October 20, 2025 US-EAST-1 Region Incident

Report Overview

On October 20, 2025 AWS reported a service event  in the US-EAST-1 region (Northern Virginia) that resulted in temporary disruptions across multiple services and a potential financial impact for AWS customers. 

This advisory will help you: 

1. Understand the cost implications of this event
2. Identify any anomalies in your usage data
3. Understand AWS SLA’s and navigate  the credits process

What Happened

On October 20, 2025, from approximately 3:00 AM to 6:00 PM ET, AWS experienced intermittent service degradation in the US-EAST-1 region.

AWS provided regular updates through its Health Dashboard and ongoing communication with customers and partners.

The event stemmed from a DNS resolution failure affecting key services such as EC2, Lambda, load balancers, and API Gateway. Major platforms and organizations experienced temporary disruptions, and AWS identified and remediated the issue, restoring stability later that same day.

This means that If you run infrastructure in US-EAST-1, some workloads may have experienced reduced performance or intermittent availability during the event window, which could, in turn, result in temporary cost anomalies. These may appear as billing errors; they are actually a normal side effect of large-scale service interruptions.

How This Affects Your Cloud Costs

Review your cloud analytics for the following cost patterns, which may have resulted from the event. 

1. Charges for Limited-Use Services

Resources that kept running but couldn't serve traffic: For example, EC2 instances with no connectivity, idle databases, and unused load balancers. 

You might be billed for services that delivered minimal business value.

2. Retry Storm Cost Spikes

Applications automatically retrying failed requests created artificial cost explosions:

• Lambda functions timing out and re-executing
• API Gateway request multiplications
• Data transfer overages because of retry logic
• CloudWatch Log surges for error logging

These patterns represent system-resilience mechanisms, not intentional use, and can be analyzed to isolate nonproductive costs. 

3. Failover & Recovery Costs

If you triggered disaster recovery to other regions, you may see unexpected charges for cross-region data transfer, temporary resource scaling, and backup restoration.

What to Look For

Map your usage anomalies to distinguish event-driven costs from legitimate business activity. 

Look for the following:

• Event-driven spikes: Unusual increases in Lambda invocations, API calls, or data transfer caused by retry loops and failures
• Idle resource charges: Normal billing for resources that were actually unavailable and not delivering value
• Real usage: Legitimate business activity that occurred outside the event window or in unaffected regions

Look for short-term deviations (Our customers’ environments often report 5-20x baseline levels during the 3:00 AM - 6:00 PM ET event period). These patterns can provide valuable context when discussing potential SLA adjustments with AWS.

Next Steps

1. Document Your Impact (This Week)

• Identify all resources in US-EAST-1 using AWS Cost Explorer
• Export hourly costs for October 20, 2025 (3:00 AM - 6:00 PM ET)
• Capture metrics showing service unavailability (CloudWatch errors, timeouts)
• Calculate costs during the event vs. normal operations

2. File Your SLA Credit Claim

AWS defines SLA thresholds in its Compute SLA.
CRITICAL DEADLINE: December 31, 2025. You must submit your claim within two billing cycles of the incident.

How to File:

• Open a case in AWS Support Center
• Subject: "Amazon Compute SLA Credit Request – Region-Level Claim" (or Instance-Level)
• Include: dates/times, affected resources, error logs showing unavailability

• Expected Credits: Based on AWS Compute SLA, this event aligns with an approximate ~98% uptime, potentially qualifying for a 10% service credit on impacted workloads. Credits typically appear on your nextbill

3. Recover Beyond-SLA Costs

Retry storms and failover costs aren't automatically credited. The following services require a separate AWS Support ticket: 

• Lambda retry overages
• Data transfer spikes
• API Gateway cost explosions
• Cross-region failover expenses

Reference the October 20 incident and present cost comparisons showing these charges directly resulted from AWS infrastructure failure.

4. Verify Credits Applied

Monitor your next billing cycle to confirm credits are reflected. If not, re-open your support case for review.

Summary & Key Takeaways

✓ File SLA claims by December 31, 2025—this deadline is non-negotiable

✓ Standard SLA provides 10% credit on affected compute resources

✓ Retry storms require separate claims through AWS Support escalation

✓ Document everything ASAP—logs and cost data become harder to retrieve over time

✓ Multi-region architectures qualify for better SLA commitments (99.99% vs 99.5%)

AWS has a strong track record of transparency and responsiveness in these processes. Be sure to engage your AWS account team early for direct support.

‍

Resources & Contact

AWS Official Resources:

• AWS Compute SLA: aws.amazon.com/compute/sla
• AWS Health Dashboard: health.aws.amazon.com/health/status
• AWS Support Center: console.aws.amazon.com/support

PointFive Contact: 

📧 Email: hello@pointfive.co
🌐 Website: pointfive.co
💼 Schedule Consultation: pointfive.co/contact

About PointFive: PointFive pioneered CEPM to help organizations proactively optimize cloud resources and identify hidden cost inefficiencies, including event related anomalies that traditional tools miss. ​​Our proprietary DeepWaste™ engine uniquely identifies hidden inefficiencies, uncovering significant cost-saving opportunities typically missed by traditional tools. PointFive integrates seamlessly into your engineering workflows with minimal setup, ensuring immediate, measurable impact.

This advisory is provided for informational purposes. While we strive for accuracy, customers should verify all information with AWS documentation and their AWS account team. SLA terms and credit eligibility are determined solely by AWS.

Published: October 22, 2025 | Incident Date: October 20, 2025 | Affected Region: US-EAST-1

‍