How does Vertica backup work?

Vertica provides a utility which fully automates the backup process. The standard backup script is designed to take hot, incremental backups. Vertica’s backup is entirely file-based. Because we never change data files in place (only add new files, and remove old ones), our backup process concentrates on getting a version of the Vertica configuration, catalog, and data files that comprise a consistent snapshot of the system. The Vertica backup script does exactly that – it stores catalog information to a file, determines which corresponding data files need to be backed up, and takes reference counts to make sure that none of the files that need to back up will be deleted before the backup completes. Once the files have been identified, the script uses rsync to transfer the files to the backup location. The first time you run the backup utility it creates a complete backup. Subsequent runs are incremental, and the utility copies only files that are new since the previous backup. As a hot backup implementation, the procedure is designed to have minimal impact on a running database. Taking the snapshot itself is extremely lightweight; the more expensive part is simply the transfer of the backed up data to the backup location.