Cloud Infrastructure
AWS Data Losses in Middle East After Regional Conflict Expose Redundancy Gaps
Infrastructure damage in UAE and Bahrain has left AWS customers unable to recover data held only in destroyed facilities, reigniting debate over cloud vendor promises and customer responsibility.
InfoQ
Amazon Web Services has confirmed it cannot restore customer data stored exclusively within damaged infrastructure across its Middle East data centers following conflict-related strikes. The losses affect two regional endpoints—UAE (me-central-1) and Bahrain (me-south-1)—where physical damage has exceeded the limits of the company's multi-availability-zone design.
The Damage and Recovery Status#
AWS provided separate assessments for each region. In the UAE, one of three availability zones (mec1-az2) suffered irreversible damage, though the company continues attempting to restore resources in the other two. Bahrain's situation is more severe: the destruction spanned multiple zones and surpassed what the company's regional redundancy architecture was built to handle. No recovery is possible for data confined to that region.
The incident traces back to previous reporting of drone strikes in March that disabled three data centers across both regions. In July, additional strikes targeted Bahrain. AWS has notified authorities and promises updates on restoration efforts, with detailed Bahrain recovery information expected in early 2027.
Renewed Debate Over Cloud Resilience Promises#
The technical failure has surfaced broader questions about what customers actually understood they were purchasing. A resurfaced 2025 interview featured an AWS executive asserting that destroying an unidentified data center would go unnoticed by users—a claim that now rings hollow. Community discussion on technology forums centered less on geopolitics than on the gap between vendor marketing and documented obligations.
The core tension: AWS markets multi-AZ architecture as a resilience solution, and documentation does promise 99.999999999% durability via redundancy across three zones within a region. Yet that same documentation explicitly requires customers to replicate critical data across geographic regions—a step many organizations apparently did not take or could not take.
Responsibility and the Shared Model#
AWS emphasizes its "shared responsibility model," where the company provides the infrastructure platform while customers design their own disaster recovery strategies. The company's disaster recovery guidance plainly states that all strategies require regional backups plus off-region copies. That caveat appears in foundational AWS certifications.
