The whole point of hyper-convergence is to get toward the all-important "infrastructure in a box" -- the one SKU that provides all you need to stand up infrastructure in your data center. The goal is to have the least amount of configuration, the least amount of licensing hassle and the least amount of restrictions possible.
The problem with VMware in this space with its EVO:RAIL offering is that this entire approach is, at its core, anathema to VMware. VMware doesn't sell "one SKU that does all things" anywhere in its portfolio. If VMware builds or acquires a major piece of functionality, then it breaks it out into a separate SKU and charges yet more licensing for it.
This is the way things worked for top-dog vendors for years, but this adamantly isn't the future of the data center. The future of the data center is not in infrastructure at all, but in the applications, services and multi-pronged solutions that infrastructure enables.
The biggest example of completely missing the point is the utter failure of the VMware Solution Exchange, formerly the VMware Virtual Appliance Marketplace . The Solution Exchange has been around in one form or another for ages, but VMware has consistently failed to capitalize on it.
VMware doesn't care what you put inside your VM. They are focused on building the bits that let you run VMs. They haven't figured out that all the various components around that VM are rapidly commoditizing.
The first vendor with a truly easy-to-use infrastructure solution and who is at least middling on price will win the data center if they have a good app store. Microsoft's Azure Stack is the standard bearer here, but plenty of other hyper-converged vendors are getting in on the game.
Consider Nutanix for a moment. Nutanix can provide appliances that run the VMware, Hyper-V or KVM hypervisors. Nuantix's Acropolis management tools will support managing nodes running all hypervisors from a single interface. Nutanix has even demoed a tool for easily moving VMs between hypervisors.
It's not a stretch to see the barrier between hypervisors vanish in the next year or two. Throw in an actively supported app store with Nutanix's marketing muscle behind it to attract developers, and they become fearsomely difficult to compete against.
Enterprises could run Nutanix clusters with multiple hypervisors, using policies to ensure that workloads requiring a specific hypervisor for support or regulatory purposes operate on that hypervisor while other workloads move to the lowest cost provider. New applications can be deployed at the click of a button from the app store.
Need more I/O speed? Add a flash-heavy node. Need more bulk storage? Add a magnetic disk-heavy node. Need more RAM? Add a RAM heavy node. So on and so forth.
Nutanix isn't the only vendor in the hyper-converged space working toward this goal, but they are the furthest along. VMware, sadly, still seems to be in the process of debating whether or not such a vision of the data center is desirable at all. Every move towards that sort of flexible, heterogeneous future seems to occur only because enough competitors got there first that VMware for forced to drag themselves forward or be left behind.
Adapt or die
Data center infrastructure matters so much that a number of companies have popped up with the express goal of making data center infrastructure matter not at all. This invalidates VMware's entire reason for being. To survive, VMware needs to accept that everything from the hardware to the software required to stand up an entire data center's infrastructure is no longer a product -- it's just a feature of a larger service.
Does VMware have a flexible enough corporate culture to evolve and survive? Moving between hypervisors is getting easier and easier. Competitors are making better management tools than VMware itself. Getting a VM lit up takes little effort, and it will get easier still in the years to come.
VMware simply can't keep charging what it does for the software required for its proprietary, single hypervisor infrastructure. If it is going to survive, then it needs to find a way to make money off the applications that people are building on that infrastructure instead.
VMware needs to leverage its lead in hybrid cloud technologies before its competitors close the gap. If Nutanix gets as good at hybrid cloud as VMware, there may be no stopping them -- and Microsoft can already match or beat VMware on every front in the hybrid cloud wars.
VMware can ride on its name for some time to come. They aren't going to fold up and disappear in a puff of smoke tomorrow. Their resources are enormous and when they choose to do something they can do it in a right hurry.
The question is: Does VMware see the train coming? Will it move aside and choose a new path? Or will it fade to black like so many IT behemoths before it?
Fixing the VMware Solution Exchange
The VMware Solution Exchange (VSX) has failed to see much investment from VMware. In a perfect world, VMware would devote substantial resources and actively solicit applications to live there. But the VSX offerings are paltry compared to what exists in similar marketplaces for Azure or Amazon.
With more development, VMware could then treat the exchange like a proper app store. Software vendors could ply their wares there, and VMware could take a percentage of sales off the top. In exchange, VMware provides promotion, certification that the application will work -- to some degree or another -- on VMware's hypervisors, and some form of assistance to developers to make the applications easy to deploy.
This could allow traditional applications to be closer to the software-as-a-service (SaaS) model. It could also allow SaaS developers to offer versions for use in on-premises solutions, expanding the potential addressable market for those developers without a lot of extra effort.
As these are virtual appliances, they are functionally "black boxes." The destination systems administrators, developers and end users wouldn't have to care much about how they work, just so long as they are certified for the existing version and integrated into the self-service UI.
As it stands now, the exchange just isn't ready. It features shoddily constructed virtual appliances that need a lot of work to get to an operable condition, many of which need that TLC through a console window, rather than being able to log in through a Web interface and go forth.
This places two difficult constraints on VMware. They need to both raise the bar on quality but also on the quantity of submissions. Typically one precludes the other. Unfortunately, if VMware can't manage this, they may well lose out to rivals that have established app stores for their public, private and hybrid offerings and are continuing to invest heavily in expanding them.