K8s cost discipline
Get Kubernetes cost under control, from requests and limits to autoscaler history, idle workloads and orphaned storage.
6 lessons. 1 quiz.
Read in order, or jump to what you need.
Requests and limits: the math behind the bill
By the end of this lesson, you will be able to calculate the K8s cost impact of resource requests and limits, diagnose over-provisioning from cluster utilization data, and set requests/limits to the right p95/p99 sweet spot.
HPA: the signal it gives you
By the end of this lesson, you will be able to read HPA signals to identify right-sizing opportunities, diagnose the four common HPA patterns (at-min, at-max, oscillating, balanced), and tune HPA settings based on observed behavior.
The seven idle workload shapes
By the end of this lesson, you will be able to identify all seven shapes of K8s idle workloads, detect each via kubectl / metrics / ZopNight, and remediate appropriately (terminate, fix, leave).
Single-replica deployments: reliability vs cost
By the end of this lesson, you will be able to decide when single-replica is acceptable vs when reliability requires more, calculate the reliability math behind replica counts, and configure monitoring + recovery for single-replica workloads.
Security signals: privileged, root, host network
By the end of this lesson, you will be able to recognize the K8s security signals that correlate with cost optimization opportunities, audit your cluster for these signals, and combine security modernization with cost savings.
Orphan PVCs and released PVs
By the end of this lesson, you will be able to identify orphaned K8s PersistentVolumeClaims and PersistentVolumes, calculate the storage cost they accumulate, and execute cleanup with the right reclaim policy going forward.
Module quiz.
Take it once the lessons are done.