Storage Overview
Vast.ai provides two main types of storage for your instances:- Container Storage - Temporary storage within the Docker container
- Volumes - Persistent local storage that can be attached to instances
Container Storage
Container storage is the default storage allocated to every instance when it’s created.Key Characteristics
- Size is fixed at creation: You must specify the disk size when creating the instance
- Cannot be resized: Once created, the allocation cannot be changed
- Persists while instance exists: Data remains even when instance is stopped
- Deleted with instance: All data is permanently lost when instance is destroyed
- Charged continuously: Storage costs apply even when instance is stopped
Default Allocation
- Minimum: 10GB (default)
- Maximum: Varies by host machine capacity
- Set via disk size slider during instance creation
Storage charges continue even when instances are stopped. To stop storage billing, you must destroy the instance completely.
Best Practices
- Estimate generously: Better to have extra space than run out mid-task
- Monitor usage: Check disk space regularly with
df -h
- Clean up regularly: Remove unnecessary files to free space
- Back up important data: Container storage is lost when instance is destroyed
Volumes
Volumes provide persistent storage that survives instance destruction and can be reattached to new instances.Key Features
- Local only: Tied to the physical machine where created
- Persistent: Survives instance destruction
- Reattachable: Can be mounted to new instances on same machine
- Fixed size: Cannot be resized after creation
- Separate billing: Charged independently from instances
Volume Limitations
- Cannot migrate between different physical machines
- Can only attach to instances on the same host
- Must destroy attached instance before deleting volume
- Size must be specified at creation time
Storage Costs
Storage pricing varies by host and includes:- Container storage: Charged per GB while instance exists
- Volume storage: Charged per GB while volume exists
- Different rates: Stopped instances may have higher storage rates than running instances
Data Persistence Strategy
Temporary Work
Use container storage for:- Temporary files
- Build artifacts
- Cache data
- Working datasets
Important Data
Use volumes or cloud sync for:- Trained models
- Datasets
- Code repositories
- Configuration files
Backup Options
- Volumes: For same-machine persistence
- Cloud Sync: For off-machine backup (Google Drive, S3, etc.)
- Instance-to-instance copy: Transfer between instances
- SCP/SFTP: Download to local machine
Common Questions
Can I increase storage after instance creation?
No, container storage size is fixed at creation. You can:- Create a new instance with more storage and transfer your data
- Attach a volume for additional space
- Use cloud storage for overflow
What happens to my data when the instance stops?
- Container storage: Data persists, charges continue
- Volumes: Data persists, charges continue
- No data is lost when stopping instances
How do I avoid storage charges?
- Destroy instances you’re not using
- Delete unneeded volumes
- Transfer important data to local/cloud storage first
Next Steps
- Volumes - Create and manage persistent volumes
- Data Movement - Transfer files between instances
- Cloud Sync - Connect to cloud storage providers
- Cloud Backups - Automate backup strategies