Busting Myths: Hyperconverged Pros and Cons

Old perceptions die hard, and in a field that’s evolving as rapidly as hyperconverged infrastructure, they become full-blown myths all the more quickly.

In its short life, HCI has undergone metamorphosis. The first systems were integrated hardware appliances used primarily for delivering virtual desktops over a network. Today, HCI is a full-featured software stack that virtualizes servers, storage devices, and storage networks into one centrally managed and highly scalable unit.

But that reality doesn’t stop myths rooted in the early days from spilling over to the present. The Forecast by Nutanix asked some HCI experts to address a few of them.

Myth #1: HCI is Just for Low-Volume Uses.

This perception hails from the days when HCI makers were just trying to get IT executives to sit still for an explanation of their new technology.

“When we started Nutanix, we targeted virtual desktop infrastructure [VDI] because it wasn’t very I/O-intensive and customers were open to something new,” said Manosiz Bhattacharyya, Nutanix senior vice president of engineering.

It turns out, VDI may be the most difficult of all use cases because it requires satisfying the demands of thousands or 10s of thousands of users, each with their own experiences, expectations and perceptions.

“Although the first use case was VDI, Nutanix HCI can handle more demanding I/O workloads than VDI," Bhattacharyya said.

The scale-out architecture is designed for massive growth because there is no one control node to create a bottleneck. “We have many large customers like the federal government that use Nutanix for high-volume tasks like processing passports and collecting taxes,” Bhattacharyya said. 

Steve McDowell, senior tech analyst at Moor Insights & Strategy, added that the low-volume myth might have been true in 2014 when HCI was an appliance. 

“But today’s HCI is about a unified management experience for all your resources,” he said. “It has evolved to unify a set of functions and deliver them as an abstraction with much more enterprise hardware underneath.”

Bhattacharyya believes the myth continues to be perpetuated by storage companies that want to niche HCI into a low-performance category so they can sell costly high-end storage. “HCI actually handles storage in many ways better than tiered storage networks because it takes out the network factors,” he said. Read activity is local and never traverses the network, he explained.

Myth #2: HCI is a Commodity, So Go with the Cheapest Solution.

That’s like saying you should buy a Mitsubishi Mirage instead of a Lexus or Tesla, said Steven Kaplan, vice president of customer success finance at Nutanix. While the Mirage may be perfectly appropriate for some use cases, your investment decision “depends on what your objectives are,” he said.

For small, stand-alone applications that don’t materially impact the business, the cheaper solution may be adequate. But if you demand a scalable, high-performing, self-healing environment, then look more closely at the specs. “You don’t want 5,000 users unable to work one day because your system is down,” Kaplan said.

Moor Insights’ McDowell believes buyers should choose a platform to match the problem they’re trying to solve. “What HCI does is democratize the platform to allow you to pick and choose the right one for your needs,” he said. “I can put a screaming high-performance system next to a print server and manage them as a whole.”

Myth #3: HCI Can’t Scale to Handle Enterprise Applications.

“That’s rich,” said Lucas Mearian, an International Data Corp. research manager who specializes in infrastructure. “HCI is certainly an excellent platform for business-critical workloads such as VDI, but because of its high availability, it’s also well suited for database workloads.”

HCI is becoming a de facto standard for data-intensive services like data protection, backup and disaster recovery and is increasingly a primary platform for deploying modern hybrid cloud infrastructure, Mearian said.

“Many enterprises are rolling out a hybrid cloud strategy from single-tenant HCI infrastructure and then investigating how to leverage the public cloud from there to support workload portability,” Mearian said. “As the basis of a hybrid cloud environment, HCI can create a consistent experience across all platforms, whether on-premises or in the cloud.”

It’s widely acknowledged that real-time business needs can’t be satisfied with traditional three-tier infrastructure.

Myth #4: The Server Uses Too Many System Resources, Especially Cores and Memory.

In truth, HCI’s aggregate volume of cores and memory is less than that of a conventional three-tiered storage network, Bhattacharyya said. “We use less CPU and memory because we don’t have to drive a lot of network I/O,” he said. 

“Three-tiered storage systems rely on “a black box where you don’t know how much compute and memory you are using. You’re using more resources, but it’s not visible.”

McDowell said this myth “is the same one that virtualization fought 15 years ago, which is that any time you put something on top of bare metal, there’s some cost.” In reality, virtualization improves performance by ensuring that virtual machines have the resources they need.

“By and large, HCI stays out of the way while you’re doing the work you need to do and only activates when you need it for management,” McDowell said. “It’s a wrapper and nothing tells me that it impacts performance at all.”

Myth #5: Choosing an HCI Platform Leads to Vendor Lock-In.

That may have been true at one time, but HCI today is a software platform that runs on a wide range of hardware. “You aren’t locked into any server vendor,” Bhattacharyya said. “We expose our storage protocols and you’re free to use them or move between them.” In contrast, migrating between different storage array vendors requires special software and a transport mechanism.

McDowell said the choice of a virtualization platform involves a certain level of commitment, but customers make that choice knowing the implications. With HCI, “there’s no hardware lock and you can move workloads to any other virtualized environment,” he said.

Myth #6: HCI Won’t Last.

Given that Research and Markets expects the global HCI market to grow more than 26% annually through 2026 and Gartner forecasts that the percentage of enterprises running some form of hyperconverged infrastructure will grow from 30% to 70% from 2019 to 2023, this myth seems almost laughable.

HCI is likely to persevere because it conserves network bandwidth, which is a precious commodity in edge computing scenarios.

“HCI inherently reduces your network traffic, meaning there will always be an economic argument for it,” Bhattacharyya said. What’s more, “it’s the only architecture that works well both on bare metal and in the cloud.”

Edge computing, a market that’s expected to more than quadruple to nearly $16 billion by 2025, will be HCI’s next frontier, IDC’s Mearian believes. 

“It will enable companies to consolidate various edge workloads onto a single platform, permit unprecedented scaling on standard hardware and enable processing at the ingestion point,” he said. 

The result is that applications like artificial intelligence at the edge will be viable. 

“All data needed for the AI application is near the HCI system, meaning no latency and faster processing,” Mearian said.

It may be overly optimistic to believe that all these myths about HCI will entirely disappear, but it helps to have the facts to counter them. In fact, HCI is not just optimal infrastructure for on-premises data centers, said Kaplan, pointing to Nutanix Clusters. 

“It’s also serving as the basis for an optimal hybrid cloud strategy.”

Paul Gillin is a contributing writer. He is the former editor-in-chief of Computerworld and founding editor of TechTarget. He’s the author of five books about social media and online communities. Find him on Twitter @pgillin. 

© 2021 Nutanix, Inc. All rights reserved. For additional legal information, please go here.