The Samba/NFS replacement that will end the VPN.
Drive is the file-server replacement on the OxiMail roadmap. The plan: mount natively as a disk on Windows (M:\), macOS (/Volumes/OxiMail) and Linux (/mnt/oximail), over one protocol — HTTPS on port 443 — behind one authentication, your OxiMail account. The same experience over the LAN or from the other side of the world. It is not shipped: the pilot opens Q3 2026, and this page describes where we are taking it.
Three arguments. One product.
The VPN will disappear
For file access, at least. Remote workers will connect to https:// and see their files — no client-side tunnel, no server-side VPN concentrator, no “I can’t connect” support tickets. A VPN may still exist for legacy LAN applications, but the file-access load is designed to drop to zero.
Ransomware lateral propagation — handled by design
The architecture leaves no LAN-accessible file share for a compromised endpoint to enumerate and encrypt. Each file access is authenticated per user, per operation, over TLS 1.3. A workstation infection is meant to stay on that workstation — an argument a cyber-insurance underwriter and a compliance officer both grasp immediately.
One identity instead of four
No more matrix of AD account + mail account + VPN account + NAS credential. One OxiMail account, one authentication surface, one revocation path when someone leaves. LDAP sync is planned for enterprises that keep an AD of record — but AD stops being the authentication boundary.
SMB (1983) vs Drive (the design).
Six dimensions where the gap is structural, not a tuning question. SMB was designed for an office LAN before the modern internet. Drive is being designed HTTPS-first, for modern threat models.
| Dimension | SMB/CIFS | OxiMail Drive |
|---|---|---|
| Network port | 445 (blocked on residential ISPs, hotels, cellular networks, many corporate proxies) | 443 HTTPS — universal |
| Remote work | VPN required | Direct HTTPS — no tunnel |
| In-transit encryption | Optional in SMB 3+; often misconfigured; silent fallbacks exist | TLS 1.3 required, no fallback |
| Offline mode | Windows Offline Files — unreliable, rarely enabled in practice | Architectural — local cache, queued changes, conflict resolution |
| Ransomware lateral propagation | An established vector in modern attacks | No LAN-accessible share — per-user authenticated access only |
| Identity management | AD + mail + VPN + NAS — separate systems to keep in sync | One JMAP identity; LDAP sync planned |
Native mount, smart cache, advisory locks.
Drive is designed to mount using each OS’s native cloud-file framework — the Cloud Files API on Windows (the same API behind OneDrive, with native status icons), FileProvider on macOS (appearing under “Locations” in Finder), and FUSE on Linux. The same Rust agent binary is planned to drive all three.
A local cache (configurable: none, session-keyed, or ephemeral) holds recently-used files. Byte-range reads mean a 2 GB file streams in chunks rather than downloading wholesale. Predictive prefetch uses four strategies — proximity (open Q1.xlsx and Q2.xlsx warms), habit (every Monday 08:00), session (top-20 most-recently-used at login) and group (a colleague pushes a file and it warms in your cache). Advisory locks carry an explicit TTL and renewal, releasing automatically when a client crashes — fixing the classic “stale SMB lock after reboot”. ACLs follow the OxiMail sharing model: principals, per-folder entries, inheritance, LDAP-synced groups.
Linux and Windows first. macOS in 2027.
The v1 pilot is planned to ship on Linux (FUSE) and Windows (Cloud Files API) in Q3 2026 — both platforms are well understood and have mature bindings.
macOS FileProvider is a separate integration that needs an Apple-signed App Container, an NSFileProviderReplicatedExtension and Objective-C bridge work — we estimate six to twelve months of focused engineering. Rather than ship a half-working macOS binary alongside v1, we target a macOS GA in 2027. Operators with mixed fleets would run Drive on their Linux and Windows seats from v1, and use the Workspace web client for macOS users during the gap.
Join the pilot program.
The pilot gives you direct contact with the engineering team, pricing held for the duration of the pilot, and priority on feedback integration. We ask for a letter of intent, structured feedback on agreed dimensions, and — optionally — the ability to reference your deployment in a case study at GA.
Pilot seats open for Linux and Windows environments with 25 or more users. macOS-majority fleets are asked to wait for 2027 or to join in hybrid form.