Microsoft Outage Disrupts Work Worldwide for 10 Hours
The Microsoft outage brought work to a standstill worldwide for more than 10 hours. The email service Outlook, the office productivity suite Microsoft 365 (MS365), and the artificial intelligence (AI) service Copilot all became unusable, hitting companies and government agencies around the world, including South Korea.
Signs of the outage began appearing from midday U.S. local time on the 12th. According to DownDetector, which tracks online service outages, reports of Outlook failures in the United States surpassed 138,000 around 1 p.m. Users reported being unable to send or receive mail or access their inboxes. The outage, which began in the United States, was also confirmed across the Asia-Pacific region, including India, Australia, and Japan.
Governments responded as well. In Japan, government IT infrastructure was affected, and the Japanese government was reportedly reviewing the situation. In Australia, the National Cyber Security Coordinator monitored the outage. South Korea was no exception. According to Yonhap News, some major Korean companies experienced delays in sending and receiving Outlook emails or were unable to access Copilot.
Microsoft initially attributed the connection errors to increased infrastructure load. However, its subsequent investigation found that rather than a rise in connection volume from all users, a large number of servers unexpectedly went offline, concentrating the load on the remaining servers. Recovery took time. Servers that were restarted initially functioned normally but stopped responding again once actual user traffic arrived. Microsoft, analyzing the cause with network and platform experts, even tested measures such as limiting traffic to some servers and rebuilding them with a previous software configuration.
The damage was not confined to Microsoft's own services. OpenAI, the developer of ChatGPT, investigated problems in which new connections through Microsoft integration features failed or some tools did not work properly. The AI company Anthropic also said that MS365 integration features had malfunctioned due to an issue with an external service provider. The ChatGPT and Claude services as a whole, however, did not go down.
As of Microsoft's outage notice released on the morning of the 13th, Korean time, services were still in a degraded state. While access had been restored in some regions, whether the entire service had been normalized remained unconfirmed.
