To understand reference architectures for hyper-converged systems, we should look at how we got to this point....
By submitting your personal information, you agree that TechTarget and its partners may contact you regarding relevant content, products and special offers.
All of the original hyper-converged systems that came to market arrived as appliances. Whatever server was in that appliance -- whatever networking was in there -- it was already packaged inside their node.
More recently, a variety of vendors that may have started with scale-out storage as a primary product line recognized they had a very similar architecture to hyper-converged systems and that their CPUs had an ample amount of processing power left over. And while they wanted to convert that storage system into a hyper-converged system, many of these vendors offered their storage as software-only.
However, they figured out quickly that an appliance, or an experience that mimics an appliance, is a must-have. So what they've done is to say, "Look, we're going to provide you with a reference architecture," which you can think of as a set of guidelines. I'll give you an example: Maybe there is a reference infrastructure for a 500-user virtual desktop infrastructure (VDI) environment. In this case, the reference architecture could be for persistent VDI, but another could be for non-persistent VDI. The vendor provides a customer with a reference architecture and tells them, "If you want to get adequate performance and adequate usability out of a 200- or 500-user VDI environment, here's what we suggest you do: Use our software and provide this type of hardware, combine it, use the best practices and you will get an appliance-like experience at the end of the exercise."
That's when reference architectures can be helpful to you -- when you want a hyper-converged experience but are only buying the software, not an entire system. Reference architectures are important because they put boundaries around the hardware-software combo so the end result for the user is a delightful experience. Simply put, you can think of a reference architecture as a set of guidelines inside a set of best practices. It allows a software-only product to be delivered to the end user in an appliance-like fashion to produce the desired results. The benefit to the user who chooses a software-only option is flexibility; as long as the boundaries set by the vendor are followed, you can choose the hypervisor, hardware and type of server you want.
Nimboxx releases Atomic Unit hyper-converged platform
DataCore Software CEO gives his take on hyper-converged storage, SDS