How Salesforce migrated from Cluster Autoscaler to Karpenter across their fleet of 1,000 EKS clusters

This article details how Salesforce successfully migrated 1,000+ EKS clusters from Cluster Autoscaler to Karpenter, an open-source Kubernetes node provisioning solution.

• Salesforce faced scaling delays, poor resource utilization, and operational complexity with traditional Auto Scaling groups
• Developed automated migration tools ensuring zero-disruption transitions with rollback capabilities
• Automated configuration mapping converted 1,180+ diverse node pool configurations to Karpenter definitions
• Phased rollout strategy prioritized lower-risk environments before production migration
• Discovered critical issues: restrictive Pod Disruption Budgets, Kubernetes label length limits, single-replica pod protection needs
• Achieved 80% reduction in manual operational overhead through automation and self-service capabilities
• Scaling latency reduced from minutes to seconds; node utilization improved significantly
• Realized 5% cost savings in FY2026, with 5-10% additional savings projected for FY2027

The migration demonstrates how modern Kubernetes auto-scaling solutions can transform enterprise platform operations while maintaining reliability at massive scale.