Хочете будувати кар’єру в Data Science та AI? Vodafone Big Data Lab відкриває новий набір —>
Oksana StepuraHot News
16 November 2026, 13:59
2026-11-16
Microsoft went unnoticed for three months after a bug caused Microsoft 365 services to go down in 74 countries. What caused the global outage?
Microsoft has finally figured out what caused a massive outage of Outlook and other Microsoft 365 services. The culprit was an update the company installed back in July. The bug went unnoticed for three months until a combination of circumstances knocked out more than 80% of servers at one of Microsoft's network gateways.
Microsoft has finally figured out what caused a massive outage of Outlook and other Microsoft 365 services. The culprit was an update the company installed back in July. The bug went unnoticed for three months until a combination of circumstances knocked out more than 80% of servers at one of Microsoft's network gateways.
This is reported by The Register, citing a Microsoft report.
The problematic update was intended for servers through which users connect to the company's cloud services. At first, everything looked fine: the code was tested, it was installed, and no noticeable problems arose.
But it turned out that Microsoft was testing the update mainly on weaker servers with fewer processor cores . Instead, the hidden bug was particularly pronounced on newer, more powerful hardware.
Modern processors with many cores can perform more tasks at the same time. But sometimes these tasks need access to the same resource. Due to a bug in the code, Windows processes began to conflict for access to shared resources. The more tasks a server could perform in parallel, the more this conflict manifested itself.
Instead of efficiently handling user requests, the server would waste more and more resources working under this conflict. Eventually, the load on the processor and RAM would become so great that the server would stop responding.
On November 12, two things happened. First, Microsoft Teams traffic was redirected to the main network gateway that also serves Exchange Online, the email system that runs Outlook. At the same time, nearly 14% of the gateway’s server capacity was temporarily unavailable.
None of these situations were critical on their own. But together they created the exact load that triggered the hidden error. Then a domino effect began. When one server stopped responding, the system automatically redirected requests to another. That server would receive additional load, also fail, and requests would move on to the next one.
In less than two hours, over 80% of the gateway servers stopped responding.
How was it fixed?
Engineers tried to get the servers back up and running, but they would almost immediately fail again under the load. Adding capacity and redirecting traffic also didn't solve the problem. Normal disaster recovery didn't help, because Microsoft's automated mechanisms weren't designed to handle such a large-scale server failure.
Simply "rolling back" the update also didn't work, because the problematic code was installed three months ago.
So engineers had to actually write a patch for the old patch—a new update that reverted the problematic part of the code to its previous working state. It was gradually installed on servers, restoring the system to working order.
Previously, dev.ua reported how Microsoft's global outage disrupted the work of companies in 74 countries, including Ukraine.
Microsoft later restored its services after a 20-hour outage, while engineers were still investigating the cause, citing issues with request processing and network infrastructure as possible explanations.
"Microsoft is down. Teams isn't working, Outlook isn't working. But Excel, damn it, is immortal!": Social media exploded with memes about Microsoft's global outage