Cloud Migration Vendors – The End of the Golden Weather

Over the past couple of years there have been a large number of startups gain attention and funding by promising the holiest of grails – effortless migration of workloads between cloud vendors. This promise, the basis of the hallowed concept of cloudbursting, was all the rage a few years ago. The list of vendors entering the space was extensive – from Ravello to CloudSwitch, from CloudVelocity to Cliqr – all were going for the high ground of migrating live workloads.

And then something changed. Almost imperceptibly the messaging that these companies used to describe what they offer morphed into something new. When CloudVelocity announced its general availability a month or so ago, I remarked about how the company had moved to articulating a disaster recovery value proposition for what it does. The spin from the company was that:

[they are] introducing a disaster recovery story to go alongside the workload portability one. In this way they believe they’re able to articulate a two step value proposition – seamless DR as an initial step. Since DR is a generally accepted first step for cloud usage, the company can leverage this fact by offering to enable the actual DR process. Thereafter they’re able to extend that message to their core workload migration story.

Joining this shifting space of messaging comes Ravello who this week announced its own worldwide general availability. This time the move is even more stark – the first paragraph of their announcement heralds the fact that:

enterprises can seamlessly use any leading public cloud to develop and test their existing on-premise applications.

Ravello goes on to (rightly) point out that organizations wishing to create real agility demand large test/dev environments and that standing these environments up on-premise doesn’t make a whole lot of sense. The barrier to doing test/dev in the cloud has of course been the differences in terms of technology stacks between the cloud provider and the on-premise production environment. By developing its “Cloud Application Hypervisor” essentially a bundled application stack that is deployable across environments, Ravello enables companies to duplicate their on-premise environment in a lower cost public cloud – currently AWS, Rackspace or HP Cloud.

Don’t get me wrong, there is some real benefit here – building specific test/dev environments on-premise is expensive and time-consuming and thus places a barrier in the way of more nimble application-based innovation That’s all good. But….

As I mentioned back in February, Ravello garnered significant attention, not to mention $26M in VC funding, with the promise of that holy grail, truly live application portability and cloudbursting. Dev/text is undeniably a useful and valuable service, but in a world where, at least over time, organization will be building production environments with similar architectures to public cloud ones (An OpenStack-based private cloud and HP Cloud public for example) the problem they’re solving is less widespread then before.

Of course one could take the view that this move into a more prosaic and less fabled technical space is a logical reaction to the reality that organizations DON’T ACTUALLY WANT CLOUDBURSTING. It’s an exciting prospect, but one which is largely irrelevant when it comes to the coalface of enterprise IT.

It seems a widespread change of focus is underway – these companies with lofty ideals will now more strongly move into the space dominated by the likes of Rightscale and enStratius – there’s a big opportunity there, but also a crowded marketplace. I suspect that the wide selection of these sorts of vendors will be greatly reduced over the coming years.

4 Comments
  • Well, I agree with Mr. Kepes that the notion of workload interoperability was the “holy grail” of cloud computing, but it is not surprising that it is looked upon a bit differently today. Hybrid cloud computing seems to be the current reduction to practice for workload interoperability. It remains largely confined to a single cloudstack to pull it off. Eucalyptus private cloud users can migrate workloads to AWS. OpenStack private cloud users look to be able to migrate workloads to Rackspace or HP public OpenStack offerings. And VMware private workloads will be able to move VMware’s Hybrid Cloud Service real soon now. This is all a far cry from where we thought workload interoperability was headed a few years ago, but the whole concept is so radical and new that it is not surprising that customers are taking a more nuanced look at it and choosing something like a DRaaS solution as an entry point. I do agree that if there is no commercial market of sufficient size then we will see M&A activity among the survivors in a couple of years or whenever their “burn rate” exhausts their funding.

  • Poignant observation by Ben and a no less poignant comment by Tim, which leads to my two cents.
    What was a hot enough pitch a few years ago to get all those companies funded and into dev production, slammed against a simple, but at that time unforeseen occurrence of Cloud-based Stack fragmentation. As various stacks and versions of stacks began to evolve, live workload interoperability became a much more distant reality, hence the DR offering naturally became the only commercially viable market entry point, for the above mentioned vendors.

  • Having worked both at RightScale and now at CloudVelocity, having spoken with thousands of people (sometimes in one month), having seen the smartest teams help clients with the most complex cloud based deployments…..gonna have to say there is a huge difference in where these companies play.

    ” these companies with lofty ideals will now more strongly move into the space dominated by the likes of Rightscale and enStratius”

    Not really, that would be a huge mistake from a business perspective. The ability to automate migration is a very different problem from actually managing multicloud infrastructure, with complex problems on either side that are, not in any way, cookie cutter. These problems are just starting to be addressed and the more you differentiate, the better.

    Cloud bursting…….I would of agreed with you but there are tons of HPC applications out there where this is a massive benefit, IT is all game to play. When scientists have to ask IT to provision resources, they wait for it (agility = time = money), get fed up, buy more hardware, rinse + repeat, cloud bursting becomes an amazing solution.

    These HPC applications are almost always not “cloud ready”. By automating the migration and synching the data, these clients can now launch the HPC application in AWS whenever the demand for computing capacity is requested. This use case was for AutoCad.

    Live workload interoperability is not far off, as it will be a combination of vendors an enterprise works with to achieve it.

    DR is not “the only” viable market entry point, it’s just a no brainer. That is a big difference. Why would anyone want to run or buy duplicate servers, when you can store it all on the cloud.

    Hope that helps give you an view from what I have seen. It might be the exact opposite of the end.

Leave a Reply