Imagery Majestic - Fotolia

Hyper-convergence: Infrastructure under the CIO microscope

Hyper-converged technology promises costs savings and efficiency. CIOs and other executives outline where the benefits are -- and where they aren't.

Editor's Note: This is part one of two articles examining the emergence of hyper-convergence infrastructure in the enterprise. Part 1 will consider the technology's benefits, while Part 2 will tackle the question of cost savings. 

Hyper-convergence infrastructure promises to shake up data center economics, reduce the IT maintenance burden, shrink the amount of space equipment occupies and provide a smoother path for adding resources as an organization grows.

That's the ambitious technology agenda product vendors now promote and CIOs have started to put to the test. At the center of the discussion is a technology architecture that combines computing, storage, networking and virtualization in the same box -- hence the hyper-convergence label. An IT manager can purchase the key infrastructure components in a single appliance rather than purchasing them separately. This can  make hyper-converged infrastructure (HCI) cheaper than conventional IT, while reducing the data center "footprint" and alleviating the care-and-feeding demands of multiple, discrete systems.

But does HCI technology live up to its billing? And, in particular, do the economics of HCI subvert traditional on-premises technology architectures and the popular public cloud?

The answer, based on interviews with CIOs and industry executives, is far from settled. Interviews with CIOs, IT analysts and other industry executives paint a complex and highly nuanced picture of HCI.

Hyper-convergence infrastructure: Benefits

HCI, first adopted by the enterprise in 2014, competes against the traditional, three-tiered data center architecture in which enterprises purchase computing, storage and networking separately and integrate them together. Hyper-converged technology also contends with converged infrastructure offerings, which typically prepackage multiple vendors' separate systems to work as one. The cloud provides yet another possibility for IT departments.

Peter O'Halloran, CIO, National Blood AuthorityPeter O'Halloran

A range of enterprises, from government authorities to technology service providers, have deployed hyper-converged technology, typically starting with a limited installation rather than an end-to-end data center overhaul.

In Australia, the National Blood Authority, a statutory agency that coordinates the supply of blood and blood products on behalf of the Australian government, has rolled out hyper-converged technology in conjunction with a virtual desktop infrastructure (VDI) deployment. VDI transfers operating systems and applications from individual PCs, laptops and other devices to the data center. From there, desktop software is streamed to users. The approach aims to provide secure remote access to applications.

The National Blood Authority adopted Nutanix's HCI technology nearly a year and half ago, replacing a traditional storage-area network in the process. The Nutanix equipment hosts the authority's VDI, which runs blood management and patient registry systems. The ability to provide rapid access to blood and patient information is critical, given the shelf life of blood is 42 days for red blood cells and five days for platelets.

The commissioning time was incredibly quick. We were ... shocked how easy and how fast it was.
Peter O'HalloranCIO, National Blood Authority

Peter O'Halloran, the National Blood Authority's CIO, said VDI helps avoid $10 million annually in cost attributable to blood wastage.

"The contribution of the Nutanix equipment is core to that, in that it hosts our VDI sessions," O'Halloran explained.

The Nutanix gear also proved easier to deploy than conventional hardware, resulting in a labor cost reduction that made the VDI rollout possible.

"The majority of the commissioning work and migration to VDI was performed by our existing internal resources -- the difference that Nutanix made was in a significant reduction in the commissioning and maintenance time," O'Halloran said. "We would not have been able to gain the additional resources required to move to VDI if we had to cost in the extra time to build up a traditional back end."

O'Halloran said the traditional approach would have called for blade servers, storage and networking among other components, which would have taken four to six weeks to install. The hyper-converged gear, in contrast, was put to work within two to three days.

"The commissioning time was incredibly quick," O'Halloran said. "We were ... shocked how easy and how fast it was."

Other hyper-convergence infrastructure adopters have noted similar advantages. IDC, in a Nutanix-commissioned report, interviewed 13 Nutanix customers in 2015 and found those deployments would produce benefits "worth an average of $2.22 million per year over five years."

While this evidence points to the potential for financial benefit, studies that directly compare the cost of HCI to other architectural approaches have only recently started to emerge.

HCI vs. Cloud: Not an either/or proposition

Peter O'Halloran, CIO with Australia's National Blood Authority, is looking beyond its initial HCI deployment, which focuses on virtual desktop infrastructure.

Mark Campbell, director of innovation research, Trace3Mark Campbell

At one point, the authority was looking to the cloud for its future IT requirements. But its experiment with VDI and Nutanix HCI technology changed its outlook.

"Based on the performance and efficiencies received to date, Nutanix provides a very compelling alternative to cloud-based hosting ... for our mainstream -- non-test, non-development -- infrastructure needs," he said.

But cloud and HCI need not be an either/or proposition.

Mark Campbell, director of innovation research at Trace3, suggested hyper-converged technology, the concept for which originated among cloud platforms such as Facebook and AWS, can readily co-exist with the cloud. Trace3 focuses on data center, big data and cloud projects with clients.

"It isn't hyper-converged vs. cloud," he said. "Often we are seeing hyper-converged plus the cloud. I don't have to pick between those two."

For example, an enterprise may offload data older than one year to a cloud archive such as Amazon Glacier to keep the data profile fresh in its hyper-converged environment, Campbell said. Or, an enterprise using HCI could tap the cloud for bursting to accommodate seasonal changes in customer demand. Backing up HCI data to the cloud is another cooperative possibility.

Cost comparisons

In January 2016, the Evaluator Group, a company that evaluates IT products and markets, published a total-cost-of-ownership (TCO) study, pitting SimpliVity, a hyper-convergence product vendor, against Amazon Web Services' public cloud.

The report, "Is Hyperconverged Cost-Competitive with the Cloud?," notes the cloud's popularity has made it "easy to assume that running IT applications on premises is not cost-competitive with a cloud-based service," but adds that "the emergence of hyper-converged infrastructures fundamentally changes the TCO economics."

The Evaluator Group compared a three-node configuration of SimpliVity's OmniCube CN-3400 product, which supported 206 virtual machines, versus an AWS Elastic Compute 2 configuration that also supported 206 virtual machines (VM). AWS' cost was based on the company's upfront pricing plan in which a customer would write a check for three years' worth of cloud capacity. The three-year upfront plan is Amazon's lowest cost option, according to the Evaluator Group.

The result? The IT analyst firm's analysis puts the monthly cost of AWS at $62.46 per VM versus $59.67 for the hyper-converged option, a difference of about 4%. The HCI economics improve vis-à-vis AWS as more VMs are added over time, according to the report. (See report for the full details of the cost comparison.)

While HCI proponents argue the math undercuts the romantic view of the cloud as the low-cost IT alternative, others are skeptical of such TCO calculations.

Any number generated in a hyper-converged-to-cloud or hyper-converged-to-on-premises cost comparison "will be completely and utterly unrealistic," said Ben Woo, storage and big data analyst at Neuralytix, an IT analyst firm in San Francisco.

"Each company looking at these comparisons will have different needs," he noted. "There is no average."

Woo also questioned the validity of comparing HCI to infrastructure as a service (IaaS), a cloud product vendor such as AWS provide.

"To compare HCI to IaaS is comparing apples and oranges," Woo said. "They serve different needs."

Next Steps

Find out how a fully hyper-converged data center might work.

Read these tips for pursuing hyper-converged technology.

See how different hyper-convergence infrastructure offerings compare.

Dig Deeper on Small-business infrastructure and operations