Kubernetes Event-Driven Autoscaler KEDA Graduates from CNCF Incubator

The Cloud Native Computing Foundation (CNCF) has announced the graduation of Kubernetes Event-Driven Autoscaling (KEDA), a single-purpose event-driven autoscaler for Kubernetes that can be easily added to Kubernetes clusters to scale applications. KEDA supports scaling workloads or scheduling long-running executions through Kubernetes Jobs and provides various authentication mechanisms for source integration.

It fills the gaps for use cases where the standard Kubernetes autoscaler, horizontal pod autoscaling (HPA), is not the right fit as they need to be scaled out on a different set of metrics, like custom metrics or metrics defined and exposed in a different system. The project was started as a collaboration between Microsoft and Red Hat in 2019 and has been used in production by more than 45 organizations, including FedEx, Grafana Labs, KPMG, Reddit, Xbox, and more. It has added more than 60 scalers, or connectors, to external services and now supports nine authentication providers. The community has also added new contributors, maintainers, and features and collaborated with CNCF's TAG Environmental Sustainability to reduce its environmental footprint using efficient autoscaling and workload scheduling. As a graduated project, the KEDA team plans to focus on improving performance, multi-tenant installation, and monitoring and observability features, as well as providing extensibility to streamline end-user experience for platform builders on top of KEDA. It plans to add capabilities to configure the scaling behavior and metrics evaluation, carbon and energy consumption to the scaling evaluation, and predictive autoscaling.