Millions of Americans lost access to email and critical business services Friday as Microsoft 365 suffered a cascading server failure that engineers struggled to contain for over 24 hours, with the tech giant admitting its recovery efforts repeatedly failed to restore normal operations.
Server Failure Spreads Across Multiple Regions
The outage began when multiple servers unexpectedly went offline, forcing remaining infrastructure to handle loads it could not sustain. Microsoft confirmed the problem was not caused by increased user traffic but by a core process on affected servers becoming overwhelmed and unresponsive. The disruption remained unresolved throughout Friday in the United States and extended into the workday across Asia-Pacific nations as the company worked to contain the damage.
By 6:30 p.m. Eastern Time, Microsoft acknowledged its initial recovery actions had not provided anticipated relief. Engineers attempted to isolate portions of infrastructure, restart affected components, and reintroduce traffic gradually. However, servers that were successfully returned to service became unresponsive again once they received traffic, forcing technicians back to diagnostic analysis.
Recovery Efforts Repeatedly Fail
Microsoft implemented multiple recovery strategies throughout the evening, including temporarily limiting traffic types to affected infrastructure and rebuilding servers using previous configurations. At 10:13 p.m., the company reported efforts to improve service availability were not producing expected results. By 11:36 p.m., Microsoft announced partial service recovery in some regions but admitted the improvements could not be sustained.
Downdetector recorded more than 130,000 problem reports by 2 p.m. Thursday. The tracking website also showed user issues with Xbox and Amazon services, though the extent of those problems remained unclear. Microsoft stated it was identifying alternate approaches to increase capacity while facing sustained traffic management imbalances.
Expert Analysis Points to Scale Problem
Attorney and cybersecurity expert Leeza Garber explained the technical challenge facing Microsoft. The company was dealing with elevated load on a reduced number of available servers, creating a cascade effect. Garber noted this type of problem appears simplistic but at Microsoft’s operational scale can shut down everything. Similar outages have hit Amazon and Google over the past decade, some affecting global operations.
Microsoft identified failed health checks potentially originating from an unidentified component responsible for processing server health. The company took steps to prevent additional servers from being impacted while engineers continued working to isolate the core issue affecting server availability and pursued multiple recovery workstreams simultaneously.
Sources
Foxbusiness: Microsoft 365 hit by multi-region connectivity issues amid elevated infrastructure load