Amazon Web Services outage history

Europe faced an elevated packet loss issue in the EU-SOUTH-2 region, with the most pronounced impact in Spain due to a single Availability Zone (eus2-az1). Users experienced degraded performance and API latencies across a broad set of AWS services, including Spain / AWS Internet Connectivity, CloudTrail, Cognito, Direct Connect, EC2, SageMaker, and Step Functions. The incident began around 4:16 PM UTC and, as engineers investigated, AWS confirmed the root cause and continued mitigation efforts, with signs of improvement appearing by 6:51 PM UTC and ongoing updates through the afternoon.

Separately, a long-running disruption in the Middle East (me-central-1) affected UAE/Bahrain resources across multiple AZs. After weeks of impact, AWS indicated progress toward full recovery, and by 1:00 PM UTC a large set of UAE/Bahrain components had moved to up, signaling recovery from the ME-region disruption that had persisted since March.

In the US East (N. Virginia) region of Amazon Web Services, users attempting to launch new EC2 instances faced failures and elevated API latency as error rates surged. The issues primarily affected new provisioning workflows (RunInstances, CreateFleet and related EC2 launch processes), while existing instances continued running. The incident began around 10:49 PM UTC and showed early signs of recovery by 12:06 AM UTC as AWS engineers actively worked to restore normal service and advised retrying failed requests.

Status updates from AWS reflected ongoing investigation and remediation efforts, with guidance to retry operations and observations of improving operation over the course of the night. The impact appeared focused on EC2 provisioning in the affected region rather than a full-region outage, and as recovery progressed, more EC2-related workflows began to return to normal.

AWS experienced a multi-region disruption that primarily affected Middle East infrastructure, with the UAE MEC1-Central region and Bahrain ME-SOUTH-1 damaged and unable to reliably support customer workloads. While some workloads continued to function, resources hosted exclusively in the affected Regions were inaccessible. AWS advised customers to migrate workloads to other Regions and restore inaccessible resources from remote backups where possible.

Users in the AWS Oregon region began experiencing authentication failures as AWS reported increased API error rates for STS and Sign-In when using SAML. The issue started at 9:12 PM UTC and there was no workaround at that time. AWS stated that their engineering team was automatically engaged at 9:19 PM UTC to investigate the root cause, with updates indicating more information would follow. The disruption affected authentication flows and credential exchange, impacting Sign-In, STS calls, and related IAM identity services in the Oregon region and services that rely on SAML-based sign-in.

The Middle East (ME-CENTRAL-1) region experienced damage tied to the regional conflict, leaving customer applications in that area unable to be reliably supported. AWS advised customers to migrate workloads to other Regions and restore from remote backups where possible. While some workloads in ME-CENTRAL-1 continued to function, the region’s reliability was significantly degraded for an extended period, affecting a broad set of services and resources in that region. Impact began around 1:50 PM UTC and persisted into early September as recovery efforts progressed.

Around 2:12 PM PDT on 2026-09-03, AWS began reporting increased API error rates for STS and Sign-In with SAML in the US-West-2 area, with engineers automatically engaged at 2:19 PM PDT to investigate. There was no workaround available at the time, and AWS committed to another update by 11:30 PM UTC. Over the next hours, partial recovery signs appeared for the impacted STS/SAML APIs in US-West-2, with updates at 11:30 PM UTC and 11:45 PM UTC noting stabilization and error rates returning toward pre-event levels while continuing monitoring.

By 2026-09-10, most services had recovered across affected regions and status updates indicated widespread restoration, with ME-CENTRAL-1 remaining the region facing disruption. Other regions and services gradually returned to normal operation as updates continued and monitoring persisted.

Amazon Web Services (AWS) experienced a regional disruption in the ME-Central-1 region (Middle East, UAE) beginning at 1:46 PM UTC. The damage affected the region's ability to reliably support customer applications, with many services in UAE and Bahrain listed as degraded. While some workloads continued to function, AWS urged customers to migrate accessible resources to other Regions and restore inaccessible resources from backups; some billing operations were affected during the incident. The disruption persisted for most of the day and caused widespread impact across compute, storage, database, and management services across the region.

An AWS regional disruption in the Middle East, ME-CENTRAL-1 (UAE), led to widespread degradation for many services in that region. AWS stated the region had sustained damage and could not reliably support customer applications. While some workloads continued to function, a large portion of services faced interruptions and degraded performance. Customers were advised to migrate accessible resources to other Regions and to restore unavailable resources from remote backups, as some billing operations were affected. A broad set of UAE-region services, including API Gateway and CloudWatch, were listed as degraded during the incident.

An AWS regional incident began around 2:06 AM UTC with increased error rates affecting real-time metrics in AP-NORTHEAST-1. A follow-up at 2:41 AM UTC described damage in the Middle East UAE region ME-CENTRAL-1 that prevented reliable support for customer applications, and AWS urged migrating workloads to other Regions and restoring from remote backups where possible. Some Tokyo-area services (including Amazon Connect) also faced degraded performance during the event but gradually recovered. The issue was regional in scope, impacting authentication and monitoring workflows for users in affected regions.

The incident began around August 18, 2026 after AWS launched a new Availability Zone in EU-WEST-2 (euw2-az4), which led to increased error rates when launching new EC2 instances or creating resources in that region. Existing running resources were reported as unaffected, and AWS recommended using alternate AZs (euw2-az1, euw2-az2, and euw2-az3) to mitigate impact while operations stabilized. The disruption persisted for several days as the new AZ and surrounding services reconciled.

Amazon Web Services Direct Connect users in the EU-CENTRAL-1 region (Frankfurt) faced sustained packet loss and degraded connectivity. The root cause was a facility infrastructure issue at the Equinix FR5 location in Frankfurt, DEU, which left customers with single-site connections particularly affected while those with redundant paths were less impacted. Impact began around 3:46 AM UTC and persisted with ongoing updates for several days.

AWS and its co-location partner undertook a phased remediation across multiple workstreams to restore connectivity, including addressing the facility issue, hardware replacements, electrical isolation, and traffic rebalancing. For immediate recovery, AWS recommended failing over to VPN where available. Over the following days, engineers made incremental progress; full recovery was not guaranteed in the near term at times, with updates continuing until most affected connections were back online by 1:01 PM UTC.

Geographically, the incident was localized to the Frankfurt FR5 site; other Direct Connect locations remained largely unaffected. The incident represented a major disruption for customers relying on that single path, with a long remediation timeline and phased restoration. By the latest update, most connections had recovered and service was returning to normal in the region.