The Cloud Native Computing Foundation today announced the graduation of KEDA, an open-source tool created by Microsoft Corp. and Red Hat to make Kubernetes clusters more efficient.

The milestone follows more than three years of development.

“Since joining the CNCF Sandbox, the KEDA community has grown and improved the project in a sustainable way, and we look forward to seeing what it can accomplish as a Graduated project,” said CNCF Chief Technology Officer Chris Aniszczyk.

The Cloud Native Computing Foundation, or CNCF, is an open-source group that operates under the wing of the Linux Foundation. It maintains dozens of open-source technologies that are used by enterprises to power their cloud environments. The main highlight of CNCF’s portfolio is Kubernetes, the ubiquitous container orchestration framework.

KEDA, the project that graduated today, was launched in 2019 as a collaboration between Microsoft and Red Hat. The two tech giants submitted it to the CNCF in early 2020. KEDA spent most of the three years that have passed since in CNCF’s incubator program, which is designed to refine the feature set of early-stage open-source projects.

The graduation from incubation is part engineering achievement, part symbolic milestone. The CNCF promotes an open-source technology to graduated status after determining that it’s reliable enough to be used in large enterprises’ cloud environments. This is a high-profile vote of confidence that can, in theory, boost the long-term adoption of a project.

KEDA is an autoscaling engine designed to extend the default feature set of Kubernetes. In particular, it’s built to help applications that run on the container orchestration platform more easily keep up with usage spikes.

A Kubernetes-powered container application consists of building blocks called pods. A pod, in turn, is a software module that performs computations. An e-commerce application might have one pod that focuses on generating shopping recommendations and another designed to process purchases.

If the hypothetical e-commerce application experiences a traffic surge, it can add more purchase processing pods to meet the increased usage. When traffic returns to usual levels, those extra pods can be removed. The process of adding and removing software modules based on user demand is known as scaling.

Application scaling is handled automatically by Kubernetes using a built-in tool called HPA. The tool is suitable for many autoscaling use cases, but not all of them. That’s where the CNCF’s KEDA project comes into the picture.

According to its developers, KEDA is designed for situations where a Kubernetes application needs to scale in response to an event that affects a different application. This is a fairly common requirement in the enterprise. 

One use case that KEDA can simplify is analytics. A Kubernetes-based business intelligence environment, for example, might comprise two components: a database and an analytics tool that runs queries on the records in the database. If the analytics tool runs an unusually complex, hardware-intensive query, the database would need to scale to address the increased workload.

Scaling an application based on an event affecting another workload is possible with Kubernetes’ built-in HPA tool. However, the task requires writing a lot of custom code. KEDA provides the necessary code in a prepackaged form to save enterprises the effort. 

According to its developers, KEDA can perform two types of scaling. It can scale up already running workloads to help them process a sudden surge in usage. Additionally, KEDA is capable of waking up a dormant application when it suddenly starts receiving user requests after a period of inactivity.

KEDA is currently used by more than 45 organizations worldwide. The project’s installed base includes Microsoft, FedEx Corp., Cisco Systems Inc. and several other large enterprises.

Following KEDA’s graduation, the tool’s developers plan to add a raft of new features. One of the items on their checklist is a predictive autoscaling feature that will enable KEDA to more efficiently scale Kubernetes applications in response to usage spikes. Additionally, the tool’s developers plan to make user experience improvements and add more monitoring features. 

Photo: Unsplash