cloudparts

Duplicati

A backup client for Windows, macOS and Linux that splits files into blocks, stores each block once, and uploads compressed, encrypted volumes to cloud storage or a file server, driven from a web UI or the command line.

What it is
Backup & restore
Type
Open source
Made by
Duplicati Inc.
License
MIT
Facts checked
24 October 2026
Website Source code

What it is

Duplicati runs on the machine being backed up and needs nothing on the other end but storage. It reads each file as a sequence of fixed-size blocks, typically 100 KiB to 1 MiB, and hashes each block with SHA-256. A block it has already stored is skipped; new blocks are packed into volumes (dblock files), which are compressed, encrypted with AES-256 or GnuPG, and uploaded. Each backup version is a file list (dlist) that says which blocks make up which files, and dindex files record which volume holds which block. Because only new blocks are uploaded, every backup after the first is incremental, yet every version can be restored on its own; there is no full-plus-incremental chain to keep.

The destination only has to support four operations, GET, PUT, LIST and DELETE, so Duplicati can write to Amazon S3 and S3-compatible stores, Backblaze B2, Azure Blob Storage, Google Cloud Storage, SFTP, WebDAV, SMB shares, consumer drives such as OneDrive and Dropbox, and others. A local database for each backup tracks files, blocks and volumes so backups run without downloading anything; it can be rebuilt from the remote files and the passphrase if it is lost. Retention rules, such as “one a day for a week, one a week for a month, one a month for a year”, delete old versions after each backup, and a compact step repacks volumes that hold unused blocks.

Most people use the built-in server with a browser UI and scheduler; duplicati-cli does the same jobs from scripts. Duplicati Inc., founded in 2024 with the project’s original author as CTO, sells a hosted console for managing many machines, with a free tier for two.

Why teams pick it

  • Encrypts on the client before upload, so the storage provider only ever sees encrypted volumes
  • Block-level deduplication across files and versions, with no periodic full backups to re-upload
  • Works with dozens of storage services out of the box, including plain SFTP, WebDAV and SMB shares
  • A web UI and scheduler make it usable on a desktop, a home server or a NAS without scripting

Watch out for

  • Fixed-size blocks: an insertion near the start of a file shifts every later block, and the block size can’t be changed after the first backup
  • Restoring even one small file means downloading the whole remote volume that holds its blocks, so large volume sizes slow small restores
  • The local database must match the remote files; when it doesn’t, backups stop until a repair, and recreating the database for a large backup can take a long time
  • A desktop and small-server tool: there’s no central server for a fleet unless you use the company’s hosted console

Alternatives

What all backup & restore have in common →Compare side by side →
PlakarOpen sourceA backup engine in Go that takes deduplicated, encrypted, immutable snapshots of files, databases and cloud sources into a store called a Kloset, which you can browse, diff, verify and sync without restoring.

How it fits with other technologies

Back up two folders to an S3-compatible bucket, keeping “smart” retention · bash

# The passphrase is read from the environment, not the command line;
# S3 keys go in the URL (aws-access-key-id=...) or in a secret provider
export PASSPHRASE='a long passphrase'
REMOTE_URL='s3://acme-backups/laptop?s3-server-name=s3.eu-central-1.amazonaws.com&use-ssl=true'

duplicati-cli backup "$REMOTE_URL" \
  /home/ana/Documents /home/ana/Projects \
  --exclude='*.iso' \
  --retention-policy='1W:1D,4W:1W,12M:1M'

# List the versions that now exist
duplicati-cli list-filesets "$REMOTE_URL"

See it in a real stack

No architecture uses Duplicati yet. Drop it into the playground and wire it up yourself.Open playground →

Worth exploring

Article· January 202615 open source backup solutions to protect your data in 2026Don Watkins, All Things OpenA short survey of 15 open-source backup tools in three groups: network and command-line tools (Amanda, BackupPC, Bacula, BorgBackup, Rclone, restic), desktop tools (Back In Time, Déjà Dup, Duplicati, Timeshift, UrBackup and others), and newer options (Kopia, Syncthing), each with one line and its licence. Good for finding names to look into; it doesn’t compare them in depth.
All resources →
Where it came from2008–2026

Kenneth Skovhede started Duplicati in June 2008 as a graphical front end for Duplicity, the Linux backup tool, along with a port of Duplicity to Windows. The port was dropped that September in favour of a clean re-implementation, and the first version came out in June 2009.

In 2012 work began on Duplicati 2, a complete rewrite around a new storage engine for incremental backups: files split into blocks, each block stored once. Version 2 then stayed in beta for years.

In 2024 Skovhede founded Duplicati Inc. to commercialize the project, taking the role of CTO, with funding from Open Core Ventures. The licence moved from the LGPL to MIT in March 2024, and the company launched a hosted console for managing backups in May. On February 28, 2025 the project announced “the first stable release of Duplicati 2.x” (the 2.1 line), saying it had been stuck in “a perpetual beta phase” until the company could hire staff. Stable releases have followed every few months since; 2.4 arrived in September 2026.

Why it’s built this way

Fixed-size blocks and deduplication instead of full and incremental chains
With every block stored once and every version a list of blocks, no version depends on an earlier full backup, so old versions can be deleted in any order.
Client-only, against plain storage
The design assumes the destination may be slow, unreliable and owned by someone else, so all logic runs on the client and storage needs only GET, PUT, LIST and DELETE.
Encrypt before upload
Since the storage provider can’t be trusted with confidentiality, volumes are encrypted on the client, and whoever holds the passphrase holds the data.
A local database of blocks and volumes
Knowing what is already stored without asking the remote side keeps backups fast, and checking the remote listing against it catches missing or damaged files early.

Timeline

  1. 2008

    Kenneth Skovhede starts Duplicati as a GUI for Duplicity in June

  2. 2009

    First release in June

  3. 2012

    Work starts on Duplicati 2, a complete rewrite

  4. 2024

    Duplicati Inc. founded; licence changes from LGPL to MIT in March

  5. 2025

    Duplicati 2.1, the first stable 2.x release, after years in beta

  6. 2026

    Duplicati 2.4 released in September

Sources11 · facts checked 24 October 2026

Improve this page on GitHub