The company said a recent change impacted some of its services and it has deployed a fix.
A vast number of Microsoft users were unable to connect to Outlook email and the Teams collaboration tools for hours Monday after the software giant fumbled a change to its internal systems.
Higher than normal U.S. outage reports began as the workday kicked off on the East Coast and peaked hours later in the early afternoon before tapering off, according to the tracking site Downdetector. Worst hit was the web version of Outlook, with the phone and desktop apps faring better, the site said.
After posting that it was looking into the issue, Microsoft eventually said it had identified a "recent change" that was behind the problems.
"We've started to deploy a fix which is currently progressing through the affected environment," the company said in a post on X. "While this progresses, we're beginning manual restarts on a subset of machines that are in an unhealthy state."
More than four hours later, just before 2 p.m. Eastern time, the company said its efforts were facing delays.
The changes involved Microsoft 365, a cloud-based package of applications that includes Outlook and Teams as well as Word and other tools for common business processes.
The outage was another jarring reminder that the move to cloud services increases dependence on outside vendors that periodically fail to function.
Related:How IT Leaders Can Assess and Evaluate Risk to Mitigate IT Outages
While security experts have long recommended that companies have a backup cloud provider, that is expensive and complicated, and it would not automatically provide access to the same email accounts handled by the main cloud vendor. Most companies rely on a sole provider for each service.
Because Microsoft is a dominant provider of business software, the effect of Monday's outage was widespread, and many users complained on social media. A minority described it as a welcome early start to the Thanksgiving holiday, with meetings and office correspondence shut down.
The ripple effects also inconvenienced customers of some businesses frozen by the outage, but the situation was far from as dire and expensive as that following a disastrous July mistake by tech security company CrowdStrike.
CrowdStrike pushed out a faulty update to its detection software that runs on Microsoft's Windows. Because that program integrates with the core-level kernel of Windows and the mistake was a serious one, it crashed the computers at large customers. For hours, those machines could only be restored one by one, typically with someone talking a user through multiple steps.
The multibillion-dollar CrowdStrike outage hit some airlines, hospitals and banks especially badly, and thousands of flights were canceled. Delta Air Lines sued CrowdStrike last month, claiming $500 million in damages, while CrowdStrike says the airline was negligent in its contingency planning.
Related:10 Microsoft Teams Alternatives, From the Free to the Pricey
Microsoft had previously pledged radical improvements in its internal security after devastating breaches of top government officials and others. Following the CrowdStrike debacle, it released a tool to allow customers to roll back the state of software on their machines to a previous time. That should make restoring crashed systems much faster.
A Microsoft spokesperson declined to comment beyond the posts on X.