In this talk, I discuss several levels to think about scheduling workloads in Kubernetes, moving from the top level kube scheduler to custom plugins and controllers and down to partition based scheduling. For this work, I talk about an experimental prototype to schedule to specific applications running in a Flux Framework MiniCluster. I call this "pancake elasticity" because different containers can each use some elastic portion of a node's total resources, to switch off easily at the jurisdiction of the user submitting the request, and orchestrated by a top level fluxion service.
For the demo, click in the top right or here: • Pancakes Elasticity Demo
Негізгі бет Ғылым және технология Pancake Elasticity: Scheduling to Application Containers in Kubernetes 🥞️
Пікірлер