Volcano: A Unified Scheduling Platform for Cloud Native AI
Talkat KubeCon + CloudNativeCon + OpenInfra Summit + PyTorch Conference China 2026 · 2026-09-07..09-09·
summary
Volcano presents a unified Kubernetes-native scheduling stack for AI workloads, covering training, inference/LLM serving, and agent runtimes. It details Volcano-Global for multi-cluster training, Kthena as an LLM serving layer (rolling updates, router observability, prefix-cache routing, vLLM/SGLang support, plus planned autoscaling, session affinity, token-cost-aware balancing), and AgentCube for agent runtimes (microVM session routing, warm pools, PicoD execution, SDKs). Infrastructure improvements include gang preemption, sharding, namespace queues, GPU sharing, HyperNode topology, and throughput/queue management as part of a converged cloud-native AI scheduling roadmap.
AI-written summary · about
related talks
08mNext-Gen AI Orchestration With Volcano On Kubernetes
43mRun Your AI Workloads and Microservices on Kubernetes More Easily and Efficiently...
30mCloud Native AI: Harness the Power of Advanced Scheduling for High-Performance AI/ML Training
32mVolcano: Intro & Deep Dive
24mIntro To Volcano: Cloud Native Batch System
28mEfficient Scheduling Of High Performance Batch Computing For Analytics Workloads With Volcano
26mIntro: Kubernetes SIG
37mImproving Performance of Deep Learning Workloads With Volcano