Amazon Web Services customers across its busiest region spent part of Sunday, November 15, locked out of logins, blocked from creating new accounts, and struggling with basic network connectivity after a multi-hour disruption hit the US-EAST-1 region. AWS’s own status updates put the start of the incident at 3:05 a.m. Pacific time, with the last lingering network problems not clearing until 6:21 a.m. — more than three hours of degraded service for some customers.
Sign-Ins, Sign-Ups, and Connected Services All Hit
The outage didn’t take down a single service so much as the plumbing several services depend on. Customers using AWS IAM Identity Center to sign into the AWS Management Console reported failures, as did users of Amazon Connect and the Kiro development environment. New account sign-ups also failed during the window, according to AWS’s own incident notices posted to its status page and social channels.
AWS engineers say they were automatically paged within five minutes of the first failures, at 3:10 a.m. Authentication problems were resolved by 5:20 a.m., restoring console sign-in and bringing most downstream services back online. A smaller slice of customers relying on Network Load Balancers in a specific infrastructure “cell” kept experiencing connectivity failures for roughly another hour, until the 6:21 a.m. all-clear.
No Attribution to an Attack
AWS’s recovery statement described the event in purely operational terms — authentication failures and connectivity problems — and stopped short of naming a root cause. The company has not pointed to any malicious activity, and nothing in its public updates suggests the outage resulted from an intrusion, a distributed denial-of-service attack, or any other adversarial action. Absent further technical disclosure, the incident is best read as a service-availability failure rather than a security breach.
That distinction matters, but it doesn’t make the outage irrelevant to security teams. The disruption landed in the same week as a separate, large-scale Microsoft 365 outage that knocked out Exchange Online, Teams, SharePoint, and Microsoft’s Defender XDR security suite — a reminder that when a hyperscale cloud provider stumbles, it can simultaneously take down the tools defenders rely on to watch for everything else.
A Familiar Pattern for US-EAST-1
US-EAST-1 is AWS’s oldest and most heavily used region, and it has a history of outsized outages given how many services and customers route through it by default. AWS has previously acknowledged delays in the same region affecting EC2 deployments and related container services. Sunday’s incident doesn’t appear to share a confirmed root cause with that earlier event, but the recurrence underscores a structural risk: a huge share of the internet’s authentication, commerce, and operational infrastructure is concentrated in a small number of regions operated by a small number of providers.
What Security and IT Teams Should Take From This
- Identity and access management that depends on a single cloud region’s availability is also a single point of failure for incident response — if your SOC can’t authenticate, it can’t triage.
- Multi-region or multi-provider failover for identity services is expensive to build and test, but outages like this one are the reason that investment keeps getting revisited.
- Outage windows are also opportunity windows: threat actors have a track record of exploiting the confusion around real outages with phishing campaigns that impersonate the affected vendor’s support channels.
- Document and rehearse what “AWS is down” actually means for your organization’s own security tooling, logging pipelines, and alerting — not just customer-facing applications.
AWS has advised affected customers to retry any requests that failed during the outage window and says all services are now operating normally. The company has not indicated whether a more detailed post-incident report will follow, which is typical practice after major regional disruptions of this scale.
Leave a Reply
You must be logged in to post a comment.