How Salesforce efficiently right-sizes Kubernetes workloads at scale on Amazon EKS
Containers Blog
This article describes how Salesforce efficiently right-sizes Kubernetes workloads at scale on Amazon EKS by combining Vertical Pod Autoscaler (VPA) and Horizontal Pod Autoscaler (HPA) for multi-dimensional autoscaling.
- Moved from static resource management to dynamic autoscaling across 1,400+ EKS clusters with 2,500+ nodes each
- VPA alone for workloads scaling by resource size; HPA alone for stateless workloads; VPA+HPA together for workloads needing both dimensions
- Implemented operational safeguards including staggered rollouts, CI/CD integration, rollback mechanisms, and comprehensive monitoring
- Achieved over $5 million in cumulative infrastructure cost savings through optimized resource reservations
- Key lessons: use VPA for steady-state optimization not traffic spikes, start with Off mode, configure independent metrics for VPA and HPA
- Future plans include in-place resource resizing for DaemonSets and AI/ML workload optimization
By combining VPA and HPA with proper testing and guardrails, Salesforce reduced infrastructure costs while maintaining workload stability across its massive Kubernetes fleet.
The AWS News Feed is currently looking for gold sponsors. If you want to support the AWS community and reach a large audience of AWS professionals, consider sponsoring the AWS News Feed.
Related articles
2025
2025
2024
2025
The AWS News Feed is currently looking for silver sponsors. If you want to support the AWS community and reach a large audience of AWS professionals, consider sponsoring the AWS News Feed.