You’ve probably heard a multitude of things around hyperconvergence or hyperconverged infrastructure as these are becoming hot new industry buzzwords, but what do these terms really mean? Are vendors that say they have hyperconverged infrastructure really living up to the promises of true hyperconvergence or is it just marketing hype?
The terms “hyperconvergence” and “hyperconverged infrastructure” originated in a meeting between Jeff Ready and Jason Collier of Scale Computing and Arun Taneja of the Taneja Group. According to these three, the term was coined to mean the inclusion of a virtualization hypervisor into a complete infrastructure solution that included storage, compute, and virtualization. Some have thought that hyperconverged is synonymous with terms like ultraconverged or superconverged but that was not the intention.
If we hold this intended definition of hyperconvergence from its creators as the standard, what does it mean to be a real hyperconverged solution? Many solutions that call themselves hyperconverged rely on third party hypervisors such as VMware or Hyper-V for virtualization. The hypervisor software in that case is developed and licensed from a completely different vendor. That doesn’t seem to fit the definition of hyperconvergence at all.
Many vendors that use the hyperconverged label are merely peddling traditional virtualization architecture designed around traditional servers and SAN storage. This 3-2-1 architecture that built a platform for virtualization from a minimum of 3 servers, 2 switches, and 1 SAN appliance has been repackaged by some as a single vendor solution without any real convergence at all, and definitely no hypervisor. It is important to differentiate these traditional architectures from the next generation architectures that the term hyperconvergence was intended for.
Before hyperconvergence there was already a concept of converged infrastructure that combined server compute with storage as a single hardware platform onto which virtualization could be added. If a solution is not providing the hypervisor directly but relying on a third party hypervisor, it seems to fall back into the converged category, but not hyperconverged.
One of the key benefits of hyperconvergence is not having to rely on third party virtualization solutions, and being independent of the costs, complexity, and management of these third parties. The idea was no more hypervisor software licensing to pay for and one less vendor to deal with for support and maintenance. Hyperconvergence should be a complete virtualization and infrastructure solution from a single vendor. Maybe it is a marketing necessity for some vendor solutions to use a third party hypervisor to get to market due to funding. These solutions are not fulfilling the promise of true hyperconvergence for their customers.
Hypervisors have been around for long enough now that they are now a commodity. The idea of having to license and pay for a hypervisor as a separate entity should be giving IT solution purchasers pause as they look to implement new solutions. Cloud providers are not requiring customers to license hypervisors so why would so-called hyperconvergence vendors do this? We are hearing more and more from IT managers about their displeasure over the “Virtualization Tax” they’ve been paying for too long. The hype cycle for virtualization is over and users are ready to stop opening their checkbooks for an operating environment that should be included at no extra charge.