Can I 'Mix and Match' Hyperconverged Appliances?
In recent weeks, we’ve held a series of webinars helping people to understand the basics around hyperconverged infrastructure and also providing a forum in which attendees can ask any questions that they may have about the technology. There have been a number of questions that arose that we simply did not have time to answer live.
This is one: Do you have to pick your hyperconverged appliances from one supplier or can they be mixed and still provide one management platform?
I’m not sure if the inquirer is curious about trying to combine appliances from totally different manufacturers or if he was wondering if it was possible to implement different appliances from different distributors with all of the appliances being the same flavor. I’m going to answer this question in two parts.
Mixing and Matching Between Manufacturers Is a No-Go
Individual hyperconverged infrastructure vendors, including SimpliVity, Nutanix, Scale Computing, Pivot3, Maxta, Atlantis Computing, and more all have significant and proprietary intellectual property wrapped up in their individual software components, and each solution supports different hypervisors. For example, Nutanix supports vSphere, Hyper-V, and KVM while SimpliVity supports vSphere and Scale Computing supports KVM.
So, all of the solutions have very different architectures. Nutanix and Scale Computing take a 100% software-based approach; SimpliVity is almost all software, but they include an accelerator card that does some of the really heavy lifting, providing comprehensive deduplication and other services without having to worry too much about the performance overhead of these services.
So, the short answer is that it is not possible to mix and match between vendors. In order to use solutions from different hyperconverged vendors, you would have to implement each as an independent silo. Perhaps you could run Nutanix for VDI and SimpliVity for the server farm, for example, but you can’t add a Scale Computing node to a Nutanix cluster nor can you directly combine any of the vendors together.
Combining Between Distributors
Let’s say bout bought four appliances from the same manufacturer, but from different resellers; two from one reseller and two from another. In that scenario, everything will work just fine.
But, let’s look at one special use case: VMware EVO RAIL. There are a number of vendors selling prepackages VMware EVO RAIL appliances. So, in theory, you could buy two EVO RAIL nodes from Dell and two from HP. I can’t see any technical reason that you wouldn’t be able to combine all four nodes in the same EVO RAIL cluster. The hardware in an EVO RAIL implementation isn’t really differentiated. The software is the key. Although it should be technically possible to mix EVO RAIL hardware sourced from different vendors, doing so would probably make support difficult, so it’s not a path I would recommend.
Summary
Bear in mind that hyperconverged infrastructure is partially intended to bring some uniformity and simplicity to the data center environment. In most cases, mixing and matching hardware between vendors isn’t going to work, not would it be the best option even if it was possible.