Your drives will fail. Your data won't.
Split backups across multiple drives so no single drive holds your complete data. Lose a drive? Pull a new one, recover, keep going. It's math, not luck.
Your backup is split into N+K pieces using Reed-Solomon erasure coding - the same math that keeps CDs readable despite scratches. Lose up to K drives and still recover everything from any N.
AES-256-GCM encryption with Argon2id key derivation, on by default - opt out with --no-enc. Each drive holds one piece: encrypted, incomplete, and useless on its own.
ZIP/Deflate compression is on by default. BFS detects already-compressed files (photos, videos, archives) and skips them - no wasted CPU, smaller pieces.
USB drives, external HDDs, NAS, FTP/FTPS and SSH/SFTP servers, network folders - anything that can hold a file. Mix and match across providers. No cloud subscription required. No vendor lock-in.
Each backup creates a new numbered version. Keep the last 7, or 30, or all of them. Verify health, prune old ones, restore from any point.
One drive offline mid-backup? BFS doesn't give up. It keeps uploading to the rest and remembers exactly which ones failed - so later you fix and retry only those, not the whole thing.
Every piece of your backup carries a map to all the others. Find one surviving drive, run bfs recovery - your setup is back and BFS knows where the other pieces live. Restoring the files themselves then needs N of them reachable.
# 3 data + 2 parity = 5 providers (local, FTP, SSH)
bfs init mybackup --ci --data-shards 3 --parity-shards 2 \
--provider "local:disk1 --path /mnt/disk1" \
--provider "local:usb --path /mnt/usb" \
--provider "ftp:nas --config-file ./nas.json" \
--provider "ssh:vps --config-file ./vps.json" \
--provider "ssh:pi --config-file ./pi.json"# Each push creates a new versioned snapshot
bfs push # backup -> new version
bfs pull # restore latest version# ANY ONE provider rebuilds the configuration - BFS reads the map from its piece.
# Restoring the files then needs N providers reachable (3 of 5 in this scheme).
bfs recovery --provider ssh --name mybackup \
--bootstrap "--config-file ./vps.json"
bfs pull # restore files as usual| Command | Description |
|---|---|
| bfs init | Initialize a backup in the current directory |
| bfs push | Backup current directory -> new version |
| bfs pull | Restore files from backup |
| bfs status | Show backup health and version info |
| bfs verify [--deep] | Check piece availability and health across providers; flag missing/damaged header files. Exits 0 healthy, 4 degraded, 5 damaged; --deep verifies every piece's full data and catches damage the header check cannot see |
| bfs recovery | Rebuild backup config from provider pieces |
| bfs repair | Fix storage locations without re-uploading - repoint a moved device, rebuild a lost piece, or restore missing/damaged header files |
| bfs prune | Delete old backup versions |
| bfs versions | List all versions with health status |