Storage Volumes

Persistent network storage for GPU instances.

Storage Volumes

Network storage volumes provide persistent block storage that survives instance termination. Mount volumes across different instances to share datasets, model checkpoints, and other files.

Creating a Volume

curl -X POST https://api.gpuworker.com/volumes \
  -H "Authorization: Bearer YOUR_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "name": "training-data",
    "size": 100,
    "dataCenterId": "US-TX-3"
  }'

Data Center Locations

Volumes must be created in the same data center as the instances that use them.

RegionIDLocation
US TexasUS-TX-3Dallas, TX
US OregonUS-OR-1Portland, OR
EU AmsterdamEU-NL-1Amsterdam, NL
EU StockholmEU-SE-1Stockholm, SE

Mounting Volumes

When creating an instance, attach a volume by specifying the volume ID. The volume mounts at /workspace inside the container.

Volumes can only be attached to one instance at a time. Detach from the current instance before attaching to another.

Pricing

Storage is billed at $0.07/GB/month regardless of whether the volume is attached to an instance.

Best Practices

  1. Store datasets on volumes — Keep training data on network volumes so it persists across instances
  2. Regular snapshots — Important data should be backed up externally
  3. Right-size volumes — Start small and resize as needed to control costs
  4. Same data center — Ensure volumes and instances are in the same region for best performance