Building a Batch System for the Cloud with Kueue
Talkat Kubernetes Batch + HPC Day Europe 2023 · on Tuesday, 2023-04-18· at 14:15· Hall 7, Room A | Ground Floor | Europe Complex·
summary
Kueue is a cloud-native job queuing layer that integrates with the Kubernetes scheduler, job controller and cluster-autoscaler to enforce quotas, hierarchical sharing and preemption; it suspends jobs and injects nodeSelectors so a single Job object controls thousands of pods, reducing API server load. Administrators declare cluster queues with resource flavors (on‑demand vs spot) and quotas (example: 40 CPU quota with 20 CPU borrow), users submit normal Job APIs and Kueue admits or preempts across cohorts; v0.3.0 adds preemption, API stability and a control library.
AI-written summary · about
chapters (4)
related talks
33mBatch Systems in Production with Kueue: Multi-Tenancy and Fungibility
35mKueue: A Kubernetes-native Job Queueing
28mMultitenancy and Fairness at Scale with Kueue: A Case Study
24mAccelerate Your AI/ML Workloads With Topology-Aware Scheduling in Kueue
37mIntroduction to the Kubernetes WG Batch
24mVolcano – Cloud Native Batch System for AI, BigData and HPC
34mWG-Batch Updates: What’s New and What Is Next?
05mKueue: Save Some QPS for the Rest of Us! How To Manage 100k Updates Pe...