Verizon has taken what may be a very important and evocative step toward maturing its enterprise cloud strategy with the purchase of privately held CloudSwitch. The significance of the move is hard to appreciate without an understanding of just what the heck CloudSwitch is, so I propose to start with that.
The classic vision of cloud computing is virtual something-hosting, where the “something” is anywhere from an entire application to the bare-bones machine image (SaaS down to IaaS, respectively). This model is useful as a way of looking at the cloud in isolation, but for most enterprises the cloud in isolation isn’t very interesting. Since they don’t expect to migrate more than a max of a quarter of their IT spending to a public cloud, the key question for them is how you hybridize.
Microsoft is the only one of the currently popular cloud leaders that has taken the hybridization to heart from the first. The Azure cloud is a PaaS cloud that, with the help of the Azure Platform Appliance, which is a partner-delivered combination of Microsoft software and server and network hardware. With APA, a user can build an “Azure cloud” that seamlessly extends between an enterprise data center and a public cloud provider (Microsoft, of course, but in theory other cloud providers who adopted the Azure architecture).
CloudSwitch can be visualized as a more generalized model of the same hybridization notion. With this approach, the user deploys a series of CloudSwitch Instances in the cloud and a CloudSwitch Appliance (which is a software component, not a gadget) in the data center. The Appliance links to all of the Instances in as many clouds as there are, and it essentially synchronizes each instance as a host for one or more virtual machines that are managed to be identical functionally with the applications’ resources in the enterprise. What you end up with is a kind of “envelope” that everything runs in and that can be made to extend to any number of clouds that can host a virtual machine. A secure Internet tunnel links the components of this architecture.
There’s a lot to be said for this approach, but it’s not a panacea for cloud issues. What CloudSwitch does is make public cloud resources (Terremark’s resources, in this case) appear to be elastic extensions of local VM hosts. The way this is done (once it’s set up) is largely transparent to users so the cloud can really appear as an elastic extension of the data center. For “cloudbursting” applications where the cloud takes up the slack when applications overload the in-house computing resources, it’s an easy way to build the framework without becoming a specialist. Compared to Azure, which is PaaS and thus imposes some application and middleware constraints, it’s more flexible.
One place the concept falls short is in the area of “equivalence”. Yes, CloudSwitch can create a virtual data center that spans the real one and the cloud, but the stuff that goes into the cloud still has to conform to the price paradigms of the cloud and is still constrained by the tunnel connection in terms of performance. For Verizon, these limitations probably won’t be critical because I think the company is likely to be targeting SMBs with the offering and because enterprises could be made to understand the limitations of the Cloud Isolation Technology and exploit the capabilities readily.
The most significant thing about the deal, I think, is that this is a NETWORK OPERATOR buying a software company. If you think about this, that’s a major twist because in the past operators would have expected their vendors to offer them a package, or would have “introduced” a startup to a big vendor in an arranged acquisition. Here’s a Tier One buying their own service-layer technology. If you need proof that the network equipment vendors have fallen asleep at the switch, this should be it.