Network File Storage Features
Last verified 25 Jun 2026
Network File Storage is a fully managed, POSIX-compliant file storage solution built for demanding workloads like AI/ML pipelines, containerized applications, and DigitalOcean Kubernetes (DOKS) clusters. It provides scalable, high-throughput shared storage that simplifies storage management for distributed applications.
Network File Storage (NFS) is a fully managed, POSIX-compliant file storage solution for Droplets and DigitalOcean Kubernetes (DOKS). It supports NFSv4.1, POSIX permissions, and VPC-based access, allowing a single share to be mounted across multiple Droplets and Kubernetes nodes.
NFS is available in two performance tiers:
- Standard Tier provides predictable, cost-effective shared storage for general-purpose workloads such as web applications, CI/CD pipelines, CMS platforms, App Platform workloads, and small-file compute use cases.
- High Performance Tier delivers sustained higher throughput and multi-node scaling for data-intensive workloads such as AI/ML training, GPU workloads, high-throughput analytics, and parallel multi-node access.
Both tiers provide up to 32 TiB of storage.
All NFS traffic uses VPC networks to ensure secure and private data transfer. NFS also supports point-in-time snapshots for file-level recovery.
You can upgrade a share from Standard to High Performance at any time without interruption, and you can increase a share’s storage size at any time. See the NFS limits page for details.
Access Points and Multi-VPC Access
Access points give clients controlled access to a share from one or more VPC networks in the same region. They let you connect to a share from Droplets and DOKS clusters in different VPC networks without VPC peering or any customer-managed networking. Access points are included with NFS at no additional cost.
Every share is created with a default access point that provides full read and write access to the share’s root directory. You can attach a share’s default access point to up to 10 VPC networks in the same region. Each attached network receives its own mount IP address, and the share stores a single copy of your data that all attached networks reach over separate network paths. Data is not replicated across networks.
You can also create subdirectory access points that restrict a client to a specific path within the share, such as /projects/team-a. A client that mounts through a subdirectory access point can read and write within that directory and its children, but cannot navigate above it. Subdirectory access points are isolated from one another: a client mounted through one access point cannot see or access directories governed by a different access point. From the same VPC network, the default access point has read-only visibility into a subdirectory access point’s contents. From other networks, it does not.
A subdirectory access point is scoped to a single VPC network, but a share can have multiple subdirectory access points across different networks, which lets different subdirectories be writable from different VPC networks. When you create a subdirectory access point for a path, write access for that path transfers to the access point. If you delete the access point, write access reverts to the default access point.
All access points provide read and write access; there is no read-only option. You cannot edit an access point after you create it. To change an access point, delete it and create a new one, then update the mount IP address on any clients that use it. For instructions, see How to Create and Delete Network File Storage Access Points.
Performance
NFS share performance varies by tier:
| Feature | Standard Tier | High Performance Tier |
|---|---|---|
| Throughput | Up to 1 Gbps aggregated | 8 Gbps up to 1 TiB of provisioned storage, then +1 Gbps per additional TiB |
| Read IOPS | 10,000 | Up to 100,000 |
| Write IOPS | 7,000 | Up to 70,000 |
Throughput scales with share size and represents aggregate throughput across all mounted clients.
Droplet Network Throughput Limits
Individual Droplet network interfaces limit the throughput each client can achieve over the VPC network. To increase aggregate throughput beyond a single Droplet’s limit, mount the NFS share on multiple Droplets or DOKS worker nodes.
| Droplet Type | Per-Droplet VPC Network Throughput Limit |
|---|---|
| GPU Droplets | Up to 25 Gbps |
| Premium CPU Droplets | Up to 10 Gbps |
| Standard CPU Droplets | Up to 2 Gbps |
Common Use Cases
Standard Tier
Standard Tier is designed for general-purpose shared storage across cloud applications.
| Use Case | Description |
|---|---|
| Web applications and media assets | Store shared web assets, user uploads, and exports accessed from multiple application instances. |
| CMS platforms | Shared filesystem access across compute nodes for themes, plugins, and uploaded media. |
| Analytics pipelines and shared staging | Shared staging area for CPU-based batch jobs, intermediate datasets, logs, and exports. |
| Shared storage for Droplets and DOKS | Shared configuration, assets, and small-file workloads with predictable baseline throughput. |
High Performance Tier
High Performance Tier is designed for data-intensive and throughput-demanding workloads.
| Use Case | Description |
|---|---|
| AI/ML workloads | Concurrent access to shared datasets, model artifacts, and checkpoints across multiple GPU Droplets. |
| Kubernetes applications (DOKS) | Stateful workloads with ReadWriteMany semantics across pods and nodes, including ML pipelines and shared caches. |
| Media and content collaboration | Multiple contributors working on shared project files without moving data between systems. |
Limits
Your maximum storage capacity and the maximum number of shares you can create are based on per-account limits. To increase your limits or request larger share sizes and higher performance, contact support. For a full list of limits, see the NFS limits page.