John Burke — Principal Research Analyst with Nemetes Research
John advises key enterprise and vendor clients, conducts and analyzes primary research, and writes thought-leadership pieces across a wide variety of topics. John leads research on private and public cloud; private cloud infrastructure; private and hybrid cloud management and security; network, server and storage virtualization; and software-defined networking (SDN), SD-WAN, and network functions virtualization (NFV).
You want your private cloud and you want it now. But are you ready?
We in IT can be what I like to call aspirational liars. That is, when someone asks us about how we are doing something IT-related, we tell them what we are working towards rather than what we are really doing. “We run an ITIL operation,” for example, often translates to “We have implemented a partial change management process, use a service desk ticketing system and are trying to get started using a CMDB.” That is, “We have cribbed a couple of good ideas from ITIL, but have had trouble implementing them fully and can’t get any further.”
Although it is rapidly being eclipsed by “We do DevOps,” I think the aspirational lie I hear most often is, “We have a private cloud.” In fact, it is so internalized, it gets subsumed into other statements, casually: “We’ll run it in our private cloud,” or “We are running out of storage in our cloud.” Unfortunately, when in our research studies we look at the characteristics of cloud (elasticity, self-service, pay-for-what-you-use, etc.) and ask about them individually, we find that most organizations do not implement most of these characteristics. No more than 20 percent of IT shops actually run a full-on private cloud. What they have are partially virtualized and software-defined resources and a whole bunch of legacy operational practices.
IT can’t sell cloud services by the drink to the rest of the organization if it doesn’t yet know how to run a cloud. Getting from conventional virtualized resources to a true cloud takes a lot of change at many levels. IT has to change its own and its users’ expectations; its resource model for the raw components of storage, networking, and compute; and its software stack, to get the automated support required for elastic consumption, self-service, simple scaling, and full orchestration.
Having all the parts isn’t the same as having to use them all immediately, or having to allow the rest of the organization to do so. IT can work with the enterprise to smoothly transition from its current operating model to a cloud model by building cloud capacity and capabilities and gradually shifting how the rest of the organization engages IT.
To learn how to run a cloud, IT must have a cloud-ready infrastructure, in both hardware and software: from a flexible and software-defined network up through a comprehensive cloud management platform for resource management and orchestration, service catalog definition, consumption tracking and billing.
IT can then expose services to the enterprise via a self-service portal, so that both sides can learn to operate in a more dynamic and demand-led environment. Most organizations need to learn how to deal with true accounting for resource consumption and the implications of self-service before they can safely run a cloud.
Once they’ve succeeded at a human-speed cloud (where virtualized resources are allocated automatically following human selection of items in the self-service portal, for example) they can make the final step, to a machine-speed cloud, by exposing services to developers via APIs.
This is not an all-or-nothing proposition, of course, and most true private clouds start out as islands of functionality within a more traditional infrastructure, expanding to provide more services and accommodate in-migrated workloads year by year. But you can’t offer a private cloud to your clients if you don’t know how to run one, and it can’t operate like one behind the scenes.
The time to begin making the transition is now. Otherwise, aspirational promises will create a string of broken promises to the organization, an insurmountable technical debt, and a disappointed client base looking for new IT leadership.