A recovery time objective (RTO) is the maximum time a system or service can be down after a disruption before the outage causes unacceptable harm to the business. It is a target agreed in advance, and it decides how fast your disaster recovery has to work.
RTO is measured from the moment the service fails to the moment users can actually use it again, not to the moment a server restarts.
RTO vs RPO
| RTO | RPO | |
|---|---|---|
| Question it answers | How long can we be down? | How much data can we lose? |
| Measured | Forwards from the failure | Backwards from the failure |
| Example | "Back within 4 hours" | "Lose no more than 15 minutes of data" |
| What it drives | How much infrastructure stands ready | How often data is copied |
How RTO shapes the DR design
The shorter the RTO, the more has to be running before anything goes wrong, and the more it costs. A rough guide to the four AWS strategies:
| Target RTO (typical) | Strategy that usually fits |
|---|---|
| Many hours to a day | Backup and restore |
| Around an hour | Pilot light |
| Minutes | Warm standby |
| Near zero | Multi-site active/active |
These are typical ranges, not AWS figures; the AWS DR whitepaper notes recovery time will always be "greater than zero" and depends on automation.
How to set one
- Start from the business, not the technology: what does an hour of downtime cost, and who is harmed?
- Check outside limits: client contracts, and for FCA-regulated firms the impact tolerance for each important business service.
- Set it per system: a payments service and a monthly reporting tool should not share one RTO.
- Include dependencies: DNS, identity, third-party APIs. The service is only as fast as its slowest dependency.
- Test it: an RTO is a guess until a failover test has timed it.
For the common gaps between the RTO leadership expects and what IT can deliver, see RTO vs RPO Explained: The Disaster Recovery Gap. Setting and testing RTOs per system is part of a Yemberzal Disaster Recovery Assessment.