What’s up with the Microsoft outage?! A tech expert explains what you need to know

Our dependence on a small number of tech companies strikes again.

Microsoft has suffered a major global outage, bringing many of its widely used tools, including email and cloud storage, to a standstill.

According to outage tracking site DownDetector, some 130,000 people reported issues accessing Microsoft Outlook as of early Thursday afternoon US Eastern time (early Friday morning in Sydney). In a statement on X, the tech giant said it was “aggressively pursuing all pathways to reduce impact” and “restarting affected servers to return them to service”.

In an article for The Conversation last month, I explained how the whole internet is unlikely to go down but several services can become unavailable together when they rely on shared systems.

This outage shows precisely what happens when that shared infrastructure fails – and why we must prepare for digital outages in a similar way we do for natural disasters.

Beyond missed emails or workplace meetings

The list of services Microsoft says are “potentially impacted” by the outage is long. It includes services used for email, document sharing, meetings and managing tasks, such as Outlook, Microsoft Teams, SharePoint Online, Microsoft Copilot 365, Microsoft Planner and Microsoft Exchange Online. It also includes Microsoft Defender, a cloud-based cybersecurity service.

With so many digital services affected, the real-world impacts of the outage have been widespread, and go far beyond missed emails or workplace meetings.

For example, family violence workers reportedly can’t access databases which contain client details. Schools can’t provide online lessons for remote students. Psychologists can’t provide telehealth consultations. Businesses are unable to process orders. Job hunters are unable to make it to planned interviews. Some government service centres have been forced to closed.

The disruption also reached services outside Microsoft. OpenAI and Anthropic reported problems with the tools that connect their artificial intelligence (AI) services to Microsoft 365.

×

Get smarter every day...

Stay ahead with ZME Science and subscribe.

Daily Newsletter
The science you need to know, every weekday.

Weekly Newsletter
A week in science, all in one place. Sends every Sunday.
No spam, ever. Unsubscribe anytime. Review our Privacy Policy.

Thank you! One more thing...

Please check your inbox and confirm your subscription.

This shows how a problem in one part of the digital world can affect tools that appear separate, and even services provided by other companies.

The outage comes after other disruptions to Microsoft services earlier this year. For example, in January, Outlook, Teams and other Microsoft 365 services were down for hours. Then in August, Microsoft Outlook and Exchange Online were disrupted by a code update which took several days to fix.

RelatedPosts

Microsoft Turns Ordinary Glass into a Permanent Hard Drive. One Tiny Square Can Store 2 Million Books for 10,000 Years
Inside the Wild World of Competitive Excel Where Office Pros Become Esports Stars
Microsoft scans photos to guess what your feelings are
Astronauts test new Holo-Lens Sidekick aboard the ISS

A shifting account of what went wrong

As of early Friday afternoon Sydney time, Microsoft has not yet named a root cause of the outage, and its account of the failure has shifted since it was first reported.

Initially the company said elevated load across multiple regions was reported first. It then clarified the load was not driven by an increase in traffic, but a large number of servers becoming unexpectedly unavailable.

Attention then moved to the automated “health checks” Microsoft uses to judge whether a server is fit to carry traffic, and to the component that processes them.

The company described a core process on the affected servers as becoming “overwhelmed”. Software deployments and configuration changes were paused while recent changes were reviewed, although none was found to correlate with the start of the disruption.

The evidence does not suggest a cyberattack.

A distributed denial-of-service attack works by flooding a service with requests from outside until legitimate ones can no longer be answered. Microsoft’s own statement that overall traffic had not increased is difficult to reconcile with that pattern. No evidence of malicious activity has been disclosed, and no group has claimed responsibility.

Indeed, Microsoft said: “There’s no indication this is a cyber attack.”

The more mundane explanation fits better. Health checks exist to protect a service by withdrawing servers that appear unwell. When the machinery making that judgement is itself at fault, healthy capacity can be removed, and the servers that remain are left to absorb work they were never sized for.

Outages of this scale are more often self-inflicted in this way than they are the work of attackers, as the August disruption to Microsoft email by a code update illustrates.

Until a post-incident review is published, however, that remains an inference drawn from Microsoft’s own updates rather than a settled account.

Microsoft has not yet been able to fix the outage entirely. It says that some of the affected servers are working again, but they “can become unresponsive again once they begin receiving traffic”.

Planning for outages like we do for bushfires

This outage reinforces the need for all of us to better prepare for digital outages. For individuals, this can involve storing essential information, such as contacts and medical prescriptions, offline in case of an outage.

Organisations that rely on one digital provider such as Microsoft or Google for tasks must also prepare back up plans. Crucially, they must also test these back plans to make sure they work in practice and don’t simply exist on paper – similar to how they conduct fire drills and how communities practice bushfire evacuations.

As more aspects of our lives become connected – from cars and homes to workplaces, transport and healthcare – and AI becomes more widely used, the consequences of digital disruption may grow.

Building resilience must be part of how we plan for that future.

Mahmoud Elkhodr, Senior Lecturer in Information and Communication Technologies, CQUniversity Australia

This article is republished from The Conversation under a Creative Commons license. Read the original article.