Incident Overview
An unexpected power outage occurred on March 1, 2026, at AWS’s me-central-1 region in the Middle East. This incident was caused by external objects striking a data center, igniting a fire and leading to a significant service interruption.
The outage primarily impacted Amazon Elastic Compute Cloud (EC2), networking APIs, and availability within the mec1-az2 zone. AWS reported that the local fire department required an immediate power shutdown, including backup systems, to address the situation, affecting multiple AWS services.
Timeline of the Outage
The disruption began at approximately 4:30 AM PST. AWS initiated investigations into the connectivity and power issues by 4:51 AM PST. By 6:09 AM PST, AWS confirmed the localized power failure affecting mec1-az2.
To mitigate the impact, AWS redirected traffic from the affected facility to other functioning zones within the region. Customers experienced errors with EC2 networking APIs, notably AllocateAddress and AssociateAddress, as AWS engineers worked to resolve these failures.
Throughout the day, AWS implemented several configuration changes to address API issues. Notably, the AllocateAddress API showed recovery signs by 2:28 PM PST, although challenges remained with the AssociateAddress API.
Mitigation Efforts and Recovery
By 6:01 PM PST, AWS confirmed the recovery of the AssociateAddress API, deploying an update that allowed customers to disassociate and reassign Elastic IP addresses from inactive resources to active zones.
Despite these efforts, the physical infrastructure remained offline, pending safety clearance from local authorities to restore power. AWS reiterated their commitment to safely restoring power and connectivity once permitted.
Customers were advised to use alternative Availability Zones or AWS Regions, launching replacement resources or restoring data from recent backups to mitigate the outage’s impact on operations.
Importance of Redundancy
This incident underscores the critical importance of deploying applications across multiple Availability Zones. AWS highlighted that customers with redundant systems faced minimal disruptions compared to those relying on a single zone.
As AWS continues recovery efforts, they cautioned that provisioning times might increase due to traffic rerouting from the affected zone. The company had no estimated time for power restoration as of the last update at 6:01 PM PST.
Stay updated with the latest developments by following us on Google News, LinkedIn, and X. For more information or to share your stories, contact us directly.
