The Cranky Admin
Surveying the Field of Private Cloud Vendors
Here's your scorecard for sorting out the players.
Who's making the next generation of private clouds? I'm not talking here about a bunch of disparate technologies poorly welded together and massively propped up by white glove support, but simple turnkey private clouds that don't require a room full of Ph.D.s to keep operating. Here at the midpoint of 2017 the list grows long.
Two of my customers make the list: Yottabyte and Hypergrid. They are smaller players with turnkey solutions that are ready to take on the large players. Yottabyte does well in education and among service providers. Hypergrid is the door you knock on when you have a heterogenous multi-hypervisor environment, or want to combine your existing environment to the public cloud.
There are more small players making a go of it. Nodeweaver is slowly evolving from a simple hyper-convergence platform toward a full-blown hybrid cloud solution. Unique among the others discussed here, they have successfully built their business around an SMB focus and are slowly moving into the midmarket.
There's also ZeroStack, who will give you a turnkey OpenStack-based cloud, but have the management layer live in the public cloud. Breqwatr is another OpenStack-in-a-can solution; making the hardware appliance part of a subscription is one of their main hooks. These are but a sampling of the OpenStack-in-a-can vendors shipping appliances into the market today.
There are also the companies selling a management layer you slather on top of a traditional(ish) IT architecture and pretend it's a cloud. Here you find OpenNebula, Platform9, Mirantis and their ilk. These aren't exactly turnkey clouds, but they deserve a mention because once you've set up a few they're pretty straightforward. They're the sort of thing a systems integrator could turn into a turnkey offering pretty easily.
And then there's Profitbricks-type solutions. These have a nested, multitenant-capable cloud. They sell capacity to service providers, who in turn can resell subsections of that capacity to their customers. You could call it hosted-private-cloud-as-a-service-as-a-service (HPCaaSaaS). My everything hurts.
All of the above ultimately cater to organizations seeking a low floor cost. The target customer are those who buy in chunks from 4 servers to four racks. That there is this level of competition in the mass market is absolutely fantastic, but who's playing at the level where compute is measured in acres?
At hyperscale, nobody's turnkey. In a world of turnkey solutions for everything from HCI to cloud, the line between hyperscale and not-hyperscale seems to be increasingly determined by the point at which consultants or specialists are required to keep the lights on. Any generalist sysadmin can keep a basic virtualization environment running forever if they're using cookie-cutter turnkey appliances. Scale changes things.
When we talk about players at scale, the number of vendors getting any sort of traction narrows. Many of the companies discussed above could absolutely play at hyperscale, if given the chance. VMware will be a serious consideration for large-scale deployments, but it has only just managed to pull all of its various bits into a workable solution. They are far from a "turnkey" solution, but they're no more miserable for large-scale deployments than anyone else.
Oracle would really like you to buy their private cloud. The problem with that is that it's Oracle. Even if it was the greatest turnkey solution on the planet, their licensing is a live wire and I have yet to encounter a single IT practitioner who has even seriously considered letting Oracle into a proof-of-concept for private cloud. They will find a niche for their private cloud solution in organizations who must deploy a lot of Oracle software, but I suspect that will be the extent of their success in this area.
OpenStack is really the only name in open source cloudy anything for people who want to feed a particle collider or a radio telescope. Managed OpenStack deployments in particular are popular.
IBM has Bluemix, and by all accounts are actually doing quite well among their base. HPE has Helion, while Cisco has Metacloud. Pick a vendor, from Fujitsu to Huawei, and they'll have an OpenStack offering. Who has how many and at what scale is a different question entirely.
And then there's Microsoft and Azure Stack. Azure Stack is a nice way to get companies "onboarded" to the IT-as-a-subscription concept and make moving workloads to Microsoft's Azure public cloud easier. One cannot simply download Azure Stack; it must be purchased from a tech titan. Microsoft does not ship its own Azure Stack hardware.
Of the current Azure Stack banner bearers -- including Dell EMC, HPE, Lenovo and Cisco -- Dell EMC stands apart from the rest.
Cisco doesn't have a cloud-in-a-can product of their own. They sort of have a Hyper-converged Infrastructure (HCI) offering in the form of their captive thrall Springpath's Hyperflex. A thrall they haven't yet purchased, which says all that needs be said. They have their OpenStack offering, but everyone has an OpenStack offering.
HPE has all sorts of HCI, but they haven't done so well with getting beyond HCI. If you want to buy your IT by the acre, they'll be happy to weave some OpenStack magic for you, but they're not ready to start cranking out turnkey mass market hybrid clouds, especially at scale.
Lenovo has spent the past few years making eyes at everyone, and this may yet save them. They'll partner with any nearly startup that wants to dance, with the result being that no matter who buys whom, Lenovo probably has another answer to whatever datacenter problem ails you. Unfortunately, while Lenovo's partnerships leave them with multiple HCI solutions, actual turnkey clouds aren't something Lenovo has done a lot of business development on.
Dell EMC, on the other hand, not only has partnerships with cloudy startups, they have so many HCI and private cloud solutions that they come in flavors. Dell EMC can deliver you vertically-integrated hybrid clouds starting at SMB pricing and going on up into the billions.
Cisco, HPE and Lenovo need Azure Stack if they want to be relevant 18 months from now. None of them have a mass market turnkey cloud offering other than Azure Stack. Unfortunately for Cisco and HPE, Azure Stack feeds Microsoft's Azure public cloud for large and hyperscale deployments, not the managed OpenStack offerings these vendors count on.
Dell is in a different position. Azure Stack is nice and all, but they have VMware-based VxRAIL for the mass market, among other offerings. And these can feed into larger VMware deployment options, including VMware Cloud on AWS for those who really want some public cloud in the mix.
Microsoft has a potential winner in Azure Stack. The key is the pipeline they've built into their public cloud offering. Get 'em hooked while they're still smallish organizations and as they grow they'll inevitably rely more on Azure.
Dell seems to be the only tech titan with any real shot at building a similar pipeline from what they have on hand. IBM is maybe the next closest, but they don't really have a mass market offering to get customers hooked on private cloud in the first place.
Meanwhile, there's a galaxy of smaller private cloud players making turnkey solutions that are not only easy to use but themselves are theoretically capable of scaling up to multi-datacenter scale. Any of them -- or maybe even multiple among them -- could become the next decade's tech titans on the back of a cloud solution with a pipeline for scale that points inward.
Much remains uncertain, except this: the future of IT lies in clouds. With so many options available, it looks like there's going to be a lot more private and hybrid in the mix than the public cloud purists would like to believe.
Trevor Pott is a full-time nerd from Edmonton, Alberta, Canada. He splits his time between systems administration, technology writing, and consulting. As a consultant he helps Silicon Valley startups better understand systems administrators and how to sell to them.