As enterprise technology has evolved in recent years, it has largely moved toward flexibility – a greater ability to meet an organization’s needs, when and where it needs them, and to scale to as these needs change.
Then came the pandemic and related office closures and redefined the meaning of “need”.
And that’s where hyperconverged infrastructure (HCI) comes in. The concept isn’t new, but it’s been gaining traction in recent years, especially with regard to the public sector, and especially since the beginning of the pandemic.
In a nutshell, it’s thanks to the flexibility of HCI. But it’s not always the best choice, and it’s not necessarily going to take over every IT organization.
WHAT IS HYPERCONVERGED INFRASTRUCTURE?
HCI is, to put it as simply as possible, a computing architecture comprised of software that orchestrates network, storage, and compute resources. It’s a paradigm by which an IT department can manage these fundamental resources in one place, and do so in ways that allocate resources more efficiently. This is the “converged” part; While organizations traditionally managed storage, network, and compute separately, HCI works with all three simultaneously.
“Essentially, we’ve taken the complexity out of mining compute resources, storage resources, network resources, and software-defined data plane resources and done it for you,” said Kapil Bakshi, Director Principal and Distinguished Engineer at Cisco. Government technology in an interview. “It’s all made for you and given to you, all you have to do is start filing application services, directly dealing with the opex part of the equation.”
Here’s another part of HCI’s flexibility: it doesn’t need to run, necessarily, in the cloud or on-premises. It can be in multiple clouds, on premises, anywhere. Vendors will describe HCI as a “cloud operating model” that can run regardless of location and can be configured to meet specific needs.
“The customer now has the choice of public cloud or public clouds,” said Mark Chuang, product marketing manager for cloud storage and data at VMware. “They can select multiple vendors they want to use, to avoid lock-in, but they get this consistent operating model because it’s the same stack.”
WHY USE HYPERCONVERGED INFRASTRUCTURE?
There are many reasons why HCI started gaining popularity. It enables more efficient use of resources, which can improve application performance because it enables better coordination between network, storage, and compute. Lee Caswell, senior vice president of product and solutions marketing for Nutanix, likened it to a video recorded on a person’s cell phone instead of being streamed on the internet.
“It turns out there are big differences in performance,” Caswell said. “So you know that yourself, don’t you?” If you want to access videos on your phone instead of accessing videos sitting in a [server]or you want to send a video to a friend, it takes time.”
There is also something to be said for the consistency of HCI. That is, because HCI is software-defined, it allows for consistent application of security principles as well as familiarity with IT personnel. In other words, an IT employee can work with the infrastructure in the same way from one application to another or from one agency to another, even if the applications are built very differently.
“Suppose a government entity says, ‘Great, I have this new application and it uses files. And then someone else comes along — “Oh, by the way, we’re deploying this other app and it’s using blocks” or “I want to get into…objects”. The idea that you have the flexibility to go and deploy to one of these storage constructs is actually really helpful,” Chuang said. “It’s liberating for clients to say, ‘I didn’t have to plan this, but now I can take advantage of all the training I’ve had, all the investment I have, and I can manage these different types of storage mechanisms.'”
CONTINUITY OF OPERATIONS, RESILIENCE AND RECOVERY
Another major benefit of HCI is its contribution to making an organization more resilient, enabling business continuity and recovery from natural hazards or cyberattacks.
There are several reasons for this. One is the flexibility to locate resources in different locations. Traditional COO strategies have called for, for example, storing backup data in a data center outside an area prone to earthquakes or other natural disasters; HCI allows this location to be done through software and the location to be changed.
“You can have totally separate geolocation,” Cisco’s Bakshi said. “It could be the east coast, the west coast, an earthquake-free region, a flood-free region, or you could have it in two different power grids. And these are all architectural design considerations that you can provide so that HCI clusters are geographically separated in different data centers connected through the network.
Hyperconvergence naturally also includes the ability to quickly restore data. An IT organization can set its fault tolerance to the appropriate level and automatically import virtualized data in the event of, for example, a leak in the data center.
It can also introduce a new ability to plug in different solutions from different vendors, making it quicker and easier to replace damaged technology because the organization has more choice.
“You’ve effectively taken the power of virtualization, and now instead of just applying it to CPU and memory, you’re now applying it to storage virtualization as well as networking,” Chuang said. . “So once you’ve virtualized, you don’t have to worry about specific underlying hardware – you can switch vendors and different hardware.”
WHEN WAS HCI INTRODUCED?
This is a difficult question, especially since not everyone has always agreed on the definition of hyperconvergence, including network resources. But all three sources interviewed for this story agreed that it was recent – generally speaking, major vendors started offering products in this area in the early to mid-2010s. Nutanix set out to first in 2009, with others like VMware and Cisco arriving closer to mid-decade.
All agreed that HCI has become much more popular in recent years, especially with recent trends towards edge computing, faster network speeds and a push towards digitalization.
“Once you realize something is reliable, then you look and say, ‘Well, OK, cool, now how fast is that?’ And that’s what’s happening right now, based in part on these new flash components that are available,” said Nutanix’s Caswell.
WHY IS HCI NOT GOOD FOR?
Since many vendors are excited about hyperconvergence, it doesn’t seem likely to support everything. There are limits to its usefulness.
It has to do with how HCI is designed. It’s intended to help IT organizations coordinate networking, storage, and compute resources at a high level, so it naturally benefits use cases involving all three. But if an organization were to simply need additional compute resources, say, HCI might not make the most sense.
“If you are looking for very large industrial computing applications…that have very large databases that require dedicated compute or large dedicated storage, then [HCI] wouldn’t serve that purpose very well in this case, since they scale vertically. Essentially any application that would scale vertically would be suitable for a different build of HCI,” Bakshi said. “Applications that tend to be virtualized or containerized and tend to scale horizontally will be a great fit for HCI.”