Sauvegardes dédupliquées avec Borg-Backup : retour d'expérience
Maurice Libes, Didier Mallarino · HAL (Le Centre pour la Communication Scientifique Directe) · 2017
The volume of scientific data stored in research laboratories is steadily increasing. Volumes from several tens to many hundreds of terabytes of data have become frequent.These very large volumes result in several problems inherent to data backup: - increased network demand, increasingly longer backup times;- backup policy scaled down due to lack of storage space.Traditional solutions are starting to suffer from excessively long backup times and may be subject to interruptions or errors that require you to restart the backup if the software does not support resume operations.In this situation, it is interesting to look for software solutions that could reduce the volume of data saved and therefore decrease backup times.We will present the BorgBackup deduplicating backup solution, software that is free in Python, which provides a satisfactory level of maturity and a very interesting set of features: deduplication, compression, encryption, resume after an interrupted or incomplete backup.Deduplication is a technology that operates at the level of file fragments (block sequences). Each file is split into fragments and only the fragments that have not been seen previously are saved to the backup repository. Across several thousand files, this pooling significantly reduces the number of bytes saved.This BorgBackup solution is showing all the signs of being a great candidate for production operations: stability, ease of use, modern features.We will detail:- the installation processes on clients and servers;- how to use it and its current features;- some performance information relating to compression and backup duration.