AWS describes four disaster recovery strategies, from cheapest and slowest to costliest and fastest: backup and restore, pilot light, warm standby and multi-site active/active. You choose by the recovery point objective (RPO), the recovery time objective (RTO) and the budget. The cheapest strategy that meets both objectives is the right one.
Last updated October 2026.
Both are business decisions. Lower values cost more, so a question that gives generous objectives and asks for the lowest cost is pointing at the cheaper end of the scale, even if a faster strategy would also work.
Only backups exist in the recovery Region: snapshots, AWS Backup cross-Region copies, S3 replication. After a disaster you redeploy the infrastructure and restore the data. AWS stresses infrastructure as code (AWS CloudFormation or the AWS CDK) here, because rebuilding by hand is slow and error-prone and can blow the RTO. RPO is set by how often you back up.
Data replicates continuously to live data stores in the recovery Region, and core infrastructure is in place, but application servers are switched off. AWS's best practice for "switched off" is not to deploy them at all until needed. On failover you switch them on and scale out. AWS Elastic Disaster Recovery uses this pattern for server-based workloads.
A scaled-down but fully functional copy of production runs in the recovery Region and can take traffic immediately at reduced capacity. Failover means scaling up, so service quotas in that Region must already be high enough to reach production size.
The workload runs and serves users in two or more Regions, routed by Amazon Route 53 or AWS Global Accelerator. AWS calls it the most complex and costly approach, able to bring recovery time near zero. Its variant, hot standby, runs full capacity in the second Region but sends it no traffic until failover.
AWS's own whitepaper admits the difference can be hard to see, then states it plainly: pilot light cannot process requests without action first, because servers must be turned on; warm standby can handle traffic immediately, at reduced capacity, and only needs to scale up. If a question says "must serve requests immediately after failover, even at lower capacity", pilot light is out.
| Strategy | Running in the recovery Region | RPO and RTO, cost |
|---|---|---|
| Backup and restore | Backups only; rebuild with IaC | Hours; lowest cost |
| Pilot light | Live data, core infrastructure; servers off | Tens of minutes; low cost |
| Warm standby | Scaled-down copy that already serves | Minutes; higher cost |
| Multi-site active/active | Full production, serving users | Near zero; highest cost |
Strategy definitions from the AWS whitepaper Disaster Recovery of Workloads on AWS, checked 27 October 2026. The RPO and RTO figures are AWS's usual relative scale, not a guarantee; real values depend on the design.
"Lowest cost", "can be down for a day", "protect against deletion or corruption".
"Replicate data continuously but keep compute costs minimal", "no servers running until a disaster".
"Serve traffic immediately, even at reduced capacity", "scale up after failover".
"Near-zero downtime and data loss", "users in several Regions served every day", "cost is secondary".
Replication is not backup. Continuous replication copies a corruption or a malicious delete as faithfully as good data. AWS's guidance is that every strategy, active/active included, still keeps point-in-time backups, and that failover should rely on data plane operations such as Amazon Application Recovery Controller routing controls rather than control plane changes.
Pilot light keeps data replicated but application servers off, so it cannot serve requests until servers are started. Warm standby runs a smaller, fully working copy that serves traffic immediately and only needs to scale up, so it recovers faster and costs more.
Backup and restore is the cheapest because only backups are stored in the recovery Region. It also has the highest RPO and RTO, since infrastructure must be redeployed and data restored after the disaster.
RPO, the recovery point objective, is the most data a business can afford to lose, measured in time. RTO, the recovery time objective, is the longest it can wait for the service to be restored. Lower values need costlier strategies.
AWS Elastic Disaster Recovery uses the pilot light strategy. It replicates servers at block level into a staging area and launches full-capacity recovery instances only when failover is triggered.
Yes. Replication between active Regions also replicates corruption and deletions, so point-in-time backups remain the only way back from a data disaster. AWS notes that recovery from data corruption always has a recovery point before the problem was found.
Chapter 39 of the Certified SAA-C03 course builds each strategy step by step and asks three scenario questions that land on different answers.
Open Chapter 39 →