One outage, millions affected. What went wrong at Microsoft?
retained text
One outage, millions affected. What went wrong at Microsoft? Problems with Exchange Online spread across Microsoft 365 and external businesses, highlighting how dependent workplaces have become on shared cloud infrastructure. Outlook marks the official start of the working day for millions globally. It also serves as a hub for daily workplace communication and scheduling across the world. So, when Microsoft’s email service began failing on Thursday, November 12, the disruption quickly spread beyond missed messages to calendars, workplace tools, and businesses that depend on its cloud systems. The massive outage continued into Friday, November 13, showing how a fault in one widely used service can cause ripples across the entire work ecosystem. What happened and when Microsoft’s incident records list Exchange Online incident EX1550543 as starting at 4.17 pm UTC on Thursday, or 9.47 pm in India. Users were unable to connect to their Exchange Online mailboxes through multiple connection methods. Microsoft Exchange Online is the back-end server that stores and manages user data, while Microsoft Outlook is the front-end app used to view and interact with the data. The wider Microsoft 365 incident, MO1550821, which disrupted multiple Microsoft 365 enterprise and consumer services worldwide, began at 6.26 pm UTC, or 11.56 pm IST. First reports came from the Americas, but Microsoft later warned that users in any region could be affected. Downdetector, a real-time platform that tracks online service disruptions and outages, recorded more than 1,30,000 reports of Outlook problems in the United States at its peak on Thursday afternoon. In India, complaints began shortly before midnight and reached 2,157 at 11.42 pm. Reports rose again to 1,960 at 8.42 am on Friday as people started work. In its latest post on X, Microsoft said traffic-optimisation measures were producing positive results for service stability and capacity. It was applying these measures more widely across affected regions as it worked towards sustained recovery. While this indicates progress, it is not an all-clear from the company. Microsoft directed users to incident MO1550821 in the Microsoft 365 admin centre or its public status page. MO1550821 is one of the official incident tracking IDs assigned by Microsoft for the ongoing service degradation and outage. Why this happened Microsoft’s explanation emerged in stages. It first said a number of servers had unexpectedly become unavailable, putting extra load on the remaining infrastructure. The higher load, it stressed, was not caused by a sudden rise in customer traffic. Later, Microsoft said a process on affected servers responsible for establishing secure connections had become overwhelmed, causing servers to stop handling requests. It also began rolling back a recent network configuration change that it said correlated with the start of the incident. Earlier updates had said no recent change had been linked to the problem. Recovery was complicated with servers becoming unresponsive again when traffic returned. Microsoft tried restarting and rebuilding affected servers, restricting certain types of traffic, moving connections to alternative routes, and improving capacity. The challenge was not to simply bring servers back online, but also to keep them stable as traffic returned. Microsoft said it had no indication that malicious activity had caused the outage. The precise root cause remained unconfirmed, and the company said it would publish a post-incident report. Why so many services were affected Exchange Online is Microsoft’s cloud-hosted email and calendar service. Other Microsoft 365 features use it for mail, contacts, calendar entries, and meeting information. When Exchange Online became unreliable, features that depended on it were disrupted even where the wider app was still running. Users struggled to access mailboxes through Outlook on the web, desktop, and other connection methods, while email delivery was delayed. Copilot for Microsoft 365 failed to load or respond, or struggled to retrieve work data. In Teams, calendars and meeting scheduling were affected, but core calls, meetings, group chats and channel messages remained available. SharePoint files stored directly in the service remained accessible, although some connected features were affected. Microsoft also listed problems with specific functions in Planner, To Do, Defender, Purview, Power BI, Intune, and Microsoft Graph, a set of programming interfaces that lets software exchange data with Microsoft services. The impact beyond Microsoft, including India Reuters reported that OpenAI and Anthropic had problems with their Microsoft 365 connectors, which link the AI services to users’ work information. In India, IndiGo and Akasa Air said their email-support services were affected. There was no indication that flights were disrupted. The effects reached public institutions too. Reuters reported that Australian government bodies, including the Australian Security Intelligence Organisation and the Australian Federal Police; Japan’s government IT infrastructure; and users in the US House of Representatives were affected. Taiwanese financial firms also reported disruption to fund-management work. Social media saw a meme fest. On X, users combined frustration with humour, as they joked about closing laptops, skipping meetings, and taking Friday off. Others linked another Microsoft outage to the long wait for Grand Theft Auto VI. Different incidents, similar risks This is not Microsoft’s first major 365 disruption this year. Another outage began on August 31 and was resolved on September 3. Microsoft attributed it to a problem with a core authentication configuration used by several services, affecting Exchange Online and parts of Copilot, OneDrive and SharePoint. However, that was a different, identified problem. A cloud-related outage similar to the current disruption happened in October at Google, when an electrical failure at a data-centre facility disrupted services in Google Cloud’s europe-west2 region, a cloud computing data centre location in London. Google’s incident report said the main impact lasted about six hours and eight minutes. The cause was different, but the incident showed how a local infrastructure failure can affect customers across a cloud region. The CrowdStrike incident in July 2024 was different again. A faulty software update from the cybersecurity firm caused Windows machines worldwide to crash. Microsoft estimated that 8.5 million devices were affected. The mechanism was different, but the disruption underlined the risks when a small number of providers underpin critical systems. What happens next The test now is whether the traffic-optimisation changes keep servers stable as Microsoft increases traffic and capacity across regions. The company has promised a post-incident report covering the root cause and measures to prevent a repeat. Customers need to know not only what failed but also why safeguards did not contain the disruption sooner. The ongoing incident shows how dependent workplaces have become on shared cloud infrastructure. For organisations, the lesson is to plan for a key provider failing. They must keep emergency contact channels outside the main email system, make critical information available by other means, and maintain alternative ways to reach staff and customers. Cloud services make apps work together, but these connections can also create shared points of failure. Products may look separate but they rely on the same underlying systems. Edited by Swetha Kannan
document fetched 2026-11-14 03:05