Horizontal Pod Autoscaling
A control loop that adjusts a workload's replica count toward a target metric, computing desiredReplicas = ceil(current * observed/target) from Metrics Server or custom metrics. Stabilization windows damp flapping; because the math divides usage by per-Pod requests, HPA is meaningless without resource requests set.
This Concept is waiting for its first lesson!
A control loop that adjusts a workload's replica count toward a target metric, computing desiredReplicas = ceil(current * observed/target) from Metrics Server or custom metrics. Stabilization windows damp flapping; because the math divides usage by per-Pod requests, HPA is meaningless without resource requests set.
Are you a teacher? Sign in to start contributing.
Sign In