Glossary›Storage Systems, Protocols, and Scale›NFS

NFS

Why does NFS matter for unstructured data?

Because a large share of research, engineering, media, and AI data sits on NFS shares. Compute farms, render nodes, and analysis servers read and write it all day over NFS. Any tool that indexes, moves, or protects that data reaches it the same way, so how NFS is set up affects how quickly the data can be scanned and moved.

What is an example of NFS in use?

A chip design team runs thousands of simulation jobs on a farm of Linux servers. Every server mounts the same project share over NFS, so each job reads the same design files and writes its results to the same folders. Engineers see one shared file tree, wherever their job runs.

What are the versions of NFS?

Three versions are common today:

• NFSv3: simple and very widely used, and it trusts the user IDs that each client sends.
• NFSv4: adds stronger security through Kerberos, richer access control lists, and a single network port that is easier to manage through firewalls.
• NFSv4.1 and later: add parallel NFS, which lets clients read from several storage nodes at once for higher speed.

How do NFS permissions work?

NFS identifies users and groups by number, called the UID and GID, rather than by name. With NFSv3, the storage system trusts the numbers each client sends, so the same person must have the same number on every server. A setting called root squash stops the administrator account on a client from acting as administrator on the share. Directory services such as LDAP or Active Directory keep the numbers consistent and map them to names.

Is NFS secure?

It depends on how it is set up. NFSv3 with default settings trusts the client machine, so anyone with administrator rights on a client can pretend to be another user. Common protections are limiting each share to known client machines, turning on root squash, and using NFSv4 with Kerberos to check each user’s identity and encrypt traffic.

How does NFS affect scanning and data movement?

Every file’s metadata is read across the network, so the delay on each request matters more than raw bandwidth when a share holds billions of small files. Scanners keep up by running many parallel workers close to the storage, and by reading only folders that changed since the last scan. For large files, read size settings and a fast network decide how quickly data can be copied.

How do data management tools work with NFS?

They connect to NFS the same way any server does, by mounting the share, and the setup decides what they can see and do:

• Access to the share: the storage system has to list the tool’s server as an allowed client of the share, usually read-only for indexing.
• Access to every folder: with root squash on, the tool cannot read folders closed to ordinary users unless its account is given read access.
• Owners by name: the tool looks up user and group ID numbers in LDAP or Active Directory to show who owns what.
• Write access only when needed: moving, archiving, or deleting files requires write access, while indexing and reporting do not.

← Back to Storage Systems, Protocols, and Scale

Last updated: November 13, 2026.