Defaulting to Karpenter for New Clusters, and Cutting Over Cleanly When Migrating an Existing One

Decision drivers

Whether the cluster is new (no legacy autoscaler commitment) vs. an existing Cluster Autoscaler deployment being migrated
Presence of a compliance framework requiring native ASG lifecycle hooks/dynamic scaling policies
Static/homogeneous vs. spiky/diverse compute shape
Whether Spot is in use, and whether the SQS interruption queue is actually wired up
Presence of long-running batch/stateful workloads sensitive to node churn, and whether Pod Disruption Budgets are configured for them

Cluster Autoscaler trades provisioning speed and Karpenter's flexibility for maturity, ASG-native compliance hooks, and predictable behavior on static, homogeneous fleets. Karpenter trades that ASG abstraction for direct EC2 Fleet API calls — bypassing Auto Scaling Groups entirely, which is both its main advantage (no 450-node-per-ASG ceiling, tighter bin-packing, per-workload instance-type selection) and its main compatibility constraint (no ASG lifecycle hooks, no ASG-native dynamic scaling policies).

Gated · a free account unlocks the full catalog

This page is part of the AWS track's gated content. Sign in — free — to keep reading. A free account unlocks the full catalog.