Backup Methods


Two main factors govern backup operations:

  • Frequency
    • how often to backup
    • configuration reflects how much lost work can be tolerated
  • Retention
    • how long a backup is kept
    • short-term and long-term

Backup Chains

Requirements for backup frequency and retention must be managed against:

  • capacity of the backup media
  • time it takes to complete a backup job

These requirements are managed by using different types of jobs in a backup chain.

Main Types of Backups

TypeData SelectionBackup Job Time and Storage RequirementRecovery ComplexityArchive Attribute
FullAll selected data regardless of when it was previously backed upHighLow (single job)Cleared
IncrementalNew files and files modified since last backup jobLowHigh (multiple jobs)Cleared
DifferentialNew files and files modified since last full backup jobModerateModerate (two jobs)Not cleared

Info

Windows uses an archive attribute to determine the backup status. Linux doesn’t support a file archive attribute. Instead, a date stamp is used to determine whether the file has changed. Most software also has the capability to do copy backups. These are made outside the chain system (ad hoc) and do not affect the archive attribute.

Synthetic Backup

synthetic full backup is not generated directly from the original data but instead assembled from other backup jobs.

  • option for creating full backups with lower data transfer requirements

How it Works

  1. The chain starts with an initial full backup as normal and subsequently makes a series of incremental backups.
  2. When the next full backup is scheduled, the backup software makes one more incremental backup. It then synthesizes a new full backup from the previous full and incremental backups.