Everyone in tech surely knows what a “public cloud” is. It scores 100% among enterprises, for sure. How about “private cloud” though? While almost all enterprises (398) tell me they know what that is too, roughly a third fail to offer any signs of being convinced when pressed. Of the remainder (271), almost half define it more as “hybrid cloud” would be defined, and that’s not as bad as it sounds, because the knowledgeable cloud users say there is a relationship between “hybrid” and “private”.
Of 398 enterprises who offered comments on cloud definitions, 104 were confident and articulate in offering a definition, and in fact offered one that was substantially the same. A “private cloud”, according to them was a hybrid cloud model based on the use of the same platform technology in the data center and the public cloud services they adopted. This could be achieved either by hosting data-center cloud platform technology in the cloud, or by hosting cloud technology in the data center, though 88 of the 104 reported doing the former. What leads to this situation? It turns out there are a number of factors that create a “private cloud” model. Sometimes it’s planned, and sometimes users sort of fall into it.
Of the 88 enterprises who said their private cloud approach was driven from the data center side, 75 were customers of Red Hat or VMware, or users of open-source tools, for data center virtualization. They elected to host the same technology on their cloud services, and 70 of that group had multiple reasons.
The top reason, cited by all 75 of the data-center-driven group, was a desire to have the same platform in both cloud and data center, the platform they were already familiar with. Number two (66 of the 70) was that they calculated the cost of public cloud services under this approach would be lower than if they used specialized cloud tools, and number three (58 of 70) was that they believed it would be easier to support multi-cloud or to change cloud providers if they adopted their current data center model.
The 16 enterprises who selected a public cloud model as their private cloud did so because they’d decided to re architect to a cloud-centric approach. Exactly half were multi-location retail companies who did most of their application heavy lifting via edge computing in the retail sites. They used the cloud both to support online ordering (which was often fulfilled at a local retail location) and to collect data for central storage and analysis in their data center. Of the remaining 8, 5 said they had no real data center virtualization platform in place, and so adopted one compatible with the cloud service. The other three didn’t provide details on their decision.
The really interesting thing here was that only 9 of the original 104 who confidently defined and articulated their private cloud strategy saw themselves as explicitly building a private cloud, which suggests to me that the majority of the companies who saw “private cloud” as any hybrid cloud model may have had similar views and drivers to the group of confident private cloud articulators. To me, what this suggests is first that the hybrid cloud model is indeed where enterprises are going almost in total, whether they say so or not, and second that the data center virtualization model (bare metal, containers or VMs, at a high level) is converging on the latter two, which also happens to be the hosting models that dominate the public cloud. Unwitting and coincidental convergence? Probably so.
It’s hard to dig out early motivating factors, especially when you’re not doing explicit surveys, but I think that there is a deep lesson here too. Enterprises don’t think as much about their “cloud strategy” as they do about application strategy. Technology is a means to an end, even for most CIOs, and so they aren’t trying to adopt one or another, but rather are looking to exploit tech as a path to a goal. This, of course, is the view of most CxOs, particularly high-level financial, company, and sales-marketing executives. The CIO has the critical task of matching technology direction to the application and business mission, and that matching process will first uncover the role of the cloud, and second the extent to which that role matches virtualization technology that’s in use in the data center. The cloud, then, is hosted virtualization. The only question is whether the way cloud virtualization and data center virtualization relate to each other. There are three choices—the cloud drives, the data center drives, and the two are not interdependent. It seems that most enterprises make the choice among the three based on their satisfaction and familiarity with their data center approach.
It’s really difficult to know whether this is a good thing or a bad thing. If enterprises saw the cloud as hosted virtualization, saw private cloud as a way of implementing the data center/cloud relationship in an optimal way, would they do anything differently? I don’t think you ever get a good answer to a hypothetical question you ask an enterprise technologist, and of course that’s not the sort of question they’d be likely to answer indirectly in a free-form chat. My own assessment based on experience and analysis of the way enterprises seem to think is that it would make a difference.
One good example of this is containerization and orchestration. Of the 75 data-centric enterprises whose cloud model tracked data center platform evolution, 48 used containers for most or all their applications, both in the cloud and in the data center. Of that group, 16 were container-centric pre-cloud, 18 became so roughly concurrent with the adoption of cloud hosting, and 14 said they had adopted containers in the cloud and data center for independent reasons. I think looking at the middle 18 offers the most interesting insight into the topic.
Containers are, IMHO, the essential cloud virtualization model in that they’re more elastic than VM deployments and thus play to the specific benefits of the cloud, but also offer deployment, resilience, the ability to respond to load changes in data center apps. This makes containers the technical bridge between the two environments. I wonder whether pushing this viewpoint out into the real world would help users plan better for hybrid cloud adoption. Personally, I think it would.
But who pushes this approach? Enterprises say it’s primarily the data center giants like Dell, HPE, IBM/Red Hat and VMware. They suggest (but only suggest) that public cloud providers encourage them to think of the cloud as not only an independent environment but also the preferred one. That view, if accepted, would tend to maximize cloud provider influence and revenues.
It may all come down to account control, but that factor isn’t always operative. First, only the four giants named above really have account control. Second, less than half of enterprises tell me any vendor really has account control. Thus, a big chunk of enterprises are really swinging in the PR cycle on cloud adoption, which of course is driven mostly by cloud providers. But there’s a third factor too; even when one of the four vendors listed does have account control, they aren’t always prepared to exercise it to promote container-based hybrid unity.
All this has to be considered in light of an interesting statistics set. The group of 75 enterprises who built hybrid clouds by transplanting their data center virtualization strategy to their public cloud choices reported less than half the problems with cost or technical project failures as cloud users overall. Private cloud, in this form at least, is delivering better outcomes for users.
