- Home
- GPU Cluster Management Software
- Koordinator
Koordinator
What it is
Koordinator is a free, self-hosted scheduling system for Kubernetes clusters. It coordinates microservices, AI, big-data, audio, and video workloads using priority and quality-of-service controls. Different workload types and pods can share a node, while application profiling supports resource overcommitment with QoS guarantees. Fine-grained orchestration and isolation address latency-sensitive workloads and batch jobs. Custom scheduling policies and profiles cover Web Service, Spark, Presto, TensorFlow, and PyTorch. QoSManager tunes pod isolation parameters using profiling, interference detection, and SLO settings. Koordinator adds control-plane and node components alongside kube-scheduler rather than replacing it, without invasive changes to native Kubernetes components. Koord-Queue supports TFJob, PyTorchJob, Spark, Argo Workflow, Ray, and native Kubernetes Jobs, with Kueue AdmissionCheck compatibility. It also provides monitoring, troubleshooting, operations tools, and YARN Copilot integration. Installation requires Kubernetes 1.18 or newer and Helm 3.5 or newer.
For and against
For
- Free plan
- Supports Kubernetes QoS and kube-scheduler
- Resource overcommitment with QoS guarantees
- Custom scheduling profiles
- Monitoring and troubleshooting tools
Against
- Requires Kubernetes 1.18 or newer
- Helm 3.5 or newer required
- Linux kernel 4.19 recommended
- v1.8 partly supported on Kubernetes 1.20 and 1.22
The verdict
Koordinator combines workload scheduling, QoS management, overcommitment, and isolation within Kubernetes. Its broad queue integrations and self-hosted deployment suit organizations managing varied workloads, provided their cluster versions meet the requirements.
Who it suits
Koordinator suits teams running mixed Kubernetes workloads that need scheduling priorities, QoS controls, resource sharing, or isolation. It is intended for self-hosted Linux environments and supports both batch and service workloads.
On the record
- CNCF status
- Koordinator was accepted to the CNCF Sandbox on April 16, 2024.cncf.io · read 1 Oct 2026
- Community support
- The project provides community support through a Slack #koordinator channel and a DingTalk group.koordinator.sh · read 1 Oct 2026
- Compatibility limit
- Koordinator v1.8 is only partially supported on Kubernetes 1.20 and 1.22 because some components require newer Kubernetes APIs, while core co-location, QoS, and scheduling continue to function.koordinator.sh · read 1 Oct 2026
- Governance
- Koordinator is a Cloud Native Computing Foundation sandbox project.koordinator.sh · read 1 Oct 2026
- Installation
- Koordinator can be installed with Helm 3.5 or newer, and the installation documentation requires Kubernetes 1.18 or newer.koordinator.sh · read 1 Oct 2026
- Integration model
- It runs with control-plane and node components in a Kubernetes cluster without invasive modifications to native Kubernetes components.koordinator.sh · read 1 Oct 2026
- Job scheduling
- Its job-scheduling mechanisms support workloads in areas including big data, AI, audio and video.koordinator.sh · read 1 Oct 2026
- Kubernetes compatibility
- Koordinator is compatible with Kubernetes QoS and its scheduler is designed to enhance kube-scheduler rather than replace it.koordinator.sh · read 1 Oct 2026
- Kubernetes requirement
- Koordinator requires Kubernetes version 1.18 or newer.koordinator.sh · read 1 Oct 2026
- License
- The Koordinator project is licensed under the Apache License, Version 2.0.github.com · read 1 Oct 2026
- Linux requirement
- The documentation recommends a Linux kernel version of 4.19 or higher for the best experience.koordinator.sh · read 1 Oct 2026
- NRI support
- NRI mode resource management requires containerd 1.7.0 or newer with NRI enabled and is enabled by default in Koordinator.koordinator.sh · read 1 Oct 2026
- Operations tooling
- Koordinator includes tools for monitoring, troubleshooting and operations.koordinator.sh · read 1 Oct 2026
- Purpose
- Koordinator is a QoS-based scheduling system for efficient orchestration of microservices, AI and big-data workloads on Kubernetes.koordinator.sh · read 1 Oct 2026
- QoS and priority
- It provides priority and QoS mechanisms to co-locate different workload types and run different pods on one node.koordinator.sh · read 1 Oct 2026
- QoS management
- QoSManager dynamically tunes pod resource-isolation parameters using profiling, interference detection, and SLO configuration.koordinator.sh · read 1 Oct 2026
- Queue integrations
- Koord-Queue supports TFJob, PyTorchJob, Spark, Argo Workflow, Ray and native Kubernetes Jobs, and is compatible with Kueue's AdmissionCheck API.koordinator.sh · read 1 Oct 2026
- Resource efficiency
- It combines elastic resource quotas, pod packing, overcommitment, resource sharing, and container isolation to improve utilization.koordinator.sh · read 1 Oct 2026
