Making backup is an important aspect of system administration. The techniques of backing up data are explained in any good document about system administration, and they won't be explained here again.
However, bakonf comes into play into a particular part of the backups: the backing up of the system's configuration.
The basic idea is that on a standard instalation of a Unix-like system you have a lot of data which can be very easily restored from the original media, thus there is no point in archiving it. For example, after a fresh install of a RedHat Linux 8.0, you have ˜4.5GB of space used. However, only a very small part of this amount is holding important information, the other part being binaries, libraries and other kind of data which will never modify in normal usage. Only the configuration files are changing (of course, also the user data is changing, but we are talking about an empty system).
If we classify the files existing on a Unix system, we have:
These are the target of bakonf; they are usually small text files, partly coming from the system installation, maybe edited by the administrator, partly created by him. Size is (on the workstation I write this) ˜15MB.
These are mostly read-only; in a package based distribution, they came from the packages and are replaced when the package is upgraded. In classical systems, they come from the install archives. Size is (in our hypothetical rh8.0 full install) ˜4.5GB.
These are emails, web pages, documents, etc. - this is important data, and needs to be backed up regularly. They also don't come from the installation media, and are not touched by the system. Size is undetermined, but is guaranteed to be exactly the amount of free space on the system :).
These are the files created and managed by the system, usually from the configuration files and other external variables. Examples: /var/lib/logrotate.status, /var/lib/slocate/slocate.db. These are not all critical files, some are needed to be included in a backup only for analysis purposes, others should not be included in backups (e.g. if you reinstall your system or restore from backup, some files will have for sure other contents, generated from the new installation).
From all these, only the configuration data and the user data are absolutely required to recreate the system. The binaries can come from the installation source. The system managed data will be recreated by the system. And since the difference in size between the configuration and user data is so great in a typical system, I believe it deserves another backup method (besides the inclusion in standard backup procedures, which are, I repeat, CRITICAL!).