KubeCon + CloudNativeCon India 2026

Kubernetes provides powerful mechanisms to protect and isolate workloads, but many teams deploy applications without leveraging these capabilities. In the context of AI workloads, which are often resource-intensive and long-running, maintaining resilience is more critical than ever.

This talk explores Kubernetes workload protection mechanisms and demonstrates how to combine them strategically for maximum resilience.

We'll examine resource requests/limits, priority classes, resource quotas, runtime classes, and strategies like pod disruption budgets and affinity rules. Through practical examples and real-world scenarios, you'll learn how to configure each mechanism, understand their interactions, and avoid common pitfalls that lead to pod evictions, performance degradation, and cascading failures.

Whether you're running mission-critical services or shared multi-tenant clusters, this session will equip you with a resilience framework that protects your workloads under pressure.