Amazon Web Services (AWS) said a localized power failure triggered by a fire at one of its data centers disrupted operations across multiple availability zones in its Middle East region over the weekend, affecting core cloud services for several hours.
The incident occurred March 1–2 in the ME-CENTRAL-1 region, impacting availability zones mec1-az1, mec1-az2 and mec1-az3. The company said the disruption originated in mec1-az2 after an object struck the facility, sparking a fire that required intervention from local fire services. Power was shut off to the data center and its generators, leading to widespread service interruptions.
Customers reported elevated error rates and latency across key services, including EC2 virtual servers, DynamoDB databases and S3 storage. EC2 API errors and instance launch failures were recorded, with new instance launches largely unavailable in the affected zones during the peak of the outage.
Certain networking-related APIs, including AllocateAddress, AssociateAddress, DescribeRouteTable and DescribeNetworkInterfaces, were also impaired, complicating recovery efforts for some customers. Existing workloads running in unaffected availability zones continued operating normally, the company said.
AWS said it implemented mitigation measures, including enabling customers to disassociate Elastic IP addresses from affected resources and redirect traffic to operational zones. Clients were advised to fail over to alternate availability zones or regions and to back up critical data.
Over several hours, AWS reported incremental improvements in API functionality and networking services, though full restoration of impacted resources was expected to take additional time. The company said customers with architectures spanning multiple availability zones were not affected.
The disruption underscores the operational risks facing hyperscale cloud providers as demand for regional data infrastructure expands. Availability zones are designed to isolate failures and provide redundancy within a region, but concentrated physical incidents can still ripple across interconnected systems. AWS said it continues to work on restoring power and normal service operations across the affected facilities.
