GitOps is increasingly becoming a popular way of delivering applications to Kubernetes clusters because of the benefits it offers such as declarative and version-controlled infrastructure configuration as well as continuous reconciliation.
With a handful of Kubernetes clusters, and the same team maintaining both application development and delivery, a static association between infrastructure Git repos and specific target clusters works well. However, when infrastructure scales to accommodate business needs, enterprises end up with a few dozens to hundreds of Kubernetes clusters. Also the need to incorporate a reliable separation-of-concerns mechanism between the application development and platform teams becomes important. In such environments, static associations do not work anymore, and a dynamic way to associate your infrastructure repos to Kubernetes clusters becomes critical to business.
In this talk, we show how to augment your team’s GitOps pipeline with a multi-cluster, multi-cloud scheduler to achieve both static and dynamic placement of workloads. We illustrate, through a number of demos, how to provide a great developer experience by separating concerns of infrastructure-choice between application development and platform engineering teams.