Cloud Provider Status Report - December 2025

This report presents incident data from major cloud providers for December 2025, covering AWS, Azure DevOps, DigitalOcean, Fly.io, Heroku, Linode, Netlify, Railway, Render, and Vercel. The data includes both officially reported incidents from provider status pages and unconfirmed incidents detected by IsDown's monitoring system.

Important: Each provider has their own framework for reporting incidents. A provider showing more incidents does not mean they are less reliable - it may indicate more transparent reporting. These providers also vary significantly in size and market share, making direct comparisons inappropriate.

Summary Overview

Provider	Official Incidents	Total Incident Time	Unconfirmed Incidents
AWS	0	0 hours	2
Azure DevOps	6	~21.9 hours	0
DigitalOcean	8	~41.7 hours	0
Fly.io	13	~64.2 hours	0
Heroku	0	0 hours	0
Linode	7	~101.1 hours	0
Netlify	4	~12.2 hours	0
Railway	7	~28.5 hours	0
Render	5	~32.4 hours	0
Vercel	6	~9.1 hours	0

Provider Incident Details

AWS

AWS reported no official incidents for December 2025.

Unconfirmed Incidents: IsDown detected 2 potential issues on December 25, 2025, based on user reports. These were never officially acknowledged on AWS's status page.

Note: Unconfirmed incidents are potential issues detected by IsDown but never officially acknowledged by the vendor. These serve as indicators only and do not confirm an actual outage occurred.

Azure DevOps

Azure DevOps reported 6 incidents (all minor) totaling approximately 21.9 hours, with an average resolution time of 219 minutes per incident. Thursday and Friday saw the most incidents with 2 each.

The incidents consisted primarily of availability degradations affecting various services including Test Case Management (TCM), Team Foundation Server (TFS), and general user functionality. These occurred on December 4th, 5th, 8th, and 10th, with durations ranging from approximately 75 minutes to over 7 hours.

DigitalOcean

DigitalOcean reported 8 incidents (all minor) totaling approximately 41.7 hours, with an average resolution time of 313 minutes per incident. Thursday experienced the most incidents with 3.

The incidents affected various services throughout the month. A significant guardrails service disruption lasting nearly 25 hours occurred on December 2nd. Database management control plane operations experienced issues on December 4th. Additional incidents impacted control panel access, app platform static websites, recovery console accessibility, and API services. Regional-specific issues were reported in BLR1 for new app creation and NYC3 for static site deployments. The longest incident involved guardrails service disruption, while shorter incidents typically lasted between 40 minutes to 3 hours.

Fly.io

Fly.io reported 13 incidents (4 major, 9 minor) totaling approximately 64.2 hours, with an average resolution time of 296 minutes per incident. Thursday saw the most incidents with 6.

Network-related issues dominated the incident reports, with instability and connectivity problems affecting multiple regions including IAD (Ashburn, Virginia), SIN (Singapore), GRU (São Paulo), and JNB (Johannesburg). The IAD region experienced recurring network issues on December 4th and 8th, with some incidents extending over 10 hours. Major outages included Depot docker builder failures, network outages in SIN, and an extended connectivity issue in GRU lasting over 33 hours. Additional incidents affected the Machines API, managed Postgres clusters, and general networking with increased latency and packet loss across regions.

Heroku

Heroku reported no incidents for December 2025.

Linode

Linode reported 7 incidents (all minor) totaling approximately 101.1 hours, with an average resolution time of 866 minutes per incident. Monday experienced the most incidents with 3.

The incidents affected various services and regions throughout the month. Managed Databases experienced configuration and provisioning issues lasting nearly 7 hours on December 4th. Object Storage faced problems in Milan and Chicago, with the Chicago incident extending over 42 hours. Compute hosts across multiple regions experienced job completion issues lasting over 24 hours. Regional connectivity problems were reported in US-IAD (Washington) and US-Southeast (Atlanta). Block Storage in the IAD data center experienced performance issues with volume operations lasting over 8 hours.

Netlify

Netlify reported 4 incidents (all minor) totaling approximately 12.2 hours, with an average resolution time of 182 minutes per incident. Monday saw the most incidents with 2.

The incidents included brief network latency in São Paulo lasting about 13 minutes, function upload failures preventing site deployments for approximately 107 minutes, significant delays in Function log delivery through Log Drains lasting over 9 hours, and increased latency with Edge Functions on the High Performance Edge network lasting about 42 minutes. The log delivery issue was the most prolonged incident of the month.

Railway

Railway reported 7 incidents (2 major, 5 minor) totaling approximately 28.5 hours, with an average resolution time of 244 minutes per incident. Saturday and Wednesday each saw 2 incidents.

The incidents included brief backend connectivity issues, network latency problems in US East, and two separate Let's Encrypt-related outages affecting custom domain verification, with one lasting over 20 hours. Additional issues involved elevated network latency with slow deploys lasting over 5 hours, failing builds for mise package installs, and Canvas performance degradation affecting project updates. The Let's Encrypt incidents significantly impacted the ability to add new custom domains to Railway services.

Render

Render reported 5 incidents (1 major, 4 minor) totaling approximately 32.4 hours, with an average resolution time of 389 minutes per incident. Tuesday and Friday each saw 2 incidents.

The incidents included certificate provisioning issues for custom domains, increased latency for service updates in Oregon, a major outage affecting service accessibility, elevated latency for requests in Frankfurt, and significant deployment delays in Virginia lasting over 27 hours. The Virginia deployment delays represented the longest incident of the month for Render.

Vercel

Vercel reported 6 incidents (2 major, 4 minor) totaling approximately 9.1 hours, with an average resolution time of 91 minutes per incident. Tuesday and Friday each saw 2 incidents.

The incidents affected various platform features including domain registrations with elevated error rates lasting about 77 minutes, Speed Insights data loading errors, deployment issues with Neon and Supabase integrations lasting nearly 3 hours, elevated error rates across deployments and the Vercel API, Web Analytics and Speed Insights data recording problems, and Support Center case creation failures. Existing deployments and live traffic remained operational during most incidents.

Summary

December 2025 incident data shows varying levels of reported incidents across cloud providers. AWS and Heroku reported no official incidents, though IsDown detected potential issues for AWS on December 25th. Linode reported the highest average incident duration at 866 minutes per incident, while Vercel reported the lowest at 91 minutes. Network-related issues were common across multiple providers, particularly affecting Fly.io, DigitalOcean, and Railway. Several providers experienced region-specific incidents, with IAD (Washington/Ashburn) being mentioned by multiple providers including Fly.io and Linode.