that we are seeing a dramatic increase in cloud data centers, and the associated power consumption it takes to run them. The stark reality is that there is simply not enough power in the world to sustain traditional spinning disks with the volumes of data we see coming…wow! That is why flash technology will soon become the norm and not the exception.


The players

The market today has three types of vendors:

  1. Venture capitalist funded start-ups who have one of two goals – to be acquired or IPO.

  2. Established storage vendors who have acquired a Flash platform and bolted it onto their existing portfolio

  3. Established storage players that have developed a flash approach within their existing architecture

The start-ups have been making a lot of noise in the market and are using their VC dollars to effectively ‘buy’ customer installations and get themselves noticed. This is typical start-up behaviour designed to lead to an acquisition. But there’s a problem – if the established players have already acquired or designed their own flash technology, why would they need to buy a start up? The technology in many cases is all industry standard hardware – there’s nothing unique in the boxes themselves that cannot be easily copied or adapted. So that leaves IPO.

To pull off a successful IPO a start-up need to establish themselves as having a decent amount of market share and/or unique intellectual property. We’ve already discounted the latter, so that leaves customer acquisition, and with the number of other start-ups and ‘me toos’ in the market, that is going to be one tough ask. Let’s not forget that one of the first flash start-ups to IPO did so at $9 per share – giving it a market valuation of $162million. Within 4 months the stock had crashed 70% and the CEO had been fired. Whichever way you cut it, that does not bode well for the start-ups who choose to take a similar path.


Let’s not forget the companies who acquired a flash offering, who have also struggled to make their mark. One particularly high profile failure being the company that took eighteen months to launch its platform following the acquisition. Not exactly striking while the iron is hot, and as a result a recent industry survey ranked them a lowly 18th in the AFA space.


Saving the best until last

In contrast to all of the above, there are companies out there who continue to release enhancements to their AFAs at a consistent and exciting pace. They are able to do this because they have built their flash offering as an integrated element of an end-to-end primary storage portfolio. They are also able to offer the high availability, disaster recovery, management and reporting tools and security that enterprise customers want.


But the most exciting news of all is that it is now possible to take the cost of flash storage below that of traditional spinning disk – essentially offering flash-optimized enterprise-class performance, scale and resiliency at disk cost per gigabyte.


In the context of the storage industry this is a first and significant milestone on the way to delivering optimised storage to cope with the vast quantities of data and managing the power consumption that goes with it.

And that is something no start-up can even hope to offer - now or in the near future.


If you’d like to find out more, check out this video: HP 3PAR and Flash – made for each other?


You can find me on Twitter @ChrisJohnsonHP if you wish to comment...

nate | ‎06-23-2014 05:33 PM

As Calvin loves to remind me I don't use tweeter so I can't reply to you there.


You forget another segment of the market - the hybrid flash market. Not tiering, but true hybrid. Similar to Nimble, Nuatix(sp), Simplivity(sp), Tintri, probably some others too. At least I think they are all hybrid I have no personal experience on any of them.


With exception to Tintri I believe the entry level pricing for the others is quite competitive for the low end of the market, a technology that HP doesn't (yet) have in their portfolio. My company has one or two Nimble systems in internal IT. They run some VMs in another location I don't have anything to do with them. All our mission critical stuff is 3PAR though.


The hybrid systems may not be good enough for true mission critical type stuff, but they do serve their market segment well I believe.


Those folks obviously are not as competitive(or competitive at all) at the upper ranges of the market, but if your in the market for something that is say $50k or less with many TBs of storage then these can be worth looking into depending on the workload. Tintri I think starts at around double that last I heard anyway.


I am planning on buying at least a 2-node 7450 if not a 4-node in the next month or so(3PAR customer since 2006), so can say with personal experience the entry level price point of such a system is still well beyond that of the aformentioned systems.


Real time compression, when 3PAR gets that will certainly help in this space too, as not all workloads are de-dupe friendly.

Showing results for 
Search instead for 
Do you mean 
About the Author

Follow Us
The opinions expressed above are the personal opinions of the authors, not of HP. By using this site, you accept the Terms of Use and Rules of Participation.