Mechanisms are implemented to achieve resilience requirements in normal and adverse situations

Implementation steps

1. 1

   Identify single points of failure in critical services

   For each service in your organization's critical path, map the components required for it to function and identify any single points of failure: a single database with no replica, a single load balancer, a single DNS provider, a single region. Each single point of failure is a risk that should be addressed, accepted with a documented recovery plan, or transferred.

   confluencemirolucidchart
2. 2

   Implement redundancy for critical components

   Address identified single points of failure based on criticality: deploy database replication, use multi-AZ or multi-region deployments for critical services, use multiple load balancers behind a DNS failover, and distribute critical services across availability zones. Test failover capability by deliberately failing components in a non-production environment.

   awsgcpazurecloudflaredatadog
3. 3

   Define and test recovery time objectives

   For each critical service, define a recovery time objective (RTO): how long is acceptable to be down? Design your resilience controls to meet that objective and test them. Run a quarterly or annual disaster recovery test where you simulate the loss of critical components and time the recovery. Use the results to identify gaps between your RTO target and actual recovery time.

   awsgcpazure

Evidence required