What is data backup?
Data backup is the process of copying files, databases, systems or virtual machines to separate storage so they can be restored after deletion, corruption, hardware failure or a cyberattack. Each copy is a restore point that records the data as it stood at a known time and is kept for a defined period.
What separates a backup from other copies?
A backup has three properties: it holds earlier versions, it sits on storage separate from production, and it is kept for a defined period. Each property rules out something that looks similar. A replica mirrors the current state, so a deletion or an encrypted file is copied across within seconds and no earlier version remains. An archive moves data out of production for long-term keeping and is not built to restore a whole environment. A snapshot on the same array as the data is lost if the array is lost.
Which recovery targets set a backup design?
Three targets set how often copies are taken, how fast restores run and how long copies are kept:
- The recovery point objective is how much recent data can be lost. It sets how often copies are taken.
- The recovery time objective is how long a system can be down. It sets how fast restores have to run.
- The reach back is how old a restore point may need to be. Corruption and intrusions are often noticed days or weeks after they start, so it sets how long copies are kept.
A fourth question follows from these: which accounts can delete or expire a restore point.
Why are backups a target in attacks?
Backups let an organization recover without paying a ransom, so ransomware operators look for the accounts that can delete or expire restore points and use them before encrypting production. A backup that one account can erase is only as safe as that account. An immutable copy on storage that enforces its own retention lock cannot be deleted before its date by any account, which closes that route.
How is a backup shown to work?
A completed job proves that data was written, not that it can be read back into a working system. The proof is a restore, performed at regular intervals and timed, of the kinds of system that would be restored in an incident. Restore testing turns an assumed recovery into a measured one and is what the last digit of the 3-2-1-1-0 rule counts.
Frequently asked questions
What is the difference between backup and disaster recovery?
Backup produces restorable copies of data. Disaster recovery is the wider plan for bringing systems back, which uses those copies along with compute, network and procedures.
How many copies of data are enough?
Common guidance is at least three copies on two kinds of media with one off-site, and for ransomware one immutable or offline copy. The count matters less than whether each copy can be restored.
Do cloud services back up their own data?
Many cloud and software services protect their platform but not a customer's deletions or compromised accounts. Customers often need a separate backup of that data.
