As hyperconverged technology matures, it’s become a clear platform for edge computing architecture.
As companies modernize, they are ditching traditional infrastructure for hyperconverged infrastructure, striving to become more agile, automated and to reduce costs.
Hyperconverged infrastructure (HCI) combines servers, storage and networking all in one box. Instead of managing and integrating separate components, hyperconverged technology brings infrastructure under one roof for simplified management.
While converged infrastructure has been around for some time, hyperconverged infrastructure provides greater capability through centralized management. This layer enables users to pool and manage resources more efficiently. A centralized management console enables data center managers to automate and simplify processes. And because of its node-based architecture, HCI enables easier consumption; companies can start small and add modular elements as needed.
In part one of our series on hyperconverged technology, we looked at the benefits and challenges associated with hyperconverged architecture. Here, in part two, we look at more innovative uses for HCI that are beginning to emerge. For instance, HCI can serve as an edge computing system. Along with the benefits come some difficulties in building such a system.
Edge computing is an architecture that moves compute processing closer to users and devices that need it rather than having processing occur centrally in an on-premises data center or a public cloud.
Sometimes it’s better for performance, efficiency, and security to move this computing to the edge. By keeping the processing as close to the source of the data as possible, you can return instant results because data doesn’t have to take a round trip back to a centralized data center or cloud, which creates latency.
Examples, including IoT systems, remote point-of-sale systems and so on. These systems can respond faster and handle data locally.
But as compute moves to the edge, it can introduce new concerns. The risk is to make something already complicated too complex to manage and operate. Note that edge systems won’t have dedicated staffers and operators, thus they need to be self-repairing and self-upgrading.
With so much data, users and devices, what we need are simple-to-install-and-operate edge systems that can be dropped in and automated such that the remote maintenance needed is minimal—from 0 to 30 minutes a year. This is doable, though, because the platform delivers high power and is on par with most cloud- and data center-centralized processing.
Hyperconverged infrastructure is the most logical alternative to achieve this data-intensive computing at the edge. Here’s why.
Hyperconverged technology’s plug-and-play approach eliminates many of the configuration and networking hassles that can be challenging at the edge. If you need to configure and install 200 edge devices all over the country, all with different types of networks and interfaces into existing local systems, for example, consultants could earn big bucks in doing custom work at each location. In reality, you can mask much of this complexity with HCI automation and software-defined infrastructure, if you’re innovative enough and do some up-front planning.
Hyperconverged technology is built from the ground up as a drop-in system. A hypervisor allows you to run multiple virtual machines (VMs) that make them easy to deploy for new applications and configurations as requirements change over time. Moreover, multi-node clusters provide redundancy to eliminate downtime, and software-defined storage eliminates the risk of data loss when drives and/or nodes fail. Finally, HCI is modular, which means you can scale-out HCI for processing, memory, and storage as requirements change.
Hyperconverged technology provides resiliency for edge deployments by logically pooling the compute resources of physical devices, thus creating a cloud-like cluster. This allows an HCI system to withstand isolated hardware failures that would stop non-HCI systems from working.
The value of using HCI is clear. It’s a fit in these kinds of scenarios:
Because hyperconverged infrastructure is software based, it can use different hardware sets to ensure that edge systems can be customized and optimized at the lowest cost. If you choose to purchase pre-configured, bundled hardware and software, making more extensive changes (such as using a specialized network interface) can be a challenge.
Edge-based HCI should support the ability to run less-powerful chipsets that are purpose-built for edge computing. The idea is to custom-build HCI edge devices to use the least amount of power and to be configured for device-specific purposes.
Since most edge computing focuses on IoT, interoperability of hyperconverged technology among various vendors and standards is important. Thus, the drop-in nature of HCI should come with pre-configured IoT connections and standards support that will quickly get the HCI edge computing systems up and running. These need to be remotely upgradeable, much like a smartphone. Considering the quickly changing nature of IoT, it’s likely that these HCI edge computing systems will be updated as often as weekly.
The other two considerations and reasons to exploit HCI for edge computing concern ROI and remote management.
HCI systems—certainly the bundled hardware and software systems—can be costly. But there are ways to minimize cost, such as by leveraging a software-only HCI system that can leverage a commodity platform. The tradeoff is that you’ll need to manage that platform. However, considering that you’ll do many deployments of the same configuration, you should be able to minimize support. A good integrator can deliver custom platforms with the HCI software pre-loaded.
The automated remote-management features are vital. If HCI edge systems don’t have uptime of 99.9999%, ROI goes out the window. HCI needs to eliminate traditional management tasks, deal with events such as network outages, and overcome the failure of physical devices without creating downtime. HCI software should handle most of this, but there will be innovative solutions needed, depending upon how much you stretch your edge computing system with customization.
HCI is a natural fit for edge-based systems, although it’s not on the radar yet for most enterprises that are considering edge-based systems. Part of this problem is the emerging nature of edge computing in general, the lack of understanding around the benefits, as well as the problems that must be overcome, such as those listed above. Your edge requirements will point you toward the right path. Just be sure to look at HCI.
David Linthicum is the chief cloud strategy consultant and a longtime contributor to a variety of technology publications.