It’s about time for a contrarian viewpoint around Hyper Convergence, and what better place to put it forth than from one of its prime cheerleaders (me). With EMC entering the Hyper Converged marketplace with VSPEX:Blue, and the impending releases of the major VMware OEM partners and their EVO:Rail solutions, I thought I would take a hard look at the state of the Hyper Converged market 5 years on and try to give a fair analysis of the market in terms of where it’s being successful and where it’s not.
Disclaimer: I have no proprietary information to share here, all of this comes from my observations of public statements made by various customers, and vendors in this space.
OMG THE TAM IS HUGE!
Nearly all of the Hyper Converged vendors will pitch that the TAM (Total Addressable Market) is huge. The three claims by the analyst community that I have seen are Gartner: 6 Billion by 2014, IDC: 17 Billion by 2016, and Forrester 40 Billion by 2018. That’s a pretty big stretch and an amazing growth rate if it turns out to be true. The reality on the other hand is something totally different. Let’s take the recent IDC analysis of the Hyper Converged market. This graph has Nutanix clearly in the lead, a point that I find to be 100% valid. It also has them with roughly 5X the sales of its nearest competitor SimpliVity and Scale Computing, which I would also say is valid. Let’s face it, they have a good head start on everyone else in the market place having had product shipping since roughly 2011. Nutanix is claiming a run rate of business of around $300 Million.
Do you see the rest of the players in this space accounting for the other 5.4 Billion that Gartner claimed would be the TAM for 2014? No, at best you can equate the rest of the markets combined run rate of business at roughly $500 Million per year. Now for an emerging market that’s fairly impressive, but lets put it into a different context, 500 Million is roughly the entire Fibre Channel HBA market world wide. For all the claims that Hyper Convergence is taking over the data center, one has to ask how is that so if these companies, who are pioneers in the space, are not taking off in terms of sales that would make a dent in the multi-billion dollar Server/Storage market, let alone eclipsing a legacy storage transport market?
Now to be fully fair, I’m pretty sure that the analyst firms are bundling in the standard Converged Infrastructure players into this market space as well. The devils always in the details, and if we were to include VCE/FlexPod/Exadata/PureSystems etc into the mix that 6 Billion number looks a lot more appealing than my estimated 500 Million. VCE hit a run rate of 2 Billion this year per their claims. For those playing at home, thats 4X the entire Hyper Converged market today.
So Easy A Caveman Could Do It.
Hyper Converged vendors are most certainly first and foremost selling Simplicity. Nearly all of their sales pitches discuss how simple it is to deploy, manage, and configure (sounds like a VMware course title), and from what I’ve seen they deliver on that promise. I took the EVO:Rail challenge at VMworld Europe and knocked it out in 15:06, that’s a full 6 seconds past the 15 minute claims by VMware for reading the document and doing any error correction in the data entry portion. So whats the problem with that? Well all of the Hyper Converged systems are Channel Sales driven, that means VARs (Value Added Resellers) are you go to market vehicle that bolsters your sales team. None of these companies are taking the business direct, even EMC is saying that VSPEX:Blue will be 100% channel.
Riddle me this, if I’m a VAR, and I already get slim margins on Hardware Sales (anywhere from 3-7%) where do I get to make the bulk of my money? That’s right, Services. What happens to all those billable hours that I used to charge for installing and designing the “legacy infrastructure” for my customers as their “Trusted Advisor” when I sell them a Hyper Converged system that they can have up and running in 15 minutes on their own?
This is a question I’ve posed to many people and no one seems to have a compelling answer so far. I know there are huge margins in the Hyper Converged space. These products are not cheap, but the OEM’s are the ones making the bulk of the money here, not the Channel partners who are selling it. Sure there will be SPIF’s and incentives to push product, that’s the nature of the beast, but what about those sweet, sweet, billable hours that are so much gravy ladled over the IT sales process? I’m sure there is a good answer out there somewhere to this question, but so far I’ve not heard a compelling one, and in my own experiences one of the first questions that arrises during the initial conversation with VAR’s I was working to recruit was: “How do I make up for this lost revenue?” Good question. Anyone have a good answer?
The Mis-Match Game
The last point I wanted to touch on was what I call the “Mis-Match Game”, how do the Hyper Converged players address the mismatch between Storage and Compute ratios that invariably will appear with an all-in-one approach? How do customers still find a way to leverage their investments in the legacy stack that they have depend on for the last two decades? Some of the vendors in this space attempt to do this by opening up the Storage functionality to external compute, but that will only let you scale compute resources. The challenge ends up being what if I want just a little bit of storage, or a whole bunch of compute?Certainly for some IT Shops it will not be a major concern if I have to scale one additional node, but what if that scalable unit is 4 like it currently is with EVO:Rail? Invariably there is a mismatch and customers end up buying more of what they truly need and as an industry, we have been pitching “right sizing” of environments since the dawn of time. This can be a fairly hefty financial impact for a target market that is currently geared towards the SMB/Mid-Market customer whose IT budgets are fairly tight to begin with.
This poses a significant challenge to customers and vendors alike, and to be honest, I find some of the arguments against this point to be very weak. What the customer makes up for in reduced complexity, they lose in terms of a very rigid infrastructure model that is not well adjusted to unforeseen change or mutation. I’ll tell you that speaking from my own experiences, the holy grail of an “All-in-One Datacenter” sounds great on paper to many customers, but it doesn’t always end up being so elegant in practice. As this space continues to mature, it will be interesting to see how the current crop of vendors attempt to address this point.
So which way do we go with Hyper Convergence??
None of above is a terminal to the Hyper Converged space. To the contrary, this space didn’t even exist 4 years ago, it’s starting to build an attractive run rate of business, and there is significant validation of the Hyper Converged concept by the entry of EMC, HP, Dell, etc. who are always looking to increase their bottom lines.
Let’s be real here, it takes time to change 20+ years of how IT has done business and it won’t happen overnight. I simply offer these examples as check points against much of the Hype that I see every day. And while it’s easy to get caught up in the echo-chamber of the influencer community to which I belong and play, customers tend to have a lot more riding on their infrastructure bets, namly their jobs. So they are going to cast a far more skeptical eye on these solutions, and in turn will be voting with their wallets.
Coming soon: Moving Beyond Hyper Converged, Rack Scale.