Hyperconverged Infrastructure

Hyperconverged Infrastructure in 2020: Continuing to Revolutionize IT

Way back near the end of the first decade of the current millennium, hyperconverged infrastructure (HCI) burst into the market; it spent the second decade of the millennium transforming organizations and IT. And now, as we enter the third decade, it’s clear that, as HCI continues its evolution, it will continue to disrupt and transform organizations for the better.

To better understand where HCI is headed and how it will continue to chip away at legacy IT, let’s look at what it’s done over the past decade since its inception. Over the past decade, HCI has:

  • Upended the storage industry
  • Commoditized enterprise storage features
  • Initiated a price war that has redefined data center economics

Looking back, it’s easy to see that HCI was a logical evolution, and the result of a culmination of trends that all happened to hit at the same time. The first trend, of course, was virtualization. That, combined with burgeoning compute capability thanks to massive multicore processors and the revolution of flash storage, gave rise to a solution that became the perfect complement to another emerging trend: a desire to simplify what had become unworkable complexity in legacy IT infrastructure.

In my opinion, hyperconvergence stands out as the key disruptor in the past decade.

In my opinion, HCI stands out as the key disruptor in the past decade. It opened minds as to what was possible, and exposed just how hungry enterprises had become for simpler IT constructs.  In fact, in the first quarter of 2019, converged systems was a burgeoning category, approaching $2 billion in spend.  Now, as we head into 2020 and beyond, we’re heading into a new era for this transformational technology.

Common Themes in 2020 Predictions for Hyperconverged Infrastructure

CRN published a hyperconverged infrastructure predictions article by Mark Haranas. In it, Mark raises some commonly-held predictions around HCI’s future.

First and foremost, Haranas believes that HCI is becoming the “backbone” of the multi-cloud world, a viewpoint he shared with IDC’s Eric Sheppard. Here, I agree with their viewpoint and I believe that this is a trend that will continue to solidify in 2020 and beyond—but possibly not to the extent that many believe.

Traditional approaches to infrastructure will refuse to die quietly, and with good reason.  Vendors that provide modern renditions of legacy solutions have learned from the mistakes of their predecessors and have brought new ways of thinking into their product lineups.

For example, whereas legacy SANs were often convoluted disasters when it came to trying to figure them out, modern solutions from modern vendors take an approach much closer to set-it-and-forget-it than anything that came before.

It seems that even HCI vendors are getting this message, especially vendors with other storage products in their portfolios. Such vendors have a focus on simplicity, at least on paper, as organizations become ever wearier of the hamster wheel that sometimes defines IT.

At the same time, there are economic considerations for organizations considering HCI, but not just in the direct monetary sense. Economic considerations can also include operational improvements, which, according to multi-year research performed by ActualTech Media, is a key expected and achieved outcome for the technology.

But there’s a whole lot more coming from HCI, the foundations of which are already here. Haranas makes the following predictions in his article, and we’re going to provide some of our own thoughts around each one:

1. The hybrid, multi-cloud world has found a foundation that will continue to solidify.

For many, HCI already forms the basis for an entry into the multi-cloud world.  With scalable economics and improved operational efficiency, HCI is well-suited to most data center workloads.

In recent years, there has been a movement toward differentiating HCI solutions with a more platform-centric approach. For example, Nutanix now positions its HCI solution as the center of a broader platform that both extends into the public cloud, and brings some of the public cloud goodness to on-premises environments. This includes providing ways for customers to better manage such environments and contain costs.

We would also put VMware and Dell in this platform category.  VMware clearly has a complete hybrid cloud platform play, and provides everything from hardware to software to partnerships with Amazon and Azure to make good on it.

2. Backup and disaster recovery use cases for HCI will continue to expand.

This is another safe bet. Storage demands never stop growing. Backups and disaster recovery grow with them. Although it’s an obvious point, it’s still worth noting.  HCI solutions increasingly include some level of backup and recovery, to varying degrees.  Some solutions have custom-crafted services, while others have deep partnerships with more traditional backup providers. Still others provide both options.

For many, the built-in backup and recovery capability is more than sufficient, and will continue to be so. This effectively lowers the total cost of ownership for the solution. But it isn’t for everyone; if HCI is but one part of a broader technology environment, the organization may prefer to include HCI in its overall data protection strategy.

Disaster recovery (DR) is a separate service, but it often bundled with data protection. To this end, HCI solutions may also offer DR capabilities aside from, or as a part of, their data protection offerings. Alternatively, they may partner with best-of-breed DR providers.

3. Edge and micro data centers will soon quadruple.

This is an understatement, but the nuance revolves around how you define the term data center.  It’s clear that the proliferation of the edge shows no signs of slowing down and, for a certain type, HCI is the perfect choice. 

There are a number of reasons for this, including compactness and ease of deployment and maintenance. Already, many HCI vendors are pivoting toward branding themselves as suppliers of edge computing systems. We even see some, like Scale Computing, betting the business on the edge and creating sprawling HCI-centric environments that are easily stamped out of the factory and deployed.

4. Will AWS overthrow hyperconverged leaders with Outposts?

This is yet another one of those things that’s been discussed ad nauseam within the industry for years. Azure Stack, Amazon Outposts, Google Anthos… we’ve known for some time that the big public cloud providers are integrating with on-premises data centers. It’s a big topic, and one we’ll discuss in depth later in this piece.

The Importance of HCI to the Midmarket

One thing that’s not on Haranas’s predictions list is the importance of HCI to the midmarket. It’s a rather important thing to point out. Yes, everyone targets the enterprise, but many of the most loyal HCI customers have been and will continue to be midmarket organizations that aren’t yet ready to bet their entire business on public cloud providers.

We expect to see a great deal of the growth in the HCI market occur in the midmarket. Price wars are driving down costs into ranges that midmarket companies can afford, and this is coinciding with increasing pressure to automate in the midmarket. HCI offers a quick and easy way to meet infrastructure automation requirements without having to pay too much, or overly disrupt existing operations.

We expect to see a great deal of the growth in the hyperconverged infrastructure market occur in the midmarket.

In terms of enterprise adoption of HCI, this seems to depend entirely upon who holds the crystal ball. An Evaluator Group study suggests that enterprises may have reservations about HCI, although other studies that support the opposite view. One place HCI is selling well is schools, and things that sell well into schools tend to sell quite well into the midmarket.

Perusing the many and varied market analyses on HCI, the “where” of HCI really seems to matter. Enterprises may start to care more about HCI in the context of edge deployments, where small footprints and ease of management may become more important.

To contrast, midmarket companies aren’t really doing the edge-computing-in-bulk thing yet, meaning their adoption of HCI is more likely to involve replacement of systems in their central data center. In other words, HCI is still nibbling at the edges of many enterprises, while at the same time taking over mission-critical workloads for the midmarket.

AWS Outposts and Azure Stack: The Cloud Fog Now Permeates the Data Center

The extension of public cloud services into the on-premises data center is inevitable. There’s nothing that technically prevents the technology stacks which power the world’s public clouds from being packaged into a “cloud in a can,” then sent to operate on-premises, where local latency and/or storage requirements may make transporting bits across the internet somewhat impractical. Azure Stack and AWS Outposts are exactly this.

Microsoft has had Azure Stack available for some time now, but it has been overshadowed by the hype around AWS Outposts. There are several reasons for this, though it takes some unpacking to uncover them.

Part of the reason for the excess of hype around Outposts is simply that Amazon has been purposefully building hype. Another reason is that Azure Stack is already an existing product family, and it hasn’t lived up to the hopes that were built up around it during the pre-release hype phase.

In creating Azure Stack, Microsoft sought to work with their existing partners (such as HPE, Dell, and so on) to deliver Azure Stack in a way that would protect those partners’ existing revenues. Azure Stack was only to be deployed by a select few trusted partners, in limited configurations; priced far higher than the market wanted it to be; and launched with a limited palette of services.

What customers actually wanted was to tap into the economies of scale and the prolific use of whitebox components that have driven costs down for the public cloud providers. Enterprises wanted inexpensive hardware married with the same software and services the public cloud providers use, but on-premises, driving down the costs of IT by lowering both CapEx (through lower cost of hardware) and OpEx (by consuming applications as a service). All without the costs, privacy, or data sovereignty risks of entrusting their data to a public cloud provider. Microsoft has delivered on only some of the dream, and the rest does not look to be on their roadmap.

Both Azure Stack and AWS Outposts have the potential to represent an existential threat for the entire on-premises IT industry, including HCI vendors. HCI vendors married compute and storage into a single product, and have been working to extend their reach into the cloud via software instances integrated with those on-premises environments.

Azure Stack and Outposts, on the other hand, are the cloud’s way of reaching back into the on-premises data center while retaining the cloud envelope. But HCI enjoys first-mover advantage in some ways and, while transformational, it’s still a familiar construct, so it enjoys some level of protection in that sense.

Although the hype factor around them is high, we see Outposts and Azure Stack as existential threats to HCI vendors that fail to keep pace with the needs of customers.

Ultimately, 2020 will be the year in which the direction that these technologies go is really determined.

Extension of Networking, Automation, and Orchestration

HCI started life as a simple mashup: bring together servers, storage, and a hypervisor and a new construct was born. But over time, it needed more. Eventually, VMware upped the ante by incorporating Software-Defined Networking (SDN), along with microsegmentation, which brought network-based information security as close to the workload as possible. Other vendors have done similar things, and have their own inherent networking capabilities and partnerships that help them handle the needs of the network.

We are seeing even more signs of network disruption among HCI vendors, too.  Recently, for example, Scale Computing announced its HCI Edge Fabric, a technology that eliminates the need for network switches in particularly small edge environments—an area of intense focus for the company.

In addition to its own Nutanix Flow VM network policy and microsegmentation service, Nutanix offers additional SDN capabilities through integrations with leading providers, including Big Switch, Arista, and others.  HPE has brought its Composable Fabric to its HCI lineup.  And they’re far from alone.  Other HCI vendors have brought their own networking feature set to the mix, too, and we expect to see many others follow.

“With advances in software-defined networking technology, we’ll see hyperconverged infrastructure move from ‘converged compute and storage with networking as an afterthought’ to fully hyper-converged infrastructure [that includes networking at its core].”
~Dan Frith, PenguinPunk.net

Disaggregated Hyperconverged Infrastructure

In addition to adding new capabilities around data protection, DR, and networking, some HCI vendors are also taking a close look at the architecture itself and decoupling the way it’s built, with the outcomes that people want to achieve.

In some cases, customers are really after simplicity, but they want a little more granularity around the components.  This has given rise to what the market is calling disaggregated hyperconverged infrastructure.  In many ways, this is a marketing term, but in real ways, it’s a forking of HCI with a focus on delivering a solution that has the management simplicity of hyperconverged infrastructure.

“Users finally figured out that ‘easy’ was the real outcome they wanted, not necessarily hyperconverged technology itself. With all the bolt-on or different solutions to accommodate Cloud, DevOps, and BR/DR, those hyperconverged infrastructure vendors don’t necessarily look as easy anymore, which erodes their primary market. Vendors that deliver outcomes through technology that are easy to consume and exploit will continue to lead the market. Nobody really cares what’s under the hood (or even where it resides) unless it affects that outcome.”
~Glenn Dekhayser, Field CTO for Red8

At the core, the physical architecture resembles a scale out storage environment and for good reason. That’s exactly what it is. In these environments, compute workloads run on separate nodes from storage workloads and each can scale independently.

Disaggregated HCI can consist of disk trays that don’t have any native compute, and controllers which are effectively compute nodes. This is different from HCI, in which compute and storage live in the same node.  We’ll continue to see the expansion of these kinds of services in 2020 and beyond, and we expect some “traditional” HCI players to adopt a more disaggregated look and feel as software continues to become an increasing focus point and further abstracts operations away from hardware.  This will continue to make it easier to customers to adopt such technologies.

The Battle Rages On

There is a clear path forward for HCI in the 2020s, but it’s just one front in a battle that continues to increase in intensity. HCI could be an enabling technology as part of a broader strategy by traditional on-premises vendors, or it could be a casualty of a revolutionary change in IT delivery toward the utility computing model that has been talked about for so long.

How much of the on-premises market share will the public cloud providers take? Will they become the first truly global utility providers? Purveyors and gatekeepers of the world’s IT? It’s impossible to know how that battle will play out, but we suspect that, for the foreseeable future, HCI adoption will continue to grow and the technology will continue to evolve.

Resources

Here some additional resources around HCI and cloud services that you may find useful as your organization heads into the next decade.