nobeastsofierce - Fotolia

Does hyper-converged infrastructure improve VDI scalability?

Hyper-converged systems are a big investment, but they also provide the flexibility to add or subtract resources to fit the demands of a VDI project -- from beginning to full implementation.

BOSTON -- When introducing a technology such as VDI to the workforce, IT administrators need the ability to scale up and down, depending on demands and adoption rate. Hyper-converged infrastructure provides back-end flexibility that can help a VDI project meet a company's needs.

One of the major benefits of using hyper-converged infrastructure (HCI) is that it combines compute, storage, networking and virtualization resources all into one tightly integrated, software-defined system. An admin can upgrade the entire system at once by adding a single node, rather than sorting through a traditional siloed setup.

"You have all of the components of the traditional infrastructure consolidated into a single device, so that allows us to quickly and easily scale out the environment," Sammie Ginger, a principal engineer at SimpliVity Corp., based in Westborough, Mass., said in a session.

HCI vendors Nutanix and SimpliVity both hosted sessions here at last week's VMware User Group UserCon event, and VMware offered a hands-on lab for one of its two HCI products -- the still unreleased EVO SDDC.

Features of hyper-converged systems

Scaling up an entire system at once is also one of HCI's weaknesses. A company might have to buy a whole new node just because it needs more of one specific resource. With SimpliVity, IT can add an extra compute node separate from the other resources. This can be especially important if you're scaling out users with 3D requirements, said Jason Shiplett, another principal engineer at SimpliVity.

"They're not only going to require more compute than your typical user, but you have to get those Nvidia GRID cards in there somewhere," Shiplett said. 

The company's OmniStack HCI platform allows organizations to start off with as few as two nodes -- each can support from 200 to 250 workers -- and add more nodes from there, which makes it relatively easy to do "a linear scale out from pilot to production," Ginger said.

SimpliVity hasn't released a reference architecture yet -- Shiplett said it should be out soon -- but its chief competitor, Nutanix, already has a reference architecture available for its Acropolis platform.

Nutanix also offers its own feature that makes scaling out a VDI project easier with its VDI Assurance program. The organization selects the number of end users that need virtual desktops, and Nutanix takes care of sizing the necessary amount of server and storage infrastructure. If its calculations are off, Nutanix will make up the difference at no cost. This way, organizations only invest in enough infrastructure to fit their needs, and gain some level of cost and performance predictability, as they scale their VDI project up or down.

"We all know that the mystery of VDI is that users don't adopt right away or at the pace we would expect them to, so being able to scale up is super important," Michael Berthiaume, a systems architect at Nutanix, based in San Jose, Calif., said in another session.

It's hard to predict how VDI projects will grow, but using hyper-converged systems -- such as OmniStack, Acropolis, or VMware EVO:RAIL or EVO SDDC -- can help put admins in position to handle whatever comes next.

Next Steps

Why should VDI shops use hyper-converged infrastructure?

Will hyper-converged systems affect VDI admin jobs?

How VMware EVO:RAIL and EVO SDDC impact VDI

Dig Deeper on Virtual desktop infrastructure and architecture