From backup-and-restore to active-active

Backups protect your data. Disaster recovery (DR) is about the rest: if the building, the data centre or the whole cloud region you run in is gone, where does the system run tomorrow, and how quickly can it be there?

AWS’s whitepaper Disaster Recovery of Workloads on AWS sorts the answers into four approaches, “ranging from the low cost and low complexity of making backups to more complex strategies using multiple active Regions”. The names are used across the industry, whichever cloud or data centre you’re in.

1. 1Backup and restore: copy backups to another region; after a disaster, rebuild everything there and restore. Cheapest, slowest.
2. 2Pilot light: data is replicated continuously and core infrastructure is ready, but application servers stay switched off until a failover.
3. 3Warm standby: a scaled-down but working copy runs all the time. It can take traffic at once, then scales up.
4. 4Active/active: both sites serve users all the time, so there is nothing to fail over to. Most complex, most costly, recovery near zero.

What separates them

• Backup and restore is, in the whitepaper’s words, “a suitable approach for mitigating against data loss or corruption”, and for many systems it’s enough. Recovery means redeploying “the infrastructure, configuration, and application code” as well as the data. Without infrastructure as code, it warns, that “may be complex” and can blow your RTO.
• Pilot light vs warm standby is the distinction people mix up. The whitepaper is precise: “pilot light cannot process requests without additional action taken first, whereas warm standby can handle traffic (at reduced capacity levels) immediately.”
• Multi-site active/active “is the most complex and costly approach to disaster recovery, but it can reduce your recovery time to near zero for most disasters”.

None of them is “best”. The right one is the cheapest that meets a system’s RTO and RPO, which is why the targets from the previous lesson come first. Most organisations end up with different patterns for different systems.