Kubernetes capacity planning: How to rightsize the requests of your cluster

Kubernetes capacity planning is crucial for infrastructure engineers as it involves balancing cluster stability and resource utilization, ensuring containers have enough resources without unnecessary over-allocation, which can lead to increased costs and scheduling difficulties. The article outlines methods to identify and rightsize cluster capacity using open-source tools like kube-state-metrics and cAdvisor, which help avoid resource underuse by analyzing current resource usage and adjusting requests accordingly. PromQL queries can detect underutilized CPU and memory, allowing for charge-backs to departments with oversized namespaces and identification of the top containers wasting resources. The goal is to set requests between 85% and 115% of average usage, ensuring efficient deployment and resource allocation. Measuring the impact of optimizations can be done by comparing current resource usage against previous benchmarks. Sysdig offers out-of-the-box Kubernetes Dashboards to facilitate this process, enabling users to quickly identify underutilized resources and streamline Kubernetes capacity planning without needing expert-level skills.