Open-source alternative to Nutanix Files

CephFS is the POSIX file layer on Ceph RADOS, and Ceph itself does not speak SMB or NFS. Clients that mount a CephFS file system use the native Ceph kernel or FUSE client. Windows clients and most NFS clients cannot use that path directly, so the design needs gateway services in front of CephFS. Samba exports CephFS to Windows and macOS clients over SMB, and NFS-Ganesha exports it over NFS. Both run as separate services on hosts you manage, and both are your responsibility to version, patch, clusterize and test. Before you commit to a design, confirm the exact Samba and NFS-Ganesha versions, the Ceph release they were tested against, and which SMB and NFS protocol features your clients need, such as SMB multichannel, file locking behavior or NFSv4 delegations. Those details depend on your own environment and should be verified with the upstream projects.

The Ceph Object Gateway (RGW) is a different service. It provides S3-compatible and Swift-compatible object APIs on top of the same RADOS cluster, and it does not serve file shares. Keep it in the design only if you also need object storage, which is covered on the sibling object storage page. Do not plan on RGW to replace SMB or NFS access to user home directories or departmental folders. Its feature set, including multisite, authentication options and the AWS-compatible IAM and STS APIs, should be checked against the RGW documentation for the release you deploy.

The practical decision is which gateway layer you will operate and who will support it. Samba handles Active Directory integration, share permissions and Windows ACL behavior, so ask your storage team to test a representative set of users, groups and nested permissions before cutover. NFS-Ganesha is the usual choice for Linux and UNIX clients. Plan for gateway redundancy, since a single gateway host becomes a single point of failure for every share it serves, and test failover with real client workloads. Ask any implementation partner for the written test plan and the list of gateway versions they will certify in your environment.

You do not have to leave AHV to move file shares. A common pattern is to keep AHV, or another hypervisor, for compute while CephFS serves SMB and NFS from a separate Ceph cluster deployed with cephadm. The licensing question comes first. Nutanix Unified Storage is licensed per usable TiB, and the mode you run determines what you still pay for. In mixed mode, NUS licenses sit on top of NCI or NCP licenses. In storage-only mode, only NUS licenses apply. Dropping Nutanix Files while keeping NCI changes which of these applies to your estate, and the effect on your NUS entitlement and your NCI renewal is not something to infer from the product pages. Ask Nutanix in writing which mode your cluster is licensed under today, whether removing Files reduces the TiB basis, and whether any licensed capacity can be released mid-term.

Identity and DNS are where overlap periods usually break. Users should see the same share names throughout the migration, so plan the DNS records before any data moves: lower the TTL on the existing Files names well ahead of the cutover window, and switch the names to the CephFS export only when the pilot share has passed its checks. Keep Active Directory as the single source of identity, and confirm with your SMB gateway vendor how it joins the domain and maps permissions, because Ceph's own documentation covers the Dashboard's SAML 2.0 and OAuth2 sign-in for administrators, not file-share authorization. For traffic between clients and the Ceph cluster during the overlap, enable the msgr2 secure mode on the network that carries it, since it encrypts post-authentication traffic on port 3300.

Treat the overlap as a rollback window, not a single switch. Pick a Ceph release with a support horizon beyond your planned retirement of Nutanix Files: Tentacle 20.2.x is the active line, and its estimated end of life is June 1, 2027, according to the Ceph release schedule, so confirm the current dates on that page before you commit. Move one share at a time, verify ownership and ACLs against a list kept from Files, and keep the Nutanix share readable until the business owner signs off. Retire the Files service only after the rollback window closes and your change log records each share's new path, owner and DNS entry.

A raw file share serves SMB or NFS paths that users mount or map, and it leaves sharing, versioning and login to the operating system and the directory. A collaboration layer such as Nextcloud sits on top of storage and adds sync clients for desktops and phones, sharing links with expiry and password options, file versioning and restore, and in-browser editing through Nextcloud Office or an integrated Collabora or OnlyOffice setup. Choose the collaboration layer when the main users are remote or mobile staff who need files outside the corporate network, or when external partners need links rather than accounts. Keep plain CephFS shares when the consumers are applications, build systems or legacy servers that expect a mounted path and POSIX semantics.

Login is the decision point most teams underestimate. Nextcloud's LDAP app lets Active Directory or LDAP users sign in with their directory credentials, and Nextcloud also documents OpenID Connect and SAML backends, so access can follow the same joiner and leaver process as other systems. Confirm with your identity team which backend matches your directory, how group membership maps to Nextcloud groups and quotas, and whether Talk calls and the Office editor should be enabled for every group or only pilot users. Plan the Talk and Office components as separate services with their own sizing, since they change the load profile of the server.

Plan the upgrade cadence before go-live. Nextcloud publishes a new major version every four months and supports each one for one year with monthly maintenance releases, so your operations team will apply a major upgrade roughly three times a year if it stays current. Nextcloud does not allow skipping major releases, so each upgrade must step through one major version at a time, with a full backup taken beforehand and third-party apps disabled until they are confirmed compatible. Budget that routine work into the platform team's calendar, or ask a support partner to take it on. Commercial Nextcloud Enterprise tiers add longer maintenance windows and response times, and that choice should be made against your own support requirements in a written quote.