For the past five years now, Microsoft has been quietly building what might be the best answer for Hybrid Cloud in the form of Azure Stack. Many non-cloud technology vendors have talked long and loud about Hybrid Cloud but none of them agree with each other, there being dozens of “Hybrid Cloud Reference Architectures” out there in the wild. Throughout this period Hybrid Cloud has increasingly looked like a weak attempt by water-treading legacy vendors desperate to stay relevant in the cloud era.
Azure Stack looks different:
Unlike most Hybrid Cloud “solutions”, the Azure Stack is not simply a virtualization layer with a cloud-like front end. It has an extendable number of modules that offer a range of cloud services, more than just IaaS.
What’s so different about Azure Stack?
There are five important differences when comparing Azure Stack to other Hybrid Cloud solutions.
- Azure Stack is a productised solution, it is downloadable and doesn’t require a services wrapper.
- Azure Stack will do more than just IaaS, with modules for higher-order cloud services.
- Azure Stack will extend the Azure reach from public cloud to on-premises.
- One API and UI experience across public and on-premises access.
- Most Hybrid Clouds are a “lock in” (see VMware’s “one cloud”) whereas Microsoft are trying to make this an open platform (in terms of extensibility and connectivity).
The problem for companies who are not Microsoft is how do they compete with this vision and execution. Executed well, Azure and Azure Stack could be the five-year play that enables Microsoft to increase growth in areas that AWS cannot reach: the enormous on-premises enterprise. Microsof could opening a new growth corridor their increasingly two-horse race with their hyper-scale public cloud competitor.
Vendors that don’t do public cloud are in line to have their market taken if Microsoft are aggressive and successful enough in reaching from public cloud into the enterprise and onto their premises.
1. Azure Stack is the first real hybrid cloud
Hybrid Cloud as a concept has been around for five years. The latest Rightscale State of Cloud report shows the pent up demand for this kind of “multi-cloud solution”:
A simple search for “hybrid cloud reference architecture” reveals pages of results from the likes of industry leaders Rackspace, VMware, IBM, EMC and Oracle. Although their solutions are all different they do have one abstract theme in common: they are solutions (not products) that are built around the vendor’s portfolio, and they mostly assume that hybrid cloud is an extension from on-premises to the public cloud. Their selling point to enterprises is to reassure them:
keep doing things the same way you have always done, but with a bit of public cloud hidden at the back end.
Microsoft offers the same comfort but they have additional weaponry: Azure public cloud let’s you do things the same way and on-premises, with virtual machines an all that stuff, but it also let’s you do things in the new, real cloudy way with APIs, event processing, database services and more.
There Azure Stack is also the most productised offering (prescriptive, doesn’t require costly consultancy head scratching) and it extends from the public cloud into the on-premises datacenter. That is, they are bringing their public cloud know-how, look and feel into the customer’s datacenter.
It is notable that executives and practitioners exposed to successful use of the public cloud find it hard to go back to the legacy, on-premises world. Microsoft have demonstrated that you can use the same Powershell, APIs and interfaces to on-premises Azure Stack that you can use for public Azure. They also demonstrated that you can access the public cloud and “reach into” the on-premises datacenter, so that you can ensure data stays on-premises but benefits from the public cloud broad access.
2. There are no signs that AWS or Google will do hybrid cloud
Today you can do “a bit” of Hybrid Cloud with AWS and Google, but they are token efforts. Quite rightly AWS and Google insist that their public clouds are great places to run any workload, and enterprises are fast catching up with that idea.
With AWS you can get plugins and connectors from your on-premises systems to back up data to AWS, and even manage some basic aspects of AWS from VMware’s vCenter. You can connect networks and fire up cloud VMs from an on-premises portal but this is barely scratching the surface of what Azure Stack will do. As for Google, well they did a deal with VMware to let you consume some GCP services from vCloud Air.
At AWS Re-Invent in 2014, Andy Jassey acknowledged that enterprises do not make binary “all cloud” or “no cloud” decisions but there are no clear plans as to what the AWS answer will be. No surprise, though, given AWS’s now legendary silence on futures! Who would rule AWS out of any market? Not me!
3. Hybrid Cloud is about more than just infrastructure
There is a bizarre obsession with infrastructure when a some people talk about cloud, perhaps because its the most familiar for virtualization stalwart practitioners. Public cloud providers offer much more than rent-a-VM and there is a fast emerging trend towards serverless architectures. These trends might be starting in public cloud but Azure Stack will bring them to hybrid cloud. Can you imagine VMware saying “virtual machines aren’t required, neither is the hypervisor”? Their foray into containers still required virtual machines, it’s in their DNA.
A serverless architecture might also be called an infrastructureless architecture, because if you don’t need the server/VM then you don’t need the network or storage or OS either (that’s someone else’s problem, of course it all has to run somewhere, but in serverless it means from the consumer’s perspective).
If you want to use a database, or a message queue, then why go to all the trouble of renting a VM, creating the network and storage, installing the OS and software, configuring the application, managing it… when you can just consume an existing service on a per-transaction basis?
The unit of purchase is shrinking from big solutions down to rent-an-infrastructure, to rent-a-container through to rent-an-application or pay-per-event.
Azure Stack will have a lot of application features in it, bringing PaaS capabilities into a coherent hybrid cloud stack. It looks like they will also play nice with enterprise PaaS vendors such as Apprenda, Cloud Foundry and even once-competitor, now-friend, RedHat with OpenShift.
What next for 2016?
It has taken Microsoft more than four years of tough product and engineering work to get Azure Stack through previous incarnations to where it is today. It isn’t clear if AWS or Google are going to attempt something similar, or perhaps just do token efforts like they have so far. The on-premises vendors don’t have the public cloud capabilities or experience to match Microsoft, so it isn’t clear how their futures will play out.
It is also important to note that big Microsoft partners such as Accenture, HP, Dell and others will likely play a driving force in adoption of Azure Stack into larger enterprises. It remains to be seen how the smaller regional partners can help Microsoft help Small to Medium Enterprises (SME).
In the past few years Microsoft has been the uncool poor relation and number two in many markets for quite some time, especially in the virtualization market. Throughout that time, however, they have grown many billion-dollar businesses and they are a familiar partner and known entity to most enterprises. In 2016 the pendulum appears to be swinging back in their favour, and perhaps rightly so given the self-disruption they have gone through.
There are three things to watch for in Hybrid Cloud in 2016:
- Will Microsoft be able to execute the Azure Stack? Are their agreements and pricing right, are their staff able to sell it, are their partners on board? Execution is hard.
- Will AWS and Google counter? Can they do what took Microsoft a number of years to achieve? But they don’t have the sales staff or partners to execute. Is it just not possible for them?
- Will Dell/EMC/VMware, IBM, Oracle and the rest see their on-premises market shares decline? Perhaps at an alarming rate? Or will they survive due to Microsoft failure to execute, or perhaps they have an answer up their sleeve that nobody has seen yet? It will be interesting for these traditional lock-in vendors to accuse Microsoft of something similar. Same old arguments.
If this goes Microsoft’s way then Azure Stack could be the coup de grâce to existing attempts for non-public cloud providers’ attempts at Hybrid Cloud. It could be their’s to lose because, at the moment, there is nothing else as good in sight.