AWS Says It Can't Restore Some Data from Mideast Facilities Struck by Iran

Commenters reacted to reports that AWS cannot restore some data after strikes on Middle East facilities by stressing both surprise and the predictable limits of cloud design. Some pointed to an earlier confident interview (cmiles8 referenced the Pogue/Wood exchange) and said that trust in provider guarantees has been undermined (krick, mitxela). Others argued this is a customer-responsibility issue: several commenters reminded readers that multi-AZ or cross-region replication is a design choice and that single-AZ storage or local regulatory residency requirements can leave data vulnerable (advisedwang, ernsheong, sparkling, rishikeshs). A few called it a basic DR failure (Art9681), while others emphasized the fine print - “designed for” durability versus guaranteed recovery and Acts-of-War exemptions in SLAs (stackskipton, advisedwang, jonahx).

Opinion diverged on cause and remedy. Some said the damage exceeded assumptions because multiple AZs were hit by the same attack and that war-level failures are often excluded from standard failure models (phendrenad2, dhx). Others warned this exposes systemic risk from centralization and data-residency policies (wewewedxfgdf, carefree-bob, nixass). Suggested mitigations ranged from cross-region replication and hardened bunkers to accepting higher DR costs or rethinking where encryption keys and administrative access reside (gregw2, tgsovlerkhgsel, dannyobrien). Political blame also surfaced, with some commenters tying the fallout to foreign policy and billionaire behavior (SmirkingRevenge, shevy-java).