Voda Scheduler
Voda Scheduler is free, open-source software for scheduling GPU resources used by elastic deep learning workloads in Kubernetes.
What Voda Scheduler is
Voda Scheduler is free, open-source software for scheduling GPU resources used by elastic deep learning workloads in Kubernetes. Built around Kubernetes, Kubeflow Training Operator and Horovod, it can scale distributed training jobs up or down while they run. Scheduling choices include FIFO, Elastic-FIFO as the default, SRJF, Elastic-SRJF, Tiresias, Elastic-Tiresias, FfDL Optimizer and AFS-L. Topology-aware scheduling and worker migration can consolidate resources to improve cluster throughput and use idle capacity during heavy demand. The scheduler tracks node additions and deletions, works with an existing autoscaler, and is designed for failing nodes and spot instances that may disappear. Deployment is described for cloud or on-premise Kubernetes clusters able to schedule GPUs. Documentation links to API endpoints and Prometheus metrics exposed by the scheduler. Related projects include Kubeflow MPI Operator and a Prometheus NVIDIA SMI exporter. The repository uses an Apache-2.0 license. The README says it was tested with Kubernetes v1.20 and warns that the project is experimental and may change significantly.
Good
- Runtime scaling without interrupting training
- Several scheduling algorithms
- Topology-aware worker migration
- Cloud and on-premise deployment
- Apache-2.0 open-source license
Watch out
- Project is experimental and may change significantly
- Tested with Kubernetes v1.20
- Requires a Kubernetes cluster able to schedule GPUs
- Self-hosted deployment
Who it suits
It suits teams running elastic or distributed deep learning workloads on Kubernetes GPU clusters. It is relevant to operators needing autoscaler awareness, migration and fault tolerance.
Verdict
Voda Scheduler targets Kubernetes operators balancing GPU workloads that change size or encounter node churn. Its experimental status and Kubernetes prerequisite should be part of any deployment decision.
The facts
Each one read from Voda Scheduler's own pages, with the date it was checked.
What it does
- API
- The project documentation links to API endpoints as part of its getting-started steps.checked 8 Oct 2026
- Changing nodes
- The scheduler is aware of node additions and deletions, works with an existing autoscaler, and is designed to tolerate failing nodes and spot instances that may come and go.checked 8 Oct 2026
- Compatibility
- The project says it was tested with Kubernetes v1.20.checked 8 Oct 2026
- Deployment
- The project says Voda Scheduler is designed to deploy in any Kubernetes cluster, on-cloud or on-premise, that can schedule GPUs.checked 8 Oct 2026
- Elastic training
- It supports dynamically scaling distributed training jobs up or down at runtime without interrupting training.checked 8 Oct 2026
- Fault tolerance
- The project lists fault tolerance as a feature.checked 8 Oct 2026
- Intended users
- The stated use case is elastic or distributed deep learning workloads running in a Kubernetes cluster.checked 8 Oct 2026
- License
- The GitHub repository identifies an Apache-2.0 license.checked 8 Oct 2026
- License and maturity
- The repository lists an Apache-2.0 license and warns that everything is experimental and may change significantly at any time.checked 8 Oct 2026
- Maturity
- The README warns that everything is experimental and may change significantly at any time.checked 8 Oct 2026
- Monitoring
- The README links to documentation for Prometheus metrics exposed by the scheduler.checked 8 Oct 2026
- Node changes
- It is aware of node additions and deletions, works with an existing autoscaler, and tolerates failing nodes.checked 8 Oct 2026
- Prerequisite
- The README says the scheduler was tested with Kubernetes v1.20.checked 8 Oct 2026
- Purpose
- Voda Scheduler is a GPU scheduler for elastic deep learning workloads built on Kubernetes, Kubeflow Training Operator, and Horovod.checked 8 Oct 2026
- Related tools
- The README names Kubeflow Training Operator, Kubeflow MPI Operator, Horovod, and a Prometheus NVIDIA SMI exporter among related projects.checked 8 Oct 2026
- Resource allocation
- The scheduler aims to use idle resources and allocate resources efficiently when a cluster is heavily loaded.checked 8 Oct 2026
- Scheduling algorithms
- The listed algorithms include FIFO, Elastic-FIFO (the default), SRJF, Elastic-SRJF, Tiresias, Elastic-Tiresias, FfDL Optimizer, and AFS-L.checked 8 Oct 2026
- Topology awareness
- Topology-aware scheduling and worker migration can consolidate resources to maximize cluster throughput.checked 8 Oct 2026
