Glossary
Disaster recovery
Disaster recovery is the set of copies, standby facilities and procedures used to restore IT systems after an event destroys a site or the data on it. Designs range from rebuilding from backups in an empty facility to running two live sites, and each system's recovery targets decide where it sits on that range.
Two kinds of disaster
Site loss means the building, power or network is gone while the data survives somewhere else. Data loss means the site is fine and the data is damaged, deleted or encrypted. A second data center fed by replication answers the first well and does nothing for the second, because the replica receives the damage along with everything else. Data loss is answered by point-in-time copies that cannot be altered after the fact, such as an immutable backup, kept apart from the replication path.
The recovery range, from cold site to active-active
NIST's contingency planning guide, SP 800-34 describes alternate sites, from the cold site to the hot site, by how much is in place before the event. The more that is ready, the faster recovery is and the more it costs between disasters.
| Design | Ready before the event | Typical recovery time | Typical data loss |
|---|---|---|---|
| Cold site | Space, power and network only | Weeks, set by hardware lead time | Back to the last backup |
| Warm site | Hardware and copies of data | Hours to days | Backup age or replication lag |
| Hot site | Running equipment and current data | Minutes to hours | Seconds to minutes |
| Active-active | Two sites serving users | No switch needed | None with synchronous replication |
These are planning ranges. Each is only as good as the last test that timed it.
Where the targets come from
A business impact analysis sets two numbers for each system: how long it can be down and how much recent data it can lose, the recovery point objective. Systems are then grouped into a few tiers, and each tier gets a design from the range above. The written result is the disaster recovery plan.
Why recovery takes longer than the plan says
A plan describes the environment on the day it was written, and new applications and changed dependencies make it wrong over time. Scale is the second gap: restoring a test set in minutes says little about restoring petabytes, which can take weeks over ordinary links. Hardware, network capacity at the recovery site and the staff available on the day all add time. Only disaster recovery testing at realistic size shows the real number.
Frequently asked questions
What is the difference between disaster recovery and business continuity?
Disaster recovery restores IT systems and data. Business continuity keeps the organization operating through a disruption and also covers people, premises, suppliers and manual workarounds. Disaster recovery is one part of it.
Is backup the same as disaster recovery?
No. Backup produces restorable copies of data. Disaster recovery is the plan for bringing whole services back, which uses those copies along with compute, network, a recovery site and procedures.