Backups
The Backup feature in VoltageGPU provides automated, scheduled backups of your pod data to ensure your important files are protected even if your rental container disappears unexpectedly.
Why Use Backups?
The Problem
If your rental machine disappears without notice due to hardware failure, network issues, or other unexpected events, you could lose all your work immediately. This includes:
- Training checkpoints and model weights
- Processed datasets and intermediate results
- Configuration files and scripts
- Research outputs and analysis results
The Solution
VoltageGPU's backup feature mitigates this risk by:
Automated Scheduling
Regular backups run automatically based on your configuration.
Secure Storage
Backups are stored in AWS S3 with enterprise-grade security.
Flexible Retention
Configure how long backups are kept to balance cost and safety.
Easy Recovery
Simple restoration process when you need to recover data.
How to Configure Backups
Step 1: Access Backup Settings
- Navigate to your pod detail page in the VoltageGPU dashboard
- Click the "Backup" button in the top right corner
- This opens the backup configuration panel
Step 2: Configure Backup Settings
You'll need to specify three key parameters:
Backup Path
The directory path that will be zipped and uploaded to cloud storage.
# Valid paths
/root # Entire volume mount
/root/models # Subdirectory of /root
/root/checkpoints # Training checkpoints
/root/project/data # Nested subdirectory
# Invalid paths
/home/user/documents # Not under /root
/tmp/backup # Temporary directory
/var/log # System directoryBackup Frequency
How often the backup job runs:
- Every 6 hours - For active development work
- Every 12 hours - For regular training runs
- Daily - For stable projects with less frequent changes
- Weekly - For long-term storage and archival
Retention Period
How long backups are kept before automatic cleanup:
- 7 days - Short-term protection for active work
- 30 days - Medium-term retention for ongoing projects
- 90 days - Long-term retention for important research
- 180 days - Extended retention for critical data
What Happens Behind the Scenes
Backup Process
- Scheduled Execution: A cron job runs based on your configured frequency
- Data Compression: The system creates a ZIP archive of your specified backup path
- Cloud Upload: The compressed backup is securely uploaded to AWS S3 storage
- Verification: The system verifies the backup was successfully stored
- Cleanup: Old backups are automatically removed based on your retention settings
Storage & Monitoring
- Backups are stored in dedicated AWS S3 buckets with enterprise security
- Each backup is timestamped and uniquely identified
- Access is restricted to your account only
- View backup history and status in the "Backups" tab on your pod detail page
Billing
Backup storage costs are passed through from the infrastructure provider with no VoltageGPU markup. You pay the exact storage rate charged by the underlying provider -- nothing more.
- No Markup: VoltageGPU does not add any margin on backup storage costs
- Storage Costs: Billed per GB stored, based on the actual size of your compressed backups
- Transfer Costs: Charges for uploading backups to cloud storage (also at direct provider rates)
- Real-time Billing: Costs are fetched from the infrastructure provider hourly
- Direct Pass-through: You pay the infrastructure provider's direct rate with zero VoltageGPU markup
Typical cost depends on the size of your data. Smaller, targeted backups (e.g., checkpoints only) keep costs minimal.
Cost Optimization Tips
- Only backup essential directories, not entire volumes
- Use appropriate retention periods for different types of data
- Consider backup frequency based on how often your data changes
- Regularly review and clean up old backups if retention periods are too long
Frequently Asked Questions
Can I backup my entire volume?
Yes, but consider the cost implications. It's more efficient to backup only critical directories rather than entire volumes.
What happens if my backup fails?
Failed backups are logged in your backup history. The system will retry on the next scheduled run. Check the backup logs for specific error details.
Can I manually trigger a backup?
Currently, backups run on the configured schedule. Manual triggering may be added in future updates.
Why am I being charged when my pod is stopped?
Backup storage costs continue even when pods are stopped because your backups remain available in cloud storage for the full retention period.
What compression is used for backups?
Backups use standard ZIP compression, which provides a good balance of compression ratio and compatibility.
Can I restore backups to a different pod?
Yes, backup restoration is not tied to specific pods. You can restore backups to any pod with sufficient storage space.
Protect Your Training Data
Configure automated backups and never lose your work again.