Hyperconvergence and the "Why?" of the Cloud

The rapid success of the cloud has created a sometimes confusing jungle of wrinkles around just how larger businesses should approach this trend... Is your organization in a position to go whole hog into the cloud? Or do you still need to keep some systems but not all on-prem, leading you to a "hybrid" cloud? Just why did you want to be in the cloud in the first place and in this all jargon what are the buzzwords that actually refer to the practices that will concretely benefit your business? "Hyperconvergence" is a big, intimidating term that could probably use a bit of demystification anyway and unpacking it provides a great window into the actually quite different reasons that different organizations might be interested in cloud computing.

The DoD JEDI cloud computing contract was an enlightening case study for me (and will likely continue to be). The military has some very conventional reasons it might want stay on-prem, including unique security concerns and the scale necessary to do a good job. The JEDI contract, for those that enjoy argument, was almost a contract to help DoD build on-prem and not in the "cloud" at all. Why the interest in the cloud? The answer I got asking around is in part that the military is interested in adoption of other technologies like machine learning consistently across a large and sprawling organization. It is the software layer that comes along with the cloud that really enables this and not any issue of where the data is hosted and who owns the physical server.

If there is something "hyper" in hyperconverged infrastructure that didn't get the name for marketing reasons, it is the hypervisor that creates and runs the virtual machines that then host the databases, servers, etc. more familiar to most engineers. Moving this task to a software layer, and not a hardware one, is a bit part of what makes the cloud scalable, cheap, and agile especially. Through this lens, you might say that cloud providers will important innovators in hyperconverged infrastructure, and what we called the cloud in the past was outsourcing of this new infrastructure. You might ask, though, if what you really wanted was to be on-prem but using the new software, and this is implicitly the sort of decision that the DoD made with the JEDI contract.

It seems like many large organizations drawn by the hype into hybrid cloud situations are having this kind of revelation. If you adopt the technologies used by the cloud providers for your own data center, particular those around virtualization and emphasis on the software layer, you might find that this is a better way to get at the value you thought "the cloud" was supposed to provide. It also makes your hybrid cloud environment much more workable when people inevitably need to collaborate across the true-cloud and on-prem components.