Redundancy adds alternate components or resources so a system can keep operating when something fails. But a duplicate is not automatically a resilient fallback: if both paths share the same flaw, both can fail together. section 10.7.2
Two computers, one failure
Ariane 5 had a backup navigation computer, but it ran the same software as the primary. When an overflow crashed one, the other hit the same code and crashed too. A useful fallback needs a different way to respond, such as returning a cached result or gracefully degrading the experience. section 10.7.2
Design the alternate path around the failure you need to survive:
- Use a queue to hold messages while a database is unavailable, then process them when it returns. section 7.6.5
- Use database replicas for failover or regional reads, but keep offline backups too: destructive changes can spread to replicas. The 3-2-1 rule calls for three copies, two storage types, and one offsite or cross-region immutable copy. section 10.5.4
- Match storage redundancy to business needs: reducing recovery time or data loss can increase cloud storage costs. section 10.5.4












