On a Thursday afternoon, people around the world found that the quiet channels of work and correspondence had suddenly gone silent, as Microsoft's email and productivity services faltered. The company, which holds much of the world's office life in its hands, acknowledged the failure only after users had already raised the alarm, and attributed it not to a rush of demand but to servers that had quietly fallen out of service. Governments and businesses, from Michigan's state agencies onward, were left waiting on replies that could not be sent, a reminder of how thoroughly everyday communication
Microsoft Outlook Outage Spreads Worldwide as Company Works to Restore Service
Servers unexpectedly unavailable, not a surge in traffic
Mark
Start with the basics. What actually broke?
Mimi
Microsoft says Exchange Online connections failed through multiple methods, so people couldn't send or receive Outlook mail. Teams and Copilot showed problems on Downdetector too, and the company later said Microsoft 365 broadly was affected.
Mark
Why does the cause matter?
Mimi
Because the first assumption was load. Microsoft's first fix was spreading traffic across its wider infrastructure. Then it said the load wasn't from more traffic but from servers unexpectedly unavailable.
Luke
Worth noting that's Microsoft's own account. We don't yet know why those servers went down.
Mark
And the fix didn't work at first?
Mimi
Microsoft itself said its efforts hadn't provided the anticipated relief. It then re-isolated part of the infrastructure and restarted components it suspected were responsible.
Luke
Notice the word "suspected." The company is still diagnosing, not declaring a cause.
Mark
Who felt it beyond ordinary users?
Mimi
Governments and companies. Michigan's technology department said email and other Microsoft apps were having issues across the state and replies might be delayed.
Luke
Michigan is the one example we have. The reporting says others were affected but doesn't name them, and there's no count of users.
Mimi
Right, and the Downdetector recovery signal at 2:10 PM was only for some users. Many were still struggling hours later.
Mark
So what's the open question?
Luke
Whether restarted servers stay up once traffic returns. Microsoft warned they might not.
Le Pouls
- Outlook mail stopped sending and arriving for users across the world, and Microsoft first labeled it only a service degradation.
- Teams and Copilot faltered alongside email, spreading the disruption across the whole Microsoft 365 suite.
- Engineers rerouted connection traffic worldwide and re-isolated infrastructure, but early fixes failed to deliver the relief they promised.
- Michigan's state technology department warned of delayed email responses, while recovery remained partial more than seven hours in.
- Restarted servers were becoming unresponsive under returning traffic, so temporary limits on certain traffic types remain in place and stability is still uncertain.
On a Thursday afternoon, people around the world found that the quiet channels of work and correspondence had suddenly gone silent, as Microsoft's email and productivity services faltered. The company, which holds much of the world's office life in its hands, acknowledged the failure only after users had already raised the alarm, and attributed it not to a rush of demand but to servers that had quietly fallen out of service. Governments and businesses, from Michigan's state agencies onward, were left waiting on replies that could not be sent, a reminder of how thoroughly everyday communication now depends on machines most people never see. By evening the work of recovery was still unfinished, with the fix itself raising new questions about whether restored systems would stay standing.
On Thursday afternoon, Microsoft's status page confirmed what users had been reporting for some time: people trying to reach their mailboxes in Exchange Online could not connect, no matter how they tried. Outlook messages were failing to send or arrive, and Downdetector showed the trouble spreading to Teams and Copilot as well. Microsoft first called the problem a service degradation and said its engineers were gathering networking logs to find the source of the errors.
The company's first response was to spread connection traffic across its worldwide infrastructure to balance the load. By mid-afternoon some users saw service returning, but for many the problems dragged on for hours. Microsoft later said the strain did not come from a surge in demand. Instead, a number of servers had become unexpectedly unavailable.
The disruption reached far past private inboxes. Michigan's department of technology, management and budget reported widespread problems with email and other Microsoft applications for the state and warned that replies could be delayed. Many organizations turned to social media to reassure the people who depended on them.
Late in the evening Microsoft admitted that its efforts had not brought the expected relief. It re-isolated part of the infrastructure and restarted components it suspected were at fault. More than seven hours after the trouble began, it warned of a new problem: restored servers might stop responding once traffic moved back through them. Engineers were studying diagnostic data to understand the cause and had temporarily restricted certain traffic types in some locations.
At the time of reporting, restoration was still incomplete. The open question is whether the restarted servers can hold steady as traffic returns, and whether the traffic limits can be lifted without bringing the instability back.
On Thursday afternoon, Microsoft's service status page posted a notice that people trying to reach their mailboxes in Exchange Online might fail to connect, whichever method they used. It was the company's first public acknowledgment of a problem that users around the world had already been reporting: Outlook messages that would not send or arrive.
Microsoft classified the Outlook trouble as a "service degradation." The status message said engineers had gathered networking logs to investigate the source of the errors and decide their next steps. Downdetector, the site many people now use to track internet disruptions, showed users across the globe also having trouble with Microsoft Teams and Copilot.
The company then described its first fix in a social media post. It had begun redistributing connection traffic across its broader worldwide infrastructure to help balance the load. By 2:10 PM, Downdetector data suggested service was starting to recover for some users. For many others, the problems continued for several hours.
Microsoft later confirmed on its service health page that the fault was affecting the wider Microsoft 365 productivity suite, not just email. It also reported progress on the cause. The elevated load, it said, was not the result of a surge in traffic but of a number of servers being unexpectedly unavailable.
The effects reached well beyond individual inboxes. Governments and companies around the world were affected, and many turned to social media to reassure customers and the public. Michigan's department of technology, management and budget said a widescale Microsoft 365 outage across North America was causing technical issues with email and other Microsoft applications for the State of Michigan, and warned that responses to emails might be delayed.
Later in the evening, Microsoft said it was still trying to open all available pathways to restore service, but acknowledged that its efforts had not provided the anticipated relief. It changed course, re-isolating a portion of the infrastructure and restarting certain components it suspected of causing the problem.
More than seven hours after the trouble first surfaced, the company warned of a new complication: servers that had been returned to service might become unresponsive once traffic began moving through them. In a post on X, Microsoft said it was analyzing diagnostic data to find what was triggering that behavior. In parallel, it was temporarily limiting which types of traffic could reach the affected infrastructure in some locations.
Restoration was still incomplete at the time of reporting. The thing to watch is whether the restarted servers hold up as traffic returns, and whether the temporary limits on traffic can be lifted without bringing the instability back.
Citations marquantes
We determined the elevated load isn't driven by an increase in traffic, but by a number of servers being unexpectedly unavailable.— Microsoft service health update
Responses to emails may be delayed during this time.— Michigan Department of Technology, Management and Budget
Envie de l'histoire complète ?
Lire l'original sur The National ↗