Alternative Facts of IT: “We have a private cloud.”

Alternative Facts of IT: “We have a private cloud.”

We’ve been wrestling with this one at Nemertes for many years.  Early on in our research around enterprise cloud use (back around 2010) we asked whether folks had a private cloud and got sensible answers, overwhelmingly in the negative. Almost nobody said they had a private cloud in their own data centers. Why? Because having a private cloud was hard and that early on pretty much meant you were writing your own cloud management platform.

A couple of years later, we got enormously different responses — a majority of the participants said they had a private cloud.  At first this worried us — had we missed some enormous change in the environment? Some massive overnight transformation?

Turns out, no, we hadn’t been asleep for the revolution.

And, it is a good thing we were asking a lot of detailed follow ups to that question, about things like automation and orchestration across silos, accounting for resource usage by specific workloads, chargeback vs showback, and self service access to services both within and outside IT. When we looked at the answer on “Do you have a private cloud?” vs the answers on all those detailed questions–things that comprise much of the functional description of a cloud’s management infrastructure–there was little correlation. People had ever-more-heavily virtualized environments, of course, but not clouds in anything but name.  Basically, they had gotten into the habit of speaking of their own resources as the “private cloud” but had not yet made that a functioning reality  behind the scenes.

Saying they had a cloud was part adaptation and part aspiration.  Adaptation, in that people outside IT (and plenty of people inside IT as well) had started to become caught up in the hype around cloud and wanted to talk in terms of private clouds, aided and abetted by many in the vendor pool, analyst community, and press. If a CIO or IT director had to start every conversation explaining how they didn’t have a private cloud, they ran the risk of either derailing every potentially productive conversation about the organization’s needs from IT and goals for services, even of being perceived as some kind of reactionary opposed to “the inevitable future” and needing to be sidelined or replaced.

Saying they had a private cloud was also aspiration.  They wanted to get there, to have an infrastructure that would support Amazon style DIY deployments for sysadmins and developers (though not for everyone everywhere in the organization), that was fluidly resourced and comprehensively redundant and software defined and virtualized in every resource.  They were hobbled in that by not being able to start building a data center and software environment and staff with that in mind. They had to — and still have to in most cases — plot a winding course to private cloud that takes into account the multiple layers and generations of infrastructure, software, and the staffing that keeps them all spinning and serving the enterprise. It also has to be built around the tools they can actually buy and deploy; buying a cloud in a box is partially possible, but only recently.

Bottom line is, I applaud the aspiration towards making a true private cloud in the data center, and the true hybrid cloud computing this will make possible. I just hope everyone remembers the difference between what they currently have and what they actually need for that to be a reality. Pressing ahead as though the private cloud is a reality leads to headaches at best, and expensive failures at worst.

Share this post