Volumes & Storage

Float16 provides persistent storage for your GPU instances. Volumes are backed by NetApp Trident for reliable, high-performance storage.

Overview

Volume storage features:

• Persistent: Data preserved across instance restarts
• NetApp Trident: Enterprise-grade storage backend
• Flexible Sizing: 50 GB to 10,000 GB
• Encryption: AES-256 encryption at rest

Volume Management

Manage your volumes at GPU Instance > Volume.

Dashboard

The volume dashboard displays:

• Total Volumes: Number of volumes and attached count
• Total Storage: Total provisioned storage and usage percentage
• Health Status: Number of healthy volumes

Creating a Volume

1. Navigate to GPU Instance > Volume
2. Click Create Volume
3. Configure your volume:
   • Volume Name: Lowercase letters, numbers, and dashes only
   • Volume Size: 50 GB to 10,000 GB
   • Access Mode: RWO (Read/Write by single node)
   • Volume Mode: Filesystem or Block
   • Encryption: Enable/disable AES-256 encryption
4. Click Create Volume

Volume Configuration Options

Creating Volumes with Instances

Volumes are automatically created when you create a GPU instance:

1. Navigate to GPU Instance > Create Instance
2. Configure Volume Size (50-10,000 GB)
3. Click Create Instance

The volume is automatically attached to your instance and available for use.

Volume Lifecycle

Cost Savings with Stop/Resume

When you stop an instance:

• Compute billing stops - No instance cost
• Storage billing continues - $1.00/GB/month
• Data preserved - Resume anytime with your data intact

This allows you to save costs while preserving your work.

Pricing

View current pricing at GPU Instance > Pricing.

Billing Notes

• Storage is billed monthly based on provisioned capacity
• Billing starts when the volume is created
• Billing stops when the volume is deleted
• Data transfer within the same region is free

Best Practices

Right-size Your Volumes

• Start with your estimated needs plus 20% buffer
• One-Click Deployment models require minimum 100 GB for model weights
• Monitor usage in the Volume dashboard

Data Management

• Keep important data backed up externally
• Clean up unused files to optimize storage costs
• Use appropriate volume mode for your workload:
  • Filesystem: General-purpose storage, ML training
  • Block: Database workloads requiring raw device access

Cost Optimization

Next Steps

• GPU Platform Overview - Create GPU instances
• One-Click Deployment - Deploy vLLM models