Mastodon Mastodon Mastodon Mastodon

24-hour cascading Microsoft 365 failure: how three factors took down 80% of gateways

Photo of author

CyberSecureFox Editorial Team

Published:

On 12 November 2026, a global Microsoft 365 outage left millions of users around the world without access to corporate email, calendars and related services. According to a preliminary report reproduced by Office Watch, the incident lasted almost 24 hours — from 16:20 UTC on 12 November to 16:10 UTC on 13 November — and was caused by a combination of a latent software defect in network gateways, automatic rerouting of Teams traffic, and a prior loss of server capacity. Exchange Online was hit hardest: most users could not open email via Outlook. At least 14 services were affected, including Copilot, Teams, SharePoint, OneDrive, Defender, Purview, Fabric, Intune, Planner, Power Apps, To Do, Power BI and Universal Print. According to the source material, Downdetector recorded more than 7.4 million problem reports, although this figure has not been independently confirmed.

Three factors behind the cascading failure

According to Microsoft’s preliminary report, the root cause was a defect in connection handling on Microsoft 365 network gateways. Under certain load conditions a Windows system process experienced lock contention, which led to exhaustion of CPU and memory resources and a complete server stall. The problematic code had been deployed back in July 2026, went through the standard staged rollout process, and had been running for more than three months without any visible consequences.

The reason this defect was not discovered during testing is telling: lab environments used less powerful hardware with fewer CPU cores, whereas production ran on newer servers where the problem manifested much more severely. The gap between test and production environments is a classic — yet still very relevant — reliability pitfall.

Two additional factors turned the hidden bug into a global incident:

  • Rerouting of Teams traffic. Around 15 hours before the outage, automation shifted Teams traffic from an alternate gateway to the primary gateway that was serving Exchange Online. This is a standard failover mechanism, but it changed the volume and pattern of load on the target gateway.
  • Loss of capacity. Due to a separate malfunction unrelated to the main defect, roughly 14% of the gateway’s server capacity was already unavailable.

When load in the US approached the daytime peak, servers began to hang. The system rerouted requests to the remaining healthy gateways, but those quickly ran into the same problem. According to the report, in less than two hours more than 80% of network gateway servers stopped responding.

Failure of regional isolation

The key question of the incident is why the cascading failure spread globally. Microsoft acknowledges that the regional isolation mechanisms (bulkheads) designed to prevent an outage from spreading between regions did not work. The company does not disclose the reason for this failure in the preliminary report, referring to it as a “separate issue”. This means the infrastructure contained at least two independent defects at the same time: in the connection-handling code and in the isolation mechanism.

Standard recovery methods proved ineffective: restarted servers immediately received the accumulated request queue and froze again. A quick configuration rollback did not help either, since the defective code had been present on the gateways since July.

Asymmetric service recovery

Service restoration was uneven, and this asymmetry is worth noting. Teams, OneDrive, SharePoint and a number of Office applications were able to be moved to alternative infrastructure, bypassing the faulty gateways. This is why Teams chats and calls continued to work for most users for a significant part of the incident.

However, Exchange Online could not be switched to a bypass route due to scaling constraints. This makes the mail service the most vulnerable component of the Microsoft 365 ecosystem when network gateways fail — a fact organizations should take into account when planning business continuity.

The scale of the impact

According to the source material, the outage affected government agencies in the US, Australia and Japan, the UK’s NHS and the London police. The European exchange Euronext reported issues with email, and a number of financial firms in Taiwan were forced to confirm trades by fax. Indian airlines Air India, IndiGo and Akasa Air warned customers about problems with email support. These data points are cited with reference to Reuters, but the original source was not available for independent verification.

Microsoft specifically stresses that the incident was not related to a cyberattack. Nevertheless, the scale of the outage clearly demonstrates how much damage an operational failure of cloud infrastructure can cause — comparable to the consequences of a targeted attack. We have previously written about how attackers attack Microsoft 365 via token theft, and an outage of this kind creates additional risks: when email is unavailable, employees may switch to less secure communication channels, expanding the attack surface for phishing and social engineering.

Measures taken and recommendations

To restore operability, Microsoft engineers prepared an update that rolled back the connection-handling code to its previous working state and gradually brought servers back online. According to the preliminary report, the following actions were carried out:

  • Rollback of the defective connection-handling code to a known working state
  • Increase of gateway server capacity
  • Addition of monitoring for traffic rerouting patterns and connection-handling state
  • Implementation of limits on traffic redistribution to prevent cascading failures

Microsoft planned to publish the final report on 20 November 2026.

For organizations that depend on Microsoft 365, this incident is a reason to revisit business continuity plans with a specific scenario in mind: prolonged unavailability of Exchange Online while other services remain partially operational. Key actions include ensuring a backup business communications channel that does not depend on Exchange, testing procedures for working when corporate email is unavailable, and tracking Microsoft’s final report to understand the long-term measures to address the regional isolation issue, which failed in this incident.


CyberSecureFox Editorial Team

The CyberSecureFox Editorial Team covers cybersecurity news, vulnerabilities, malware campaigns, ransomware activity, AI security, cloud security, and vendor security advisories. Articles are prepared using official advisories, CVE/NVD data, CISA alerts, vendor publications, and public research reports. Content is reviewed before publication and updated when new information becomes available.

Leave a Comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.