Due Diligence Report: Disaster Recovery & Single-Provider Dependency
Testing whether a platform's stated resilience survives a real regional provider outage, not just a planned drill.
Executive Summary
This review mapped the full dependency chain behind the platform's resilience claims and tested it against a simulated regional outage of its primary cloud provider.
Findings
DNS, authentication, and the primary database were all hosted with a single corporate cloud provider, with no tested failover to an alternate provider.
Recommendation: Re-architect at least authentication and DNS across providers; this is the highest-leverage resilience investment available.
Internal status dashboards reflected only the monitored services' own health, not the health of the full dependency chain, producing false-positive "nominal" status during a simulated outage.
Recommendation: Extend monitoring to the full dependency chain, including upstream provider status, not just owned services.
The most recent full disaster recovery drill predated the platform's most significant architectural changes, making its results no longer representative.
Recommendation: Run disaster recovery drills on a fixed cadence tied to architectural change, not a calendar default.
Conclusion
The platform's resilience claims were accurate for the architecture that existed at the time they were last tested, which was materially different from the architecture in place during this review. Resilience claims decay silently as systems change.
Frequently Asked
Why do disaster recovery drills need to be re-run after architecture changes?
A drill validates resilience for the architecture that existed when it was run. Subsequent changes can silently invalidate those results without anyone noticing until a real outage.