Microsoft’s Azure cloud platform suffered a widespread global outage on 29 October 2025, affecting — among others — Microsoft 365, Xbox, gaming services, and a wide range of enterprise customers, forcing organisations and consumers to contend with interrupted services.

The disruption first registered as spikes in outage reports on Downdetector earlier in the day. Microsoft’s Azure status page later confirmed problems with Azure Portal access and with the Azure Front Door (AFD) service.

In its update, Microsoft said that “starting at approximately 16:00 UTC, we began experiencing Azure Front Door issues resulting in a loss of availability of some services. … Customers may experience issues accessing the Azure Portal. The company would go on to note that the incident was caused by “an inadvertent configuration change” to AFD.

Users reported difficulties accessing a range of services that depend on Azure.

On Downdetector, thousands of reports covered Microsoft 365, Minecraft, Xbox Live, Copilot and other Microsoft services.

The outage comes in the wake of a large Amazon Web Services (AWS) incident last week; Microsoft said the root cause here related to its domain name system (DNS) and an Azure Front Door misconfiguration.

Microsoft’s note on the Azure status page said users of the Azure cloud portal “may not be able to access Office 365, Minecraft or other services due to issues with its domain name system.”

DNS problems and the internet’s translation layer

Microsoft’s update outlines the mechanics of the issue: the domain name system, or DNS, is the service that translates internet addresses into machine-readable IP addresses that connect browsers/apps with websites and underlying web services.

The company warned that DNS errors disrupt this translation process, interrupting the connection, and noted that because so many sites and services run on Microsoft’s cloud, a DNS failure can have far-reaching consequences.

On its status page, Azure’s network infrastructure was showing as “critical” in multiple regions worldwide, meaning this was certainly not a localised issue. In fact, voting at Holyrood was temporarily suspended after technical issues linked to the Microsoft outage disrupted the parliament’s electronic voting system. The Presiding Officer informed MSPs of the issue during Wednesday evening’s session.

Microsoft said it believed the outage was a result of “an inadvertent configuration change”, and that it planned to remedy the situation by rolling the service back to a recent backup known to be functioning correctly.

The company provided no exact timescale for full recovery, though later updates suggested services had largely returned to normal.

Microsoft acknowledged that parts of its infrastructure were experiencing connectivity issues and said it had deployed a fix to address the configuration change. The firm described its mitigation work as rerouting traffic to healthier infrastructure while investigating and attempting to restore service health.

As of the latest updates:

• Microsoft said the AFD service is now operating above 98 % availability.

• Services such as Azure, Microsoft 365 and others “returned to pre-incident performance levels”, though “a small number of customers may still be seeing issues”.

• The company remains monitoring systems to ensure full recovery and validation of the rebuild/rollback.

What this means for organisations

For businesses and consumers affected, the incident underscores how critical cloud-infrastructure resilience is – and how a single mis-configuration in a globally distributed system can cascade widely. Organisations using Azure-based services should check their own dependencies, prepare for future incidents (including multi-cloud or regional fail-over strategies) and review how they will handle such events in future.

For now, affected businesses and consumers are left waiting for Microsoft’s remediation to fully take hold. The fact that Microsoft chose to revert to a recent “last known good” configuration illustrates the severity of the error it has diagnosed – and the lack of a firm timescale means disruption could have lingered for some users longer than expected