When looking good is no longer enough

Last week I attended Storage Field Day 7. One of the companies I met at the event was Kaminario, producer of an All-Flash Array.
This startup has a potentially good product, with its pros and cons, but the question here is: is “good” still enough to have a serious chance of success in this market segment?

A few words about Kaminario

Kaminario was founded in 2010 in Boston with offices in Israel, New York and Silicon Valley. Its objective is to produce next generation storage systems based on flash-memory. The startup raised a total of $143M in funding (last $68M round just a few months ago) and is targeting the general purpose array market with a product that has an average declared price which is below the threshold of $2/GB.

We are talking about a scale-out design architecture starting with a minimum of 1 K-Block (a controller pair with 1 or 2 disk trays) up to a total of 8 K-blocks at the moment. Backend is Infiniband and Front-end is FC or iSCSI.
The list of features is on par with most of its competitors:
– in-line compression/dedupe,
– non disruptive upgrades,
– advanced data services (but no remote replica yet!!!),
– integration with VMware (VVOL will be added soon),
– easy to use GUI, APIs,
– A smart double parity RAID mechanism (which allows a high utilization rate of flash),
– cloud-based analytics and support (___ which I haven’t actually seen in action),
– long term warranty for flash memory.

On paper, this product definitely has all the characteristics to cover most enterprise use cases regarding performance and availability. The lack of remote replication is a downfall, but I’m sure this flaw is only temporary. The K-2 array also has some good benchmarks to show.

Why good, is no longer enough

AFA is a tough market. When you look around you can find plenty of solutions with four of them at the head of the pack (HP 3PAR, Pure Storage, EMC XtremIO and SolidFire). Two primary vendors and two startups. I’m not questioning the single feature of each single product or its potential based on its large sales force, but no doubt these four are already well known, represented and, for one reason or another, are most likely to participate in the majority of the dealings.
Just to give a few examples: in Italy, Pure Storage launched its subsidiary a few weeks ago with 3 people on staff and there is talk already about hiring more to get at the market faster! But the others are very aggressive too. HP has a very good solution and a broad product range, starting at ridiculously low prices (thanks to its 7200 model). EMC has a huge sales force and they already have access to all high-end (aka VMAX) end users. SolidFire is making a dent in the very high end and ISP customers (especially when OpenStack is involved).
I’ve only mentioned the most visible vendors. In some cases solutions provided by vendors like HDS seem to suit their customers’ needs just fine.

How many end users will allow to bring in another vendor? The qualification and testing process alone (involving a PoC, maybe) could be too onerous to justify it. The list of potential vendors is already good and long, why make it longer?

Looking for a differentiator

But storage is storage, and don’t get me wrong when I say this. I completely agree that implementation counts but, on the other hand, basic data protection and data services are implemented quite well on all modern storage platforms. And all the most sophisticated characteristics, like QoS for example, are not needed by all end users.
Now, to find the real differentiation, we need to climb up the stack.

All-Flash or not (Flash memory is just a faster media after all), modern primary storage must be more clever and smarter than in the past. Cloud-based system analytics can make a difference and can help to lower the TCO, but it is only the first step. Some storage vendors are now looking ahead and doing more.
Qumulo, a startup which came out of stealth a few days ago, has presented a smart scale-out NAS solution capable of storing and analyzing files. Nimble Storage came out with impressive features for analyzing VM/App workloads a few months ago. And, Data Gravity is capable of analyzing content of data stored in its array.
These are only examples, but they should give you an idea of what is happening.

Closing the circle

A good implemented array feature affects infrastructure TCO, which is good but it is always difficult to calculate (and to sell, at any level). On the other hand, a good implemented analytics system can reflect its advantages directly on business activities and helps IT to meet the changing needs of an enterprise much faster/quicker. Isn’t this a huge value?

Getting back to Kaminario, at the risk of being repetitive, let me once again say that the product looks good but it’s a little bit late to the party and I can’t find an absolute differentiator that makes it outstanding. The technology is good but what kind of exit strategy is foreseeable for a startup of this sort? Acquisition most likely, or living in limbo for the time being. Time will tell.

Disclaimer: I was invited to this meeting by GestaltIT and they paid for travel and accommodation, I have not been compensated for my time and am not obliged to blog. Furthermore, the content is not reviewed, approved or edited by any other person than the Juku team.