Introduction
Software-Defined Storage (SDS) platforms are called to make storage architectures simpler and cheaper to maintain. Hence, with respect to the latest developments in the field, it’s no wonder that more and more businesses of any scale go hyperconverged.
Hyperconverged infrastructures are often contrasted with compute and storage separated (AKA converged) ones. In such hyperconverged scale-out models, hardware building blocks are rack-mounted. Industry-standard servers with internal drive bays can be populated with direct-attached spindles and SSDs. Speaking of off-the-shelf hardware, will any server and a bunch of disks work for your hyperconverged solution? Let’s find it out!
Diving into hyperconvergence
Why is the whole world deploying this tech? Within hyperconverged environments, hardware and software are integrated very tightly, making it easier for admins to build an environment that fits the needs of their company. Another strong point of this tech is that you can use the commodity storage and the server hardware. The freedom to build an environment with the components of your choice enables to use familiar management and monitoring tools, and leveraging the existing servers within your environment. If you love innovations, you can extend your infrastructure to public cloud. Alternatively , you can just build a stretched cluster. With hyperconvergence, maintaining environments like that won’t be rocket science!
So, how to build an environment that rocks
Successful deployment of hyperconverged solution mainly depends on the two key components: storage devices and servers.
Interoperability must always be considered because putting a random server and storage together sometimes won’t result in a resilient hyperconverged solution. In a true hyperconverged model, conjoining the software and the hardware should not limit any of these components’ functionality and performance. The software provider should do the interoperability testing between a large selection of server and storage hardware as well as virtualization platforms, providing you with the choice to deploy hyperconvergence on any x86 server, using any hypervisor, and utilizing any combination of storage devices. Some vendors develop proprietary hardware to ensure the interoperability of all components within a solution. This approach leads to a vendor lock-in which is undesirable by most
There’s another side of the coin though. Having both storage and compute resources in one box, you may often end up with situations where you have a surplus of one of those resources. I guess that it’s not that good for snug IT budgets… Often, one runs out of storage very fast. The solution may be hooking up some storage boxes on-site. Alternatively, you can just switch over to higher capacity disks or just extend your infrastructure to public cloud.
Conclusion
Hyperconvergence is all about choices. Admins know that they should always rely on their experience. Slapping together a random server and some storage components into a rack and installing some OS and SDS on top sometimes won’t deliver the gains you seek. You need to think through many factors and decide which solutions would be better for your infrastructure. Try various blends of hardware and software but make sure you look through hardware compatibility lists first. Of course, one can buy an appliance and save yourself a hassle. Vendors always make sure that hardware (server and storage) and software run smoothly in appliances. It is a good idea if you don’t have enough experience. Otherwise, if you have any preferences, you can just use the components that you are familiar with.
