Better together: IBM Research and Red Hat make OpenShift enterprise-ready

Through this collaboration, Red Hat and IBM have been able to use our joint expertise to make an even greater impact on our clients.

Red Hat now has some 15 engineering scrum teams that include IBM Research partner engineers. Some of these projects are closer to the product phase, while others are longer-term research efforts. For example, in July of 2020 our teams started collaborating on the core container platform to further-boost resource management in OpenShift, the application development platform on OpenShift serverless and observability, and application services on cloud-native Java (a project called Read more about the cloud native journey to Red Hat OpenShift using Quarkus.Quarkus), as well as cloud security.

It all started with open source

In July 2020, we launched a formal collaboration effort — matching up the expertise we needed between our teams to work as partner engineers. It’s been just over a couple of years, but we’ve already achieved a lot together.

We’ve worked on eventing capabilities in OpenShift serverless to enable event-driven, serverless applications on OpenShift. We’ve jointly launched Konveyor, an open-source project aimed at modernizing and moving enterprise applications to the open hybrid cloud.

Our extension to the Linux Foundation’s open source sigstore project, which Red Hat considers strategic, allows developers to sign, verify, and enforce cloud infrastructure configuration. And then there is Iter8, a Kiali extension that automates the progressive rollout of new microservice versions of applications and analyzes the performance of competing versions.

Collaboration goes both ways

Take Red Hat OpenShift. Based on Kubernetes, it provides a consistent foundation for developing, deploying and managing cloud-native applications and services across the hybrid cloud.

Kubernetes has a generic, open-source resource manager that’s a great default scheduler for general application workloads. It struggles, though, with specialized workloads when it comes to AI, machine learning and analytics. And this is where IBM Research and Red Hat come in.

For years, IBM had been developing more advanced workload schedulers for use in hybrid cloud environments. When the two companies came together, Research presented the work it had already been doing on more sophisticated scheduling with Kubernetes. Red Hat showed IBM how to apply some of that work to drive client use cases on OpenShift.

Working as one team, Red Hat and IBM Research turned their combined efforts into two components:

• One is Trimaran, a set of load-aware scheduler plugins that rely on the actual use on the worker nodes — something Kubernetes doesn’t take into account.
• And then there is the Vertical Pod Autoscaler (VPA), a controller that lets developers automatically resize containers in real time during runtime.

PayPal, the global payments giants, is one of the joint customers putting this new load-aware scheduling capacity into production.

IBM and Red Hat both continue to contribute to open source projects. For example, some clients needed OpenShift to manage virtual machine workloads, as well as containers. For that, IBM Research made key contributions to OpenShift Virtualization, based on the open-source project KubeVirt.

IBM and Red Hat are now collaborating to build compatibility between IBM Digital AI Cores and Red Hat OpenShift. Digital AI Cores serve as accelerators — using custom architecture, software and algorithms — to transform existing semiconductor technologies to speed computation and decrease power consumption while maintaining model accuracy. We also plan to enable AI hardware accelerator deployment across hybrid cloud infrastructure: multi-cloud, private cloud, on-premise and edge.

It’s now been three years, and there is still so much to do together. As IBM’s hybrid cloud and AI strategies evolve, this partnership will continue to contribute to the open source community, while providing better tools for our enterprise customers.