Is Northern Virginia still the least reliable AWS region?

97 points by colinbartlett 20 hours ago on hackernews | 75 comments

This updated analysis is based on StatusGator outage data collected from January 1 to December 9, 2025. We decided to review our AWS analysis of outages in 2022 due to several new AWS incidents, especially another widely discussed AWS outage in us-east-1 (N. Virginia) that occurred on October 20, 2025.

We’ve expanded the report with fresh 2025 regional data as well as a new breakdown of affected AWS services.

The Data Behind the Study

StatusGator continuously monitors the official AWS status pages and aggregates incidents across every public AWS Region. This analysis reflects:

  • Major, publicly acknowledged AWS outages
  • All commercial AWS regions (GovCloud excluded)
  • Data timeframe: January 1, 2025 – December 9, 2025

AWS Outage Ranking by Region

So let’s take a look at the number of outages, duration, and components affected.

RegionNumber of outagesDurationComponents Affected
Regionless1231:55:1914
Canada-Central13:49:5719
Hyderabad10:44:5946
Ireland10:44:5110
N. Virginia1033:49:33126
Ohio21:20:452
Oregon32:59:413
Osaka12:15:0111
Sao Paulo10:44:519
Singapore10:54:591
Stockholm211:54:4981
Sydney10:50:001
Tokyo31:24:5118
Zurich14:54:557

Key Findings

N. Virginia (us-east-1) is once again the least reliable AWS Region.

It leads the dataset in:

  • Total number of outages (10)
  • Total downtime (33 hours, 49 minutes)
  • Total components affected (126)

No other region even comes close. Stockholm ranks second in downtime (11+ hours). Despite only 2 outages, each incident had a massive regional impact.

Regionless outages were unusually high. This category recorded 12 outages and 32 hours of downtime, indicating:

  • More widespread AWS service disruptions in 2025
  • More failures affecting multiple regions simultaneously

AWS Services with the Most Outages

AWS doesn’t just experience regional outages. Service-level incidents are just as impactful.
We analyzed the most frequently disrupted AWS services in 2025, ranked by the number of outages.

ServiceNumber of OutagesDuration
Amazon EC21419:14:01
Amazon SageMaker1120:40:21
AWS Glue1015:51:40
Amazon EMR1021:39:31
Amazon ECS1019:54:32

Key Findings

  • Compute services dominated the outage list, especially:
    • Amazon EC2 (core compute)
    • Amazon ECS (containers)
    • Amazon EMR (big data)
  • EMR had the longest duration among the top five (21 hours and 39 minutes).
  • SageMaker experienced more outages than expected for an ML service, an emerging reliability trend.

AWS Services With the Longest Outage Duration or Broadest Impact

These services didn’t always have the highest count, but had the longest or most severe incidents.

ServiceNumber of OutagesDuration
Amazon OpenSearch Service625:36:36
Amazon EMR Serverless725:30:08
Amazon CloudWatch624:58:49
Amazon Connect522:52:42
AWS STS522:48:39
Amazon VPC Lattice722:35:47
Amazon EMR1021:39:31
Amazon EventBridge521:24:32
Amazon Kinesis Data Streams521:15:00
AWS DataSync920:36:52
Amazon Elastic Load Balancing912:34:20
Amazon DynamoDB913:19:18
AWS Transit Gateway817:14:51
AWS Lambda813:50:15

Key Findings

  • OpenSearch, EMR Serverless, and CloudWatch each exceeded 24+ hours of cumulative downtime.
  • Mission-critical systems like STS, DynamoDB, Lambda, and ELB saw prolonged disruptions.
  • EMR appears in both spreadsheets, indicating it experienced frequent and long-lasting ones.

Across Tables 1–3 above, we see a consistent pattern emerge:

1. Many of the affected components were concentrated in N. Virginia

With 126 components affected, us-east-1 experienced the widest service disruption footprint.

2. Region-level outages and service-level outages are correlated

Major incidents involving:

  • EC2
  • SageMaker
  • EMR
  • CloudWatch
  • OpenSearch
  • STS

…almost always touch N. Virginia due to:

  • Higher customer density
  • More service deployment fronts
  • More inter-service dependency points
  • Heavier API traffic
  • Higher multi-AZ coordination complexity

3. The longest-running outages disproportionately affected us-east-1

Duration-heavy outages (CloudWatch, OpenSearch, EMR Serverless) frequently included N. Virginia, driving up the region’s total downtime.

Conclusion:

N. Virginia is not only the region with the most outages, but it is also the region where service outages cascade the widest and run the longest.

AWS Outage on October 20, 2025

On October 20, 2025, AWS experienced one of the most significant cloud outages in its history. 76 individual AWS components in the N. Virginia region alone showed disruption, by far the most heavily affected region.

Portions of Amazon Web Services were down for nearly 15 hours, causing cascading failures across thousands of SaaS platforms.

StatusGator’s Early Warning Signals detected the incident approximately ten minutes before AWS officially acknowledged it, ultimately identifying outages across more than 2,000 of the 6,000 services in our monitoring network.

However, the magnitude of the event meant StatusGator was also impacted, experiencing two periods of dashboard and status page downtime due to a surge in global traffic and failures in upstream infrastructure.

Despite these disruptions, StatusGator delivered over 100,000 outage notifications throughout the incident and has since implemented architectural improvements to strengthen reliability during future large-scale cloud failures.

Why Is N. Virginia Still the Least Reliable Region in 2025?

We revisited the three common theories from our 2023 AWS outage analysis and compared them against this year’s dataset.

Assumption 1: “N. Virginia Has More Services, So More Things Can Break”

In 2023, we found this explanation to be weak. But the 2025 “Components Affected” numbers tell a new story:

  • N. Virginia affected 126 components
  • Next highest: Stockholm with 81
  • Most regions affected ≤ 20 components

This indicates:

  • Broader blast when outages occur in N. Virginia
  • More interconnected or high-density service dependency
  • More potential points of failure

Still, high service count alone doesn’t explain the full scale:
Regions like Oregon and Ireland offer nearly as many services but have far fewer issues.

So the number of components contributes to complexity, but not the root cause.

Assumption 2: “N. Virginia Is the Most Used and Most Heavily Loaded Region”

This remains the strongest and most likely explanation. StatusGator monitoring AWS status data historically shows:

  • N. Virginia is monitored by almost 2× as many users as Oregon
  • And over 3× as many as many other U.S. and global regions

More customers → heavier load → more real-world stress → more outages that reach public visibility.

So this assumption is very likely true, and reinforced by 2025 data.

Assumption 3: “N. Virginia Is Older and Built Differently”

AWS provides no evidence that us-east-1 uses a fundamentally different architecture. And our 2025 numbers don’t suggest “old region issues”:

  • Tokyo and Sydney (both older) had minimal downtime
  • Newer regions, like Zurich and Hyderabad, had multi-hour outages

Like in 2023, we still see no evidence supporting this theory.

Summary: AWS Reliability in 2025

With only weeks left in 2025, the data is clear:

  1. Us-east-1 (N. Virginia) remains the least reliable AWS Region
  • Most outages
  • Most downtime
  • Most components affected
  1. Compute, analytics, and AI/ML services were the most outage-prone

EC2, SageMaker, Glue, EMR, and ECS led the list.

  1. Several AWS services experienced extremely long-running disruptions

OpenSearch, CloudWatch, EMR Serverless, and STS had over 24 hours of cumulative downtime.

  1. Multi-region outages increased

The Regionless category shows a notable rise in cross-region or global incidents in 2025.

Get Notified of AWS Outages Before AWS Reports Them

StatusGator aggregates every AWS service and region into a single unified dashboard.
We alert you instantly, often before AWS posts the incident publicly.

Get instant, account-specific AWS outage alerts through StatusGator’s unified dashboard, now enhanced with AWS Health integration for Enterprise customers. It delivers trusted, direct notifications about incidents, outages, and maintenance affecting your services, with built-in filtering to reduce noise, and seamless delivery to Slack, Microsoft Teams, Discord, Google Chat, and more.

Monitor AWS outages in real time with StatusGator — free to try.