Kubernetes v1.35: In-Place Pod Resize Goes Stable and What It Means for Your Infrastructure

Kubernetes v1.35 brings significant advancements with the stable release of in-place pod resizing, allowing dynamic adjustments of CPU and memory resources without restarting pods. This development addresses longstanding challenges in managing stateful workloads, machine learning tasks, and long-running batch jobs by enabling resource scaling in response to actual demands while maintaining system state and reducing operational complexity. The release enhances performance and cost efficiency by avoiding the need to overprovision from the start or disrupt workloads, and it integrates with the Container Runtime Interface (CRI) to adjust resources seamlessly. Complementary technologies like live migration further optimize infrastructure by enabling workload relocation when necessary, offering zero-downtime maintenance, and enhancing geographic optimization. Additionally, new features such as Pod Certificates for workload identity and Node-Declared Features improve security and compatibility, while the PreferSameNode traffic distribution option enhances performance for latency-sensitive tasks. This version of Kubernetes marks a major step toward more adaptive and resource-efficient cloud infrastructure, with platforms like DevZero providing advanced orchestration to maximize the benefits of these capabilities.