Volcano v1.13 Released: Comprehensive Enhancement of Scheduling Capabilities for LLM Training and Inference
volcano
New Features: LeaderWorkerSet support for large model inference, Cron VolcanoJob, Label-based HyperNode auto-discovery, Native Ray framework support, HCCL plugin support, Enhanced NodeGroup functionality, ResourceStrategyFit plugin, Colocation decoupled from OS, Custom oversubscription resource names, Kubernetes v1.33 support, and more
New features: Network Topology-Aware Scheduling (Alpha), Dynamic MIG Partitioning for GPU Virtualization, DRA Support, Queue Capacity Management in Volcano Global, Security Enhancements, Performance Optimizations, Gang Scheduling for Generic Workloads, Job Flow Improvements, and Kubernetes v1.32 Support.
New Features: Network topology-aware scheduling, Elastic hierarchical queues, Multi-cluster AI job scheduling, online/offline workloads colocation with dynamic resource oversubscription, Load-aware descheduling, Fine-grained job fault recovery policies, Volcano Dashboard for resource visualization, Supports for Kubernetes v1.31, Volcano Job supports for Preemption Policy, and Performance optimizations for large-scale scenarios
New features: Support Queue Priority Scheduling Strategy, Enable Fine-Grained GPU Resource Sharing and Reclaim, Introduce Pod Scheduling Readiness Support, Add Sidecar Container Scheduling Capabilities, Enhance Vcctl Command Line Tool, Ensure Compatibility with Kubernetes v1.30, Strengthen Volcano Security Measures, Optimize Volcano for Large-Scale Performance, Improve GPU Monitoring Function, Optimize Helm Chart Installation And Upgrade Processes, etc.
New features: Support elastic queue capacity scheduling, Supports affinity scheduling between queues and nodes, GPU sharing feature supports node scoring scheduling, Volcano Support for Kubernetes v1.29, Enhance scheduler metrics, Add license compliance check, Improve scheduling stability, etc.
New features: Support for vGPU scheduling and isolation, support for vGPU and user-defined resource preemption capabilities, addition of JobFlow workflow scheduling engine, node load-aware scheduling and rescheduling support for diverse monitoring systems, optimization of Volcano's ability to schedule microservices, optimization of Volcano charts packages for publishing and archiving, etc.
New features: enhanced plugin for PyTorch Jobs, Ray on Volcano, enhanced scheduling for general Kubernetes services, multi-architecture images of Volcano, and optimized queue status info
ING Bank: How Volcano Empowers Their Big Data Analytics Platform
Use case description: Ruitian Capital, an investment company, uses Volcano to build their offline computing platform.
OpenI-Octopus: How to Avoid Resource Preemption in Kubernetes Clusters
Junmao Li/System Architect in Pengcheng Laboratory
Volcano use case in scientific research