zpaq is a journaling archive for incremental backups with deduplication. Journaling means that the archive is append-only and you can revert to an earlier state to recover old versions of files.

Compression is based on the open ZPAQ standard, a self-describing format that preserves compatibility as improved algorithms are developed. The default compression level is faster than zip with better compression. Typical speed for 100 GB is 1-2 hours for a full backup and 1-2 minutes for incremental backups of files whose date has changed.

zpaq includes the public domain libzpaq API in C++ providing compression and decompression services to developers. zpaq includes development tools for writing and debugging custom compression algorithms in the ZPAQL language.

Download from http://mattmahoney.net/dc/zpaq.html