Last week, SimKube v1.0 was released. This project leverages KWOK (Kubernetes WithOut Kubelet) to simulate Kubernetes scheduling and autoscaling behaviour.

Announcing this release, its author, David R. Morrison, provides an example of using SimKube to compare how autoscaling works in Kubernetes Cluster Autoscaler vs. Karpenter. To run a simulation with SimKube, you need to collect a trace from your production cluster and create virtual K8s nodes (managed by KWOK) with fake Pods. SimKube also features:

- Autoscaling support (with Kubernetes Cluster Autoscaler or Karpenter);
- Collecting metrics from your simulations;
- Exporting traces to Amazon S3, Google Cloud Storage, and Azure Storage;
- Hooks to run arbitrary scripts at different points.

▶️ GitHub repo
📢 v1.0 announcement

UPDATE (September 19th): the second part of the SimKube 1.0 announcement was published. It covers running the simulations to compare Kubernetes Cluster Autoscaler & Karpenter.

#tools #news