Redundancy: Backups: What is backed up and how often?
Options:Option 1: Never backup anything.
Option 2: Backup critical information when you feel concerned about losing it.
Option 3: Backup everything periodically.
Option 4: Create and maintain different backup processes and schedules for different information based on its criticality.
Option 5: Continuously and automatically backup everything all the time.
Basis:Never backup anything.
Some things should not be backed up because they unnecessarily consume space. This includes standard distributions of operating environments where there are many existing backups and a standard regeneration process and other similar collections that are readily recreated. Other things not to backup include data that is for temporary use only, like authentication codes used on specific transactions (e.g., credit card present digits), cookies for systems where there is no desire to retain long-term state information, temporary files such as those used in processing other data, data on kiosk machines in lobby areas, and other content that has little or no utility after its initial use. In the forensics arena, there are also cases where backup is not permitted, for example under court orders relating to making copies of certain content, and in cases where old copies may create confusion or legal issues associated with discovery limited backups or automatic overwrite of backups with newer data may be used.
Backup critical information when individuals feel concerned about losing it.
This strategy is used predominantly by enterprises that do not have a standard approach and by individuals who may wish to save their own work but where that work has little impact on the enterprise. Examples include computers operated by students and professors at universities, personal computers used by individuals at home, and research computers used for experiments but not part of operations.
Backup everything periodically.
This generally only applies to systems with medium or high consequences of failure and in cases where these systems do not have standard configurations or other methods for rapid recreation. This is common in systems designed for special purposes, research and development systems, and for enterprises in which there are relatively small numbers of individuals working on independent projects. The most common approach is a monthly compete backup with annual permanent storage, weekly incremental backups kept for a month, and daily incremental backups kept for a week. As disk-to-disk backup has replaced tape and other media backup, backups to file servers has become increasingly dominant.
Create and maintain different backup processes and schedules for different information based on its criticality.
In most enterprises, there are enough differences between systems that a mixed strategy is applied using many of these approaches depending on the specifics of the systems involved.
Continuously and automatically backup everything all the time.
In transaction systems or increasingly in resilient infrastructure systems and to a lesser extent in systems with built-in incremental backup systems, like Apple's "Time Machine", backups are done either immediately or with frequencies high enough that almost everything ever done is recorded.