Storage overview

Compare Runpod network volumes and global volumes, and learn how to attach persistent storage to Pods, Serverless endpoints, and Runpod SDK apps.

Runpod offers persistent storage that exists independently of your compute resources. Data on a persistent volume stays available after a Pod stops or a Serverless worker shuts down, and you can attach the same volume to multiple resources.

Pods and Serverless workers also include temporary disks that are tied to the resource’s lifecycle. To learn about those, see Pod storage options and Serverless storage.

Storage options

• Network volumes: Persistent storage located in a single data center. Network volumes are available in a standard tier and a high-performance tier that provides up to 3x the throughput and 4x the IOPS.
• Global volumes (beta): Persistent storage that’s available across data centers, so you can access the same files from compute in any supported region.
• S3-compatible API: Manage files on your network volumes using standard S3 tools, without starting a Pod or worker.

Compare network and global volumes

	Network volume	Global volume
Location	One data center. Compute that mounts it runs in that data center.	Not tied to a data center. Compute in any supported data center can mount it.
Best for	Workloads that write often, such as training and checkpointing.	Workloads that write rarely and read often, such as loading model weights for inference.
Consistency	Changes are visible right away.	Eventually consistent. Changes made after a worker starts may not be visible immediately.
File system	Full file system behavior.	Object-backed. No file locking, atomic rename, hard links, or permission bits.
Compute	GPU and CPU.	GPU only.
Capacity	You set the size and can increase it later.	Grows automatically with your data.
Pricing	Billed per GB per month. See Network volumes.	Billed for storage plus read and write requests. See Global volumes.

Choose a storage option

• Use a network volume when your workload runs in one data center and needs fast, consistent reads and writes, such as training checkpoints or a shared model cache for an endpoint.
• Use a global volume when you deploy to multiple regions and want the same models or datasets available everywhere without copying them.
• Use the S3-compatible API when you want to upload or download files on a network volume from your local machine or a CI pipeline.

Where volumes mount

Compute	Mount path	Limits
Pods	Network volumes mount at /workspace. Global volumes mount at /workspace, or at /workspace-global when a network volume is also attached.	One network volume and one global volume per Pod.
Serverless endpoints	Both volume types mount at /runpod-volume.	An endpoint can attach network volumes or one global volume, not both.
Runpod SDK apps	Set with the mounts parameter. Queue and API resources mount one volume at /runpod-volume.	Tasks can mount one network volume and one global volume at different paths.

For SDK examples, see Storage in the Runpod SDK.

Next steps

• Create a network volume.
• Get started with global volumes.
• Transfer files with the S3-compatible API.