Video Demonstration

Prefer a hands-on visual walkthrough? Watch the complete live demo on YouTube: Cold Archive your NAS without breaking SMB shares.

The SMB Archiving Dilemma

Every shared network environment faces the same reckoning eventually. Your high-speed SSD or NVMe pool hits 90% utilization. You have video editors, 3D artists, automated rendering queues, and office users connected across Windows, macOS, and Linux workstations via SMB (Server Message Block).

Until now, storage administrators had only two bad choices:

Traditional Option A

The "Buy More NVMe" Trap

Keep buying flash storage indefinitely. 80% of the files on the volume haven't been accessed in six months, but they sit on premium, power-hungry, high-end flash just in case someone needs them.

Traditional Option B

Manual Offloading (Broken Paths)

Move older folders to an unmounted "archive" share, tape, or cloud bucket. The result? Broken project links in Premier/DaVinci, broken relative paths in scripts, and angry users opening IT tickets.

With HuskHoard, there is a third option: leave the paths, names, and attributes untouched in the share, migrate the cold payloads to cheap media (tape, object storage, or cold JBOD disks), and fetch them back automatically on demand.

The Client Experience: Zero Broken Paths

From the user or client application’s perspective, nothing moves. When an editor mounts \\nas\projects\2024\promo.prproj or browses the directory in Windows File Explorer or macOS Finder:

When the user or application opens the file, it seamlessly stages back to the local SSD tier. The client application experiences only a brief standard I/O pause while the bytes are retrieved from the archive tier. Once hydrated, the file stays on high-speed flash where subsequent reads and writes happen at local wire speed.

Under the Hood: Samba Meets fanotify

How does this work over a protocol like SMB without modifying client workstations or installing proprietary client-side agent software?

Samba (the service presenting your SMB share on Linux) operates in user space and interacts with the host filesystem using standard POSIX system calls like open(), read(), and stat(). HuskHoard leverages the Linux kernel's native fanotify permission subsystem on the host storage server.

5
Client Machine
Windows Explorer / Mac Finder / Premiere Pro requests \\nas\share\render.mov over SMB.
client os
4
Samba Daemon (smbd)
Receives the SMB packet and issues a local POSIX open("/mnt/share/render.mov") call.
user space
3
HuskHoard (fanotify)
Intercepts Samba's open() before it completes. Checks the catalog, pulls blocks from cold tier to SSD, then emits FAN_ALLOW.
← intercept & hydrate
2
Local SSD Cache Pool
The file is now physically staged back on fast NVMe/SSD storage.
local storage
1
Archive Volume
LTO Tape Cartridge, Cold S3 bucket, or spun-down SMR/CMR disk.
cold media

Because the interception happens entirely inside the Linux host kernel below Samba, the remote SMB client doesn't need to know anything about HuskHoard. To the Windows client, Samba simply took a moment to return the file handle. No special drivers. No client plugins. No protocol modifications.

The Complete Lifecycle: Aged Out, Recalled, Recycled

Data tiering is not a one-way trip. HuskHoard manages an active, cyclic lifecycle on your share:

01
Normal Ingest & High-Speed Work
Files are written directly into the SMB share. They sit on fast local SSD pools for low-latency editing and collaboration.
02
Files Age Out & Are Archived
After sitting untouched for a configured duration (e.g., 45 days), HuskHoard migrates the data payload to your archive tier (cloud, tape, or secondary disk) and leaves an extended-attribute stub file in place. Flash capacity is immediately freed.
03
Transparent Retrieval (Hydration)
A year later, an application or user opens the file. HuskHoard catches the request, retrieves the payload from the archive volume, and places it back onto the SSD pool.
04
Warm State & Re-aging
The restored file remains on the SSD tier for as long as it is actively used. Once it becomes dormant again, the timer resets and it quietly ages back out.

Target Media: Cloud, Tape, or Cold Disk

HuskHoard doesn't lock you into a single backend storage format. The archived payload can be offloaded to whatever tier matches your budget and operational model:

01 — Object Storage
S3 & Cloud Glacier
Ship payloads to cheap offsite cloud buckets. Perfect if you don't want to manage physical tape hardware or separate on-prem cold storage racks.
02 — Linear Tape
LTO Tape & Libraries
The gold standard for air-gapped, ultra-low-cost archival ($0.005/GB). HuskHoard interacts with standalone drives or robotic autoloader libraries natively.
03 — Cold Hard Drives
SMR / JBOD Disks
Secondary spinning rust arrays or spun-down external disks that only receive sequential writes and spin up when retrieval is commanded.
04 — Redundancy
Dual-Target Tiering
Configure policies to duplicate stubs across both on-prem LTO tape for instant air-gap safety and S3 for disaster recovery.

Setting Up HuskHoard with an SMB Share

Configuring HuskHoard alongside Samba takes only a few straightforward steps on any standard Linux distribution.

1. Configure the Archive Policy

Define what counts as "cold" for your SMB root. In this example, any file over 500MB in the /hot_tier share that hasn't been accessed in 30 days is queued for archival:

2. Standard Samba Configuration

Your smb.conf requires virtually no adjustments. Because HuskHoard handles extended attributes and transparent interception seamlessly, standard POSIX share directives work out of the box.

3. Run the Daemon

Launch the HuskHoard background service to begin monitoring the share.

A Note on Client Timeouts

If you are archiving to tape autoloaders or deep cloud tiers with retrieval latency of more than 30–60 seconds, configure standard SMB lease and session timeout settings on Samba to prevent aggressive client timeout popups while the media loads.

Stop paying a premium to keep cold data on high-speed flash, and stop breaking production workflows with disconnected archives. HuskHoard gives you the best of both worlds: uncompromised SSD performance for active jobs, and transparent archival for everything else.