AWS US‑EAST‑1 Outage Knocks Popular Apps Offline

On October 20, 2025, a major Amazon Web Services (AWS) outage in the US‑EAST‑1 (Northern Virginia) region caused widespread interruptions across many popular websites, apps and services. Users reported disruptions to Amazon’s Alexa, payment app Venmo, Snapchat, Fortnite, Canva and more as DNS resolution problems and subsequent EC2 launch errors rippled through the cloud provider’s systems.

According to AWS’s service health notices, engineers traced the root cause to a DNS resolution issue affecting the DynamoDB API early Monday. The DNS problem was reported as “fully mitigated” by AWS later, but dependent services continued to see increased error rates because of problems launching new EC2 instances and a backlog of requests.

• Timeline highlights: DNS/DynamoDB issue detected; mitigated later in the morning; residual problems persisted with EC2 instance launches.
• Services reported affected include: Alexa, Venmo, Snapchat, Lyft, Fortnite, Roblox, Reddit, Disney+, Apple Music, Pinterest and major news sites.
• AWS response: multiple mitigations applied, rate limiting of new EC2 launches in US‑EAST‑1 to aid recovery, and guidance to avoid tying deployments to specific Availability Zones.

Experts say the outage underscores the risks of heavy reliance on a small number of cloud providers. As of mid‑2025, AWS held roughly 30% of the global cloud infrastructure market — a concentration that can magnify the impact when a major region experiences trouble.

For the official timeline and updates, see the AWS Service Health Dashboard. For a contemporary report that compiled affected services and updates, see an updated news roundup here.

The outage illustrates how a single regional failure can cascade across internet services, affecting both major consumer apps and enterprise operations. Even after mitigation, recovery can take time as queued operations and instance launches are processed.

Discussion: Were any of your apps or services affected by today’s outage? How should companies design systems to reduce risk from single‑region cloud failures?