Table of contents:
- Some Vendors Are Retreating, and Software Products Are Getting Bigger
- HCI “Disaggregates”?
- HCI: Always Good For Peripherals?
- HCI As A Service
- HCI And Containers
HCI promises to simplify IT by consolidating storage and computing, and typically virtualization environments, into a single system or compliance. This one-box approach leverages and integrates the flexibility of virtualization and networked storage.
The result, its proponents argue, is a flexible, high-performance system that is suitable for small businesses, branch offices, and edge applications. In addition, vendors are increasingly looking to provide hyper-convergence through software and, especially, software-defined storage systems.
It is noteworthy that as soon as some vendors decided that HCI was unacceptable to them, others entered the market, especially with software-defined offerings. HCI is now firmly established as a premium option, especially for applications where ease of operation is important.
But the market is changing. This is partly in response to customer demand and the growing focus on cloud computing and services during the pandemic, and this trend is likely to continue.
Below are some of the key trends in the hyperconverged systems market.
Some Vendors Are Retreating, And Software Products Are Getting Bigger
Industry analyst firm Emergen Research estimates the HCI market at $7.34 billion in 2020 and forecasts a compound annual growth (CAGR) of 26.8% through 2028. Drivers include backup and recovery and improving application performance.
According to Naveen Chhabra, Senior Analyst at Forrester, HCI is being used for virtual desktops (VDI), databases, analytic workloads, and virtual machine farms.
While hyperconverged systems are used primarily in data centers, the technology is becoming more versatile as vendors improve the controls available to IT administrators, he says.
However, HCI is best suited for workloads that use scale-out (adding more nodes) rather than scale-up (adding more CPU, storage capacity, memory, etc.).
Perhaps surprisingly, given the overall market interest, some vendors are losing interest in hyperconverged solutions. NetApp has dropped (direct) support for HCI, and some analysts note that other vendors are paying less attention to hyperconverged solutions than they used to.
Nutanix remains the most prominent HCI vendor, with other vendors with a strong presence: Cisco, VMware, Dell EMC, Microsoft (via Azure), and Huawei.
Gaining strength and options based only on software. Nutanix is now more focused on selling independent software than proprietary hardware. Scale Computing, StarWind, and Pivot3 are vendors to keep an eye on.
Stormagic, with its software-defined virtual SAN, is also often associated with hyperconverged infrastructure projects.
Initially, hyperconverged infrastructure attracted attention because its key components were integrated. This allows IT teams to rapidly deploy systems and reduce management overhead. This integration is an important part of HCI’s appeal, especially to branches and peripherals.
However, HCI scales better horizontally than vertically. Vertical scaling, for example, for large-scale transactional databases, is not its forte.
“If your application requires vertical scaling, you should not consider hyperconverged systems,” says Chhabra. For the most part, monolithic systems are less suitable for HCI (although there are exceptions, such as running SAP Hana on HCI). The reason is that HCI typically cannot scale its (component) resources independently.
This encourages vendors to “disaggregate” (separate) HCI components to make them easier to scale. VMware, for example, allows users to share storage across HCI clusters with its HCI Mesh system. Users can also connect to Dell storage subsystems by connecting VxRail through their vSAN infrastructure.
HCI: Always Good For Peripherals?
HCI solves some of the challenges organizations face when deploying technology in small offices, branch offices, or remote locations. There may not be dedicated IT teams or even dedicated facilities for larger, more complex equipment.
At the same time, a hyperconverged system should be suitable for edge applications, especially if it comes in a robust compliance form factor. Eliminating the need for separate storage, computing, and networking hardware also reduces power consumption and cooling needs.
In addition, the use of a single supplier means fewer parts. Perhaps this is an exaggeration: hyperconverged systems can be complex, but IT departments need to be able to control all of their systems with a single management tool.
But, as Chhabra warns, there is no single sectoral definition of the term “periphery”. Vendors consider a wide range of use cases, and those who claim to be able to serve all of them are best avoided.
“For some HCI users, it’s a retail outlet, or a branch office, or a telco base station, or an outback installation,” he says. “The question is, can all HCI implementations be made to fit into this wide variety of deployments? The answer is no.”
A CIO should start by looking at the HCI usage scenario in their edge environment and then see which vendors have the best deal for a given workload and configuration.
HCI As A Service
The “as a service” approach is being promoted by HCI vendors such as Dell EMC with its VxRail, which in turn is part of the Dell Apex product line. Cisco offers a service option with HyperFlex. HPE Greenlake customers can use Nutanix Era or Microsoft Azure Stack HCI.
This operating system is also available for IT teams to purchase directly from Microsoft. It includes Hyper-V for computing, Storage Spaced Direct, and a software-defined networking module. HPE offers SimpliVity HCI, which it says is an HCI optimized for an edge, VDI, and “general virtualization.”
The rise of hyperconverged solutions as a service may be driven more by vendors seeing the ability to provide resources on a subscription basis than by advances in technology.
It probably makes sense for organizations buying infrastructure as a service (IaaS) to purchase HCI in the same way. If the workload is more amenable to horizontal rather than vertical scaling, then scaling cloud infrastructure based on adding nodes should reduce management overhead.
It can also make it easier to replicate on-premises HCI workloads to the cloud. With this in mind, CIOs need to evaluate whether HCI is the right fit for the workloads they want to migrate to IaaS or to the public cloud in general.
One of the advantages of the cloud is the ability to buy computing and storage resources separately, increasing or decreasing them as needed. HCI, as a service, loses some of its inherent flexibility.
However, vendors are investing in an “as a service” model, and this should make it easier to fine-tune hyperconverged instances for different workloads and combine this with the ability to leverage HCI’s operational rather than capital costs.
HCI And Containers
Container support is one area where HCI is clearly growing rapidly.
Established hyperconverged vendors such as Cisco, Nutanix, and VMware support containerized workloads, as does IBM. Spectrum Fusion works with Red Hat’s version of OpenShift Kubernetes, and Nutanix also works with Red Hat.
The ability to support containers and hypervisors expands the applicability of HCI. As more and more cloud – or cloud-native – applications are being developed for containers rather than virtual machines, this support is becoming more important to customers.
HCI was not designed for containers, but vendors are adapting hyperconverged nodes to support the flexibility required by environments such as Kubernetes. For example, Google Anthos is a hybrid cloud powered by NetApp technology. Kubernetes applications are available on the Google Cloud Platform marketplace.
As more enterprise applications move to containers, HCI vendors are expected to follow suit.