Red Hat’s CTO Explains Edge Computing Buildout

Red Hat is building out its suite of tools—including its container platform OpenShift and container-native virtualization (CNV) packages, which are add-ons to the OpenShift platform—to enable clients to develop and better manage edge computing projects as part of their broader cloud strategies.

Edge computing is a category of distributed computing that takes place close to the relevant data it needs to process—rather than at the server or cloud location. By bringing compute power closer to the location it is needed, Chris Wright, chief technology officer at Red Hat, says that in the world of capital markets, edge computing can reduce latency on the trading floor, decrease strains on bandwidth, and allow for more complex real-time data projects.

In one area, the IBM subsidiary is working with clients to scale down the number of Kubernetes clusters being deployed in edge locations and produce smaller footprints of computing activities for the OpenShift platform to run on.

Red Hat’s OpenShift platform is built on Kubernetes, an open-sourced system for deploying, scaling, and managing containerized applications. Within the Kubernetes system are clusters, or machine nodes that run applications. By scaling back the size of Kubernetes clusters, the vendor hopes to improve the processing power and speed of computations carried out in the edge environments.

Wright says the open-source technology provider is focused on helping clients deploy and refine edge environments to support low-latency processing, real-time analysis, and data-heavy workloads.

“A pattern is emerging across the industry of deploying a larger number of smaller clusters, maybe focused on certain application types, or production versus development environments, and the ability to manage all of those clusters consistently is a key requirement for our customers’ success,” Wright says.

In April, the vendor introduced a new offering called Advanced Cluster Management, which can manage multiple clusters of Kubernetes across the OpenShift network and distributed locations. That same month, at the Red Hat Virtual Summit, the vendor announced the rollout of OpenShift CNV, a set of extensions designed to enable firms to manage applications across on-premise, cloud, and edge-computing locations. This is a step forward, as the virtualization and system management capabilities now extend to non-containerized applications and bare-metal, on-premise servers. 

“Bringing OpenShift to bare metal is something that we’ve been working on for quite some time and it enables this edge use case. Combined with OpenShift Virtualization, you create that flexibility of the types of applications you can run in an edge deployment. We have been making the tools you need to develop an edge infrastructure available and it’s something that we’re continuing to iterate, improve, and refine,” Wright says.

During the Summit, the vendor announced that it is working with Goldman Sachs as part of the investment bank’s multi-year pilot program to modernize its customized virtualization environment and leverage the use of container-native virtualization. The partnership will help the bank develop a virtualization environment and allow it to manage its virtual machines effectively.

Red Hat is also exploring the use of edge computing to deploy complex models. In this scenario, Wright explains that trained models can be deployed onto an edge environment, where it can access real-time data through inference, as it is located closer to the relevant data center within the hybrid or multi-cloud environment.

The idea is that specialized physical edge environments can be used to process machine-learning workloads faster and more effectively.

“If you go all the way down to the hardware level, a machine learning environment is typically dependent on not just some software tools, but specific types of hardware that accelerate machine learning environments to make it practical to do work in it in a meaningful timeframe,” Wright adds.