Hyper-converged infrastructures are quickly capturing the fancy of end-user organizations everywhere. They bundle hypervisor, server and storage in a single node and provide the flexibility to scale-out to form a single logical entity. In this configuration, they offer a very real opportunity for organizations to economically and practically collapse their existing infrastructure of servers and storage arrays into one that is much easier to implement, manage and upgrade over time.
These benefits have many contemplating whether hyper-converged infrastructures foretell the end of big box servers and storage into a data center that is solely hyper-converged. There is a great deal of merit in this viewpoint as an infrastructure shift to hyper-converged is already occurring with a high likelihood that it will go hyper-growth as soon as 2016.
The driving force behind the implementation and adoption of hyper-converged infrastructures is largely two-fold. The advent of server virtualization – which has already been going on for the better part of a decade – and the more recent rise of flash memory as a storage media.
Server virtualization has reduced the number of physical servers that organizations have had to implement as, using this technology, they can host multiple virtual machines or VMs on a single physical server. The downside of server virtualization to date has been the inability of storage media – primarily internal hard disk drives (HDDs) – to meet the performance demands of hosting multiple VMs on a single physical server.
This has led many organizations to use externally attached storage arrays that can provide the levels of performance that these multiple VMs hosted on a single server required. Unfortunately externally attached storage arrays, especially when they are networked together, become very costly to implement and then manage.
In fact, many find that the costs of creating and managing this networked storage infrastructure can offset whatever cost savings they realized by virtualizing their servers in the first place. Then when one starts to factor in the new levels of complexity that networked storage introduces, the headaches associated with data migration and the difficulty in getting hypervisors to optimally work with the underlying storage arrays, it makes many wonder why they went down this path in the first place.
The recent rise of flash memory – typically sold as solid state drives (SSD) – changes the conversation. Now for the first time organizations can get the type of performance once only available in a storage array in the form factor of a disk drive that may be inserted into any server. Further, by putting the storage capacity back inside the server, they eliminate the complexity and costs associated with creating a networked storage environment.
This two factors have led to the birth and rapid rise of hyper-converged infrastructures. Organizations can eliminate implementing today’s networked server and storage solutions by using a single hyper-converged solution. Further, due to their scale-out capabilities, as more computing resources or storage capacity are needed, additional nodes may be added to an existing hyper-converged implementation.
These benefits of hyper-convergence have every major big box IT vendor from Dell and HP to Cisco and EMC proclaiming that they now have a hyper-converged story that competes with up-and-comers like Maxta, Nutanix, Simplivity and Springpath. These big box vendors recognize that a hyper-converged solution from any of these emerging providers has the potential to make their existing big box server, networking or storage array story one that no one wants to hear any longer.
One question that organizations need to answer is, “How quickly will hyper-converged solutions go hyper-growth? Near term (0-12 months) I see hyper-converged infrastructures cutting into the respective market shares of these different systems as organizations test drive them in their remote and branch offices as well as in their test and development environments. Once these trials are and, based upon what I am hearing from enterprise shops and the level of interest that they are displaying in this technology, 2016 could well be the year that hyper-converged goes hyper-growth.
At this point, it is still too early to definitively conclude the full impact that hyper-converged infrastructures will ultimately have on today’s existing data center infrastructures and how quickly that will happen. But when looks at how many new vendors are coming out of the woodwork, how quickly existing vendors are bringing hyper-converged infrastructure solutions to market and how much end-user interest there is in this technology, I am of the mindset that the transition to hyper-converged infrastructures may potentially happen much faster than anyone anticipates.