Edge computing and Kubernetes: Understanding the landscape

James has a passion for how technologies influence business and has several Mobile World Congress events under his belt. James has interviewed a variety of leading figures in his career, from former Mafia boss Michael Franzese, to Steve Wozniak, and Jean Michel Jarre. James can be found tweeting at @James_T_Bourne.

Edge computing differs from cloud computing in that information is processed nearer to the source of data. Yet the need for greater speed and efficiency in maximising resources as ever-increasing amounts of data get utilised are a common source of pain in both technologies.

It’s not just the data, either. In edge use cases, which range from a security video camera to a remote oil rig, conditions may not always be optimal. As a result, the complexity needs to be managed. Kubernetes, as the de facto standard for container orchestration, is a well-regarded solution for cloud – and increasingly so in edge.

The benefits of containerisation will be well-known to anyone versed in cloud technologies, from agility, to portability, to flexibility. Software developers can create and deploy applications faster and more securely – write once, run everywhere – and allow organisations to modernise legacy applications and move towards cloud-native software.

Yet it is by no means easy, and trying this at the edge means another environment with its own issues, as edge computing specialist ZEDEDA explains. “The edge is complex both in terms of technology and logistics,” the company explains. “Edge solutions require a multi-layer stack developed and operated by a variety of providers and skillsets. Integrating this diverse landscape without an open anchor point can result in siloed and fragmented solutions at the expense of flexibility and vendor lock-in.”

A solution for deploying Kubernetes at the distributed edge, ZEDEDA adds, needs to address the challenges of flexibility, scale, security, and orchestration. Selecting the right Kubernetes distribution to meet hardware and deployment requirements, and then working out how to manage this Kubernetes cluster infrastructure in the field at scale, is another major headache.

A recent report from analyst GigaOm explored the landscape for edge computing and Kubernetes, and the vendors at the vanguard of this area.

The GigaOm Radar Report uses two either/or segments to assess where companies sit in a bullseye-like diagram; maturity versus innovation and feature versus platform. The ideal position is dead centre, a perfect balance between the two areas. Vendors are designated as either leaders, challengers, or new entrants.

Ultimately, the GigaOm report put seven companies in the leaders’ category, though most of them are at the edge between leader and challenger; in alphabetical order, NodeWeaver, Platform9, Rafay, Red Hat, Spectro Cloud, SUSE (Rancher), and ZEDEDA. A further two vendors, Microsoft and VMware Tanzu, were placed in the challengers’ category, noting – particularly with Microsoft – the relative lack of maturity of their product offerings to date.

This verdict underlines the nascence of the area, with a variety of approaches and a lack of consensus. The platform-based vendors, such as Red Hat, SUSE Rancher and VMware Tanzu, are thinking beyond scalability, and according to GigaOm, believe customers will tolerate additional operational complexity as a result. Use cases and customer requirements continue to evolve.

The differences between certain vendors are interesting in their own right. NodeWeaver, for example, ‘bridges the world of private cloud platforms and hyperconverged infrastructure with a unified cloud fabric that brings all pieces together in building-block style’, as the company puts it. This means, when it comes to something like disk failure, the system will replace unavailable resources without any administrative intervention – something the GigaOm research noted, saying high availability failover was a ‘particular speciality’. NodeWeaver supports both full Kubernetes (K8s) and the more lightweight K3s distribution.

For another example, the aforementioned ZEDEDA is a vendor specifically focused on edge computing, and offers a cloud-native SaaS-based solution as a complete edge orchestration platform. The company announced support for Kubernetes at the distributed edge as far back as 2021; the GigaOm analysis noted how ZEDEDA has a particularly strong focus on security and assurance deploying into environments ‘where physical device security cannot be relied on.’

Microsoft, by contrast, only made its Azure Kubernetes Service (AKS) Edge Essentials service generally available in March. The service is an ‘on-premises Kubernetes implementation of AKS that automates running containerised applications at scale, as the Microsoft documentation puts it, and was developed in response to customer needs for Kubernetes at the operational edge ‘on small-footprint devices for orchestrating workloads and driving business optimisation.’

There is one other important offering to note, direct from the Cloud Native Computing Foundation (CNCF), which may be a case of ‘back to the future.’ KubeEdge is a distribution of Kubernetes as an open source framework for edge computing, which helps developers ‘deploy and manage containerised applications in the cloud and on the edge using the same unified platform.’ The goal is to ‘bring Kubernetes and cloud-native to edge computing’, though vendors such as ZEDEDA will note its solution supports any Kubernetes distribution.

If you are thinking of utilising Kubernetes for your edge deployments, there are plenty of ways to solve the puzzle – but it is worth making sure you know which one is right for your business.

You can read the GigaOm Radar for Kubernetes for Edge Computing here (courtesy of ZEDEDA).

Photo by Tom Fisk

Want to learn more about edge computing from industry leaders? Check out Edge Computing Expo taking place in Amsterdam, California and London. 

Explore other upcoming enterprise technology events and webinars powered by TechForge here.

View Comments
Leave a comment

Leave a Reply