Spend more than an hour in a meeting with any major software company and you’re bound to hear the buzzword “hyperconverged infrastructure,” but what is it, and why should you care? Industry analyst Zeus Kerravala explained it for us in a question-and-answer session. We played the role of skeptic.
SEE: Virtualization policy (Tech Pro Research)
TechRepublic: We think we understand what hyperconverged infrastructure means, but how would you explain it?
Zeus Kerravala: “It’s kind of a weird term. There was already a converged infrastructure market [lacking the software aspect] when this technology came around. Hyperconverged platforms are turnkey products that include all the hardware and software one needs to run a contained little data center in a box. … When you look at running data center infrastructure there’s a lot of different choices for buyers. If you use Cisco networking, EMC storage, and Dell computing, which is a pretty standard thing, there’s over 800 configurations. [In HCI] the vendor’s done all the heavy lifting. They’re not plug-and-play… it’s data center technology, nothing’s ever going to be plug-and-play. But customers have told me the deployment time for these is days vs. months if you’re trying to cobble it all together yourself.”
TechRepublic: Do you think most corporate sysadmins and CIOs understand this?
Zeus Kerravala: “I’m not sure the CIO does. I think technology has been somewhat niche. It’s been used primarily for virtual desktop deployments. Those are workloads that tend to be demanding… unified communications are a likely next thing. I don’t really understand where the ‘hyper’ came from, to be honest with you.”
TechRepublic: Most good ideas in information technology are cyclical. How much of this is truly novel and how much is just a new name?
Zeus Kerravala: “We used to have converged platforms a long time ago, and we called them mainframes. The reason the hyperconverged market exists is to simplify the deployment of all the stuff we need to run data centers.”
TechRepublic: Why is this happening now?
Zeus Kerravala: “I talk to CIOs. More and more, CIOs are less concerned about the technical aspects of running stuff. They want stuff to work so they can run the business. There’s a theme of digital transformation that’s cutting across all businesses. If you talk to a CEO about running a business, it’s about speed today. It’s Darwinism.”
TechRepublic: What are the risks of changing from traditional to hyperconverged infrastructure?
Zeus Kerravala: “I haven’t really talked to anyone who hasn’t had a good experience [except] using the technology for the wrong workloads. If you’re going to run hyperconverged infrastructure, the development has to be done on a product that’s at least similar from a hardware perspective.”
TechRepublic: What about hardware upgrades?
Zeus Kerravala: “Applications that have the most demanding hardware requirements, I’d probably keep those on a platform that I have a little bit of control over, and I can upgrade the processors when I need to. If you wouldn’t run it on a virtual machine, then certainly don’t run it on this.”
TechRepublic: Which companies are offering hyperconverged infrastructure the right way?
Zeus Kerravala: “The market leader right now in terms of brand and share is still Nutanix. They’ve done a lot of work in software. The one to watch is Dell/EMC but for specific use cases [such as with VMware’s vSphere]. If your hypervisor is Microsoft or Citrix, then I might look at a different platform… 8kpc is a startup. They’ve done a lot of work on the hardware optimization phase.”
TechRepublic: Which companies aren’t doing so well at it?
Zeus Kerravala: “I think HPE is bit of a confused company right now… Lenovo is another one that I’ve expected more of by now.”
TechRepublic: What is your advice for customers considering a hyperconverged infrastructure product?
Zeus Kerravala: “There’s a lot of products on the market and they all kind of pitch the same message. But the performance from box to box, from vendor to vendor, is going to be quite different depending on what you’re running on it. Do your own testing. How does it work in a hybrid cloud situation? I’d also want to know from a roadmap perspective about flash storage, 100Gb Ethernet, and then NVMe.”
TechRepublic: What else should people know?
Zeus Kerravala: “Try to have a good understanding of what it means to the operational team. Things may be easy to deploy initially but take a look at the ongoing management. That’s really going to determine whether you get value out of these products or not.”