PROVIDER Β· The classic deduplicating encrypted backup
BorgBackup
BorgBackup forked from Attic around 2010 under BSD, centring on deduplication, encryption, and snapshots with native SSH remote repositories and snapshot mounting, a classic choice for Linux server backup.
4pts
Pros
Cons
Product Information
Detailed Notes
Overview
BorgBackup, or Borg, forked from Attic around 2010 with primary authors in Germany, is BSD licensed and has long been a mainstream choice for Linux server backup. Its core is deduplication plus encryption plus snapshots, storing only changed chunks after the first full pass.
The typical pattern pushes backups to another server over SSH, which suits Backup & Migration work well since backups land on your own backup host rather than object storage, keeping cost low and restore fast. Snapshots can also be mounted to browse files, making single-file recovery simpler than unpacking an archive.
The limits are object-storage second class citizenship, with native backends being local directories and SSH and cloud storage requiring tools like rclone, no GUI so automation lives in shell scripts, and repository migration across major versions that needs maintenance windows in long-lived projects.
Core Strengths
- Maintained since 2010 with block dedup keeping multi-version archives far smaller than stacked fulls
- End-to-end encryption keeps repositories ciphertext on third-party or owned hosts
- SSH remote repositories push backups to your own backup server for fast restores
- Snapshots mount for browsing and single-file recovery without unpacking archives
Product Ecosystem
Repositories and snapshots
init, create, prune, and compact form the main flow. Prune needs compact to actually reclaim space, and running prune alone makes usage look unchanged.
Remote repositories and checks
Repositories over SSH live on another machine with check for consistency. Give the backup host a dedicated system account with restricted rights so a site compromise cannot take the backups with it.
Drawbacks
- No native object storage, so cloud backup detours through tools like rclone
- No GUI, leaving automation and monitoring to custom scripts
- Major version upgrades need repository migration and maintenance windows
- The dedup index lives in memory, demanding more RAM on large repositories
Use Cases
- Teams with their own backup host (β β β β β ) - SSH repositories are cheap and fast to restore
- Scheduled Linux server backups (β β β β β ) - Dedup, encryption, and snapshots form a solid combination
- Hot single-file recovery (β β β β ) - Mounted snapshots beat unpacking archives
- Object-storage-first environments (β β β ) - Requires rclone detours and extra configuration
Pricing
| Plan | Price | Scope |
|---|---|---|
| Licence | $0 | Open source (BSD), no licence fee |
| Remote repository | $0 | Your own backup host at no licence cost |
| Operations effort | Team dependent | Scripts, monitoring, and upgrades are yours |
The software is free and cost sits in the backup host and operations. The hidden cost is retention plus prune frequency, where skipping compact keeps disks looking full while running it too often costs performance, so test against your data volume.
FAQ
-
How is it different from restic? Borg targets SSH and local repositories with the dedup index in client memory, while restic natively supports object storage and travels better across platforms. Borg fits an existing backup host and restic fits cloud-first setups.
-
Why does space not free after deleting snapshots? compact is required to reclaim unreferenced chunks. prune alone only removes snapshot references, the most common misunderstanding.
-
Can it back up databases? Not live data files directly. Take a logical dump and back that up, or use the database's own physical backup, since copying live files tends to produce an inconsistent snapshot.