What the three copies actually mean.
Production data is the first copy, supplemented by at least two backup copies. They must not all be exposed to the same failure, account or administrator access. A second file on the same server is not an independent backup.
Separation matters more than the number.
Ransomware and compromised administrator accounts can deliberately delete backups. At least one copy should therefore be separated so that an attack on production cannot easily reach it. Separate credentials, another platform or immutable storage can provide this boundary.
Off-site alone does not solve everything.
An external copy protects against fire, theft or total site failure, but it still requires monitoring and restore tests. Bandwidth also matters: retrieving a large dataset from remote storage may take longer than the business can tolerate.
A simple control question for SMEs.
For every important system, identify where the three copies are held, which one is independently protected and when recovery was last tested. Any unclear answer indicates an incomplete strategy.
A managed backup combines copying, monitoring, separation and restore testing in one operating process.
Decide which copy will be restored first.
A local copy may be fast, while an external copy offers better protection against a site-wide failure. The recovery plan should state which copy is preferred for each scenario.
Include encryption and key management.
External backups should be protected against unauthorised access. At the same time, decryption keys and recovery codes must remain available if the main server fails. Store them through a separate, documented emergency route.
Retain enough history for slow-moving errors.
Deletion or corruption may be noticed weeks later. Retention should therefore include several historical recovery points, chosen according to the data type, change rate and business requirements.
Review 3-2-1 regularly.
New cloud services, sites and growing data volumes change recovery needs. Reassess copies, separation, retention and recovery paths after major infrastructure changes.