The VMware, Nutanix mud wrestle is hilarious, but which one is crying with fear on the inside?
Hint: It's VMware
Sysadmin blog So Nutanix is in yet another spat with VMware. The big ones are getting to be a yearly affair. That's great for them: these little soap operas seem amusing, but underneath it all, there are some very serious issues being hashed out.
VMware wants control of our data centers. The whole stack, top to bottom. It wants to be the only vendor in your data center making any margin, and everyone else can be commoditized into interchangeable (and irrelevant) suppliers of steel and silicon.
Nutanix wants to make VMware (and all other hypervisor vendors) irrelevant and commoditized, and wants to be the one supplying the software that makes all the margins. But VMware has the money, the very smart engineering bodies and brutal ruthlessness to play this game for keeps.
Nutanix, founded in 2009, takes clusters of servers, turns them into pools of storage, and runs virtual machines in those nodes so they can splash around in all that capacity. It supports VMware's ESXi and Microsoft's Hyper-V hypervisors, and is saying: "Don't get locked in to VMware." I'm paraphrasing, of course.
Meanwhile, VMware is saying, and again I'm paraphrasing, "Nutanix isn't as good as us because it can't offer you our VMware-certified rubber-stamp of approval."
VMware has a point. There is a large gaggle of people out there who simply see a logo they recognize and then disable the soggy thing residing between their ears. The VMware brandname means something.
But do not underestimate Nutanix. It has fantastic people; many of the very best in the industry. Not only that, but Nutanix's CEO Dheeraj Pandey is wicked smart: he knows how to play the game, he knows how to motivate the troops, and he knows how to make a great product. He is not a cookie-cutter startup engineer-turned-CEO who knows how to turn bits on a disk into neat things but can't lead a team to save his soul.
The chains that bind
More to the point, Pandey is right. Brandname recognition means something, but there is considerable pushback against VMware's lock-in. This is seen not only by Nutanix, but by every other hypercovnergence vendor I've worked with. Why – oh please, do tell me why – would the IT industry blindly refuse to learn from our experiences with Microsoft, IBM, Oracle, EMC and Cisco? Lock-in is a killer.
If you're filling data centers with equipment, you should rely on fewer features if it means being in a better bargaining position with your suppliers. This is better than being 18 months ahead of your competitors on a handful of infrastructure ease-of-use features while being stuck with paying through the nose for the privilege.
Is a fully integrated, fully tested all-VMware solution more efficient than what its competitors can build? Absolutely. But by how much? 6 per cent? (Which is about what I see in actually production environments as an efficiency difference between VSAN and Maxta.)
How about 16 per cent? (Which is the difference between VSAN and Maxta I can create in the lab if I really work at it.)
What increase in efficiency is worth the increased cost and the loss of bargaining position? What percentage of efficiency is worth the cost of lashing yourself and your company to an IT mega-vendor that (judging by its vtax and pricing) thinks little of its customers, less of its partners (PEXtanix, among others), and believes it deserves your custom rather than working to earn it?
We've all been down that road before. Microsoft. Oracle. IBM. Cisco. Do we really want to go there with VMware? Do we want to let VMware be the one to build the first true software-defined infrastructure, and then effectively own us until the end of time?
What is VMware terrified of? Let me tell you. And I will
I'm a great believer in the joined-up stack. I really am. I champion the whole idea of software-defined infrastructure as the ultimate goal of the hyper-convergence trend.
But who controls that stack matters. The more integrated the products you buy, the more you have to trust your vendor. Let's look at some hyper-converged suppliers for a moment, and I'll try to explain what I mean by example.
I don't have to trust Maxta all that much. They sell me software that goes on top of a hypervisor and binds the storage in my servers together into one shared storage pool, allowing me to run my workloads on that storage and still get high availability if a node dies.
That's great, that's awesome. So does practically everyone else today. I choose Maxta because it offers me some features I care about – namely deduplication, compression, and multiple hypervisor support – at a price I am willing to pay.
If Maxta get uppity, I simply bin 'em. There are eleventeen squillion hyper-converged vendors, and if Maxta doesn't meet my needs I can replace it with only minor inconvenience. Maxta has to work to earn my custom, and I rather like being the one in control of that relationship, thank you very much.
Nutanix is another company I put in this same category. It is functionally disposable. As much as it won't like me saying this, there is nothing that Nutanix does that I can't find from other vendors as well. If I choose Nutanix – and a huge number of folks do – it's because it's earned that sale the hard way: offering better TCO, ease of use, and a general sense of warm fuzzies than its rivals.
If I am using Nutanix's VMware appliance, and Nutanix makes funny faces at me that I don't like, I can make my next cluster a SimpliVity cluster, and then I'll get the advantage of inline deduplication and telco-grade performance predictability for roughly the same price. Groovy.
If I am using Nutanix's KVM appliance, and I decide that I don't like the look of its bezel, I can ring up Scale Computing and get a KVM appliance delivered, and we're off to the races. Nutanix's OpenStack offering displeases me? Maxta's excellent Mirantis-based OpenStack appliances to the rescue!
Hyper-V on Nutanix not working out? Ring up Gridstore! Decided I want my own non-VMware, non-Microsoft, non-OpenStack hybrid cloud setup made out of unicorn farts, rainbows and Intel's ridiculously awesome erasure coding libraries? Yottabyte will change my religion.
If only one company builds a full software-defined infrastructure stack then our faith in that company not screwing us has to be absolute. If a dozen companies start putting out SDI stacks then we don't have to trust vendors, we can simply walk away if they act improperly.
This is exactly what VMware is terrified of.
When you're just not that important
VMware's investors demand an ever-increasing – and completely unrealistic – level of year-on-year growth. EMC-owned VMware is never going to get there if all these uppity startups keep commoditizing its offerings and driving down margins. VMware needs to own the enterprise data center, and it needs to be in a position to dictate terms and pricing to all the companies with revenue high enough to matter.
While VMware's aspirations may lean towards IBM's genitalia-crushing hold over its mainframe customers, it has some pretty big problems getting to that level of customer base control. Short version: it's just not nearly so important as it likes to think it is.
Nutanix, SimpliVity, Scale Computing, Maxta, Yottabyte, Gridstore, ScaleIO, Nimboxx, Stratoscale, Atlantis, Datacore, Springpath, NodeWeaver and others that I am sure I am forgetting are all hyper-converged vendors. Not only is the hypervisor a commodity (shut up VMware, it bloody is) but VSAN and EVO:RAIL are too.
Server SANs in general, and hyper-convergence very specifically, are not new, they are not wow, they are not even products unto themselves anymore. Hyper-convergence is a feature. A tickbox item, like deduplication or compression. You just have to have them, or you're so far behind the competition it's probably not worth talking to your sales nerds at all.
That's a problem for VMware. A truly frightening amount of research and money went into VSAN. VMware's marketing and social media relations around VSAN burned a whole lot of bridges with its ecosystem partners, its customers and its own hardcore evangelists. And for what? Just another feature?
VMware's epistles about its own importance are full of righteous fury, but they signify nothing. In the time it takes for VMware to research, write and pass through legal the corporate character assassination of a competitor, a dozen other startups will have exited stealth peddling the very same feature VMware is desperate to protect.
If VMware's offerings are so mighty then they will win on merit. If VMware's offerings aren't good enough to win on merit then for the love of $deity, spend the money on R&D! There is no good reason to give platform and voice to those within VMware whose hegemonic advocacy threatens to alienate VMware's partners and customers.
With a corporate attitude such as has been demonstrated, we're never going to allow VMware to be in a position where it is the only possible choice. If we won't trust it enough to let it be our "one true vendor", what's all its vertical integration, single stack and brandname bravado worth?
I'm not saying Nutanix is full of saints. It isn't. It engages in some really awful marketing and social media practices at times. Some of the other startups do too. Sometimes – OK, most of the time – it's like dealing with a group of children with no social skills whatsoever. Except that these particular children collectively have billions of dollars and the influence over one of our planet's more critical industries.
But two wrongs don't make a right, and VMware is big enough that it should be the adult in the room. So to VMware I have only this to say: be the vendor we want to buy from, VMware. Give us a reason to trust you, and to believe that you're better than Nutanix, or any of the other buzzing mosquitoes that flit about. ®