Pavan Kumar Madduri

Learn how GPU-aware autoscaling helps Kubernetes respond faster to AI workloads, reduce idle GPU costs and protect inference performance. Key Takeaways Kubernetes’ CPU-based autoscaling can miss overloaded GPUs, causing latency spikes and SLA violations. Direct NVML-based scaling cuts GPU metric latency from 15-30 seconds to roughly 2-4 seconds. Workload-specific scaling profiles can improve performance while reducing the cost of idle GPU capacity. Your vLLM pod shows 8% CPU usage.

Most Kubernetes tutorials explain networking with boxes and arrows: Pod A talks to Pod B. But what happens inside those arrows? Which Linux kernel subsystems move the packet? How does a Service IP — which exists on no interface — actually route traffic? I spent two weeks tracing packets through a live Kubernetes cluster using tcpdump, ip route, iptables-save, conntrack and bpftool.

Crossplane v2 (released in late 2025) introduced a cleaner, namespaced resource model and removed a lot of the v1 ceremony around Claims and cluster-scoped composites. Upgrading the control plane to v2 is usually painless — if you're not using the v1 features that changed, your existing claims keep working thanks to backward compatibility. The hard part is the next step: migrating your existing v1-style workloads onto v2-style namespaced resources.