Batch Scheduling Across Multiple Kubernetes Clusters

Kubernetes has become a standard in the industry for deploying microservice applications, but because of its architecture, one area that provides challenges is running batch workloads. 

Batch workloads consist of finite lifetime computations such as AI or scientific computing workflows, which can consist of extensive resource requirements such as high memory, CPU and/or GPU requirements. Various Cloud Native Computing Foundation (CNCF) projects have been founded for running batch workflows (Argo Workflows, Volcano, Kubeflow, etc) but there is a significant limitation in that they do not support multi-cluster environments. 

Fortunately, there is a solution! In this talk, we propose an open source application for running batch workloads across multiple Kubernetes instances called Armada. Join Kevin Hannon, an Armada maintainer working at G Research Open Source, on April 4, 2023 at 11am ET to learn more about how you can overcome this fundamental Kubernetes limitation to pass your competitors.