The Outlook email service has begun to be restored after an outage that has lasted more than 20 hours, but some users may still have difficulty accessing their accounts or receiving messages, according to Microsoft.
In its most recent update, released late this Friday morning (13), the company reported that the queue of pending emails has been significantly reduced and is approaching normal levels.
The company, however, warned that some users may still face problems, as the full recovery has not yet been completed. Microsoft also said it had taken additional steps to restore the rest of the affected infrastructure.
But why did the outage take so long to resolve?
Professor James Ferryman, from the University of Reading, in the United Kingdom, says that the information released by Microsoft gives some clues.
According to him, the company reported that the servers responsible for establishing secure connections were becoming overloaded — until they simply stopped responding.
"The problem is that this type of failure ends up getting worse in a chain," he explains.
"One server stops working, its tasks are transferred to another, while a huge number of cell phones and computers continue trying to reconnect."
The problem, therefore, does not end when the initial failure is corrected, according to the expert.
The servers that continue to work need to absorb the tasks of the affected equipment and, at the same time, deal with new connection attempts.
"It is not possible to simply reactivate everything at once. It is necessary to restore services little by little", says Ferryman.
This means that Microsoft needs to recover the infrastructure gradually, controlling the volume of devices trying to connect to avoid further overload.
Microsoft shares fell 0.2% in early U.S. trading on Friday, suggesting investors aren't too worried about what the disruption could mean for the tech giant.
This seems remarkable considering the enormous level of disruption the failure caused around the world: banks, shopping, schools and medical services all reported problems. But stock price movements depend on earnings expectations, not public sentiment.
So for Microsoft investors, there are only two big questions today: will the company be held responsible for paying damages as a result of this problem? And will its customers cancel their subscriptions to the company's services en masse in response to the incident?
The stock's movement, or lack thereof, indicates that most investors believe the answer to both questions is "no", writes BBC Business Reporter Mitchell Labiak.
What happened and which services were affected
The interruption began on Thursday (12) and affected, in addition to Outlook, tools such as the Teams messaging application, the Microsoft 365 package (formerly Office) and the Copilot artificial intelligence assistant.
A Microsoft spokesperson told Reuters that the problems were focused on Exchange Online, the service that manages Microsoft 365 mailboxes.
Among the problems, users were unable to view certain files or details in their calendar applications.
Some users, however, were able to use Teams normally, both to send messages and to organize meetings and consult calendars.
Microsoft denied that the failure in its services was caused by a cyber attack and attributed the problem to "high load" on its infrastructure.
The term refers to situations in which services are dealing with "a much greater volume of stress" than usual.
In the case of Microsoft's failure, a large number of servers became "unexpectedly unavailable," according to the company, which overloaded the remaining servers.
In addition, AI services with connections to Microsoft products were also impacted.
OpenAI, responsible for ChatGPT, reported that it was investigating an issue with integrations to Microsoft services, in which new connections could be interrupted, causing the tools to not work as expected.
For its part, Anthropic said it has identified an issue with a top-tier provider that affects its connection to Microsoft 365.
How many people were affected?
Microsoft did not disclose the exact number of people affected, but its tools have around 430 million users, including companies around the world.
Tens of thousands of people in more than 70 countries have reported problems through the Downdetector platform — a number that is probably well below reality.
According to the internet failure monitoring website, 140,000 reports of problems with Outlook around 3 pm (Brasília time) in 74 countries on Thursday night (12/11).
The number fell to around 34,490 in the late afternoon, according to website data cited by Reuters. At 7:47 pm, the site also registered around 13 thousand reports about Microsoft 365 and 763 about Teams.
How is Microsoft trying to prevent emails from being lost?
One of the measures announced by the company was to extend the waiting period for the delivery of messages that were delayed during the interruption to 48 hours.
In practice, this period determines how long the system tries to deliver an email before considering it definitively undelivered and returning a notification to the sender.
With the expansion, Microsoft seeks to preserve messages sent during the failure, giving more time for them to be delivered when services are normalized.
The company states that it will continue to monitor the situation until complete recovery and that it will release a report identifying the causes of the incident and the additional protective measures adopted.
