A technical issue in Microsoft Corp.’s Azure cloud platform today disrupted several online services operated by the tech giant and its customers. 

Starbucks Corp., Costco Wholescale Corp. and Capital One Financial Corp. were among the affected Microsoft customers. The tech giant, for its part, saw several of its consumer services go offline. The outage also affected Microsoft’s investor relations page, which briefly became inaccessible before the release of its latest earnings report.

The outage was caused by an error in Azure Front Door, Microsoft’s content delivery network. The CDN enables website operators to store copies of their content in dozens of server clusters around the world. When a user visits a webpage, the CDN downloads the webpage’s contents from the nearest server cluster to speed up loading times. 

Azure Front Door includes several features designed to minimize the risk of outages. 

When one of the server clusters, or edge locations, the CDN uses to store customer content goes offline, user traffic is automatically rerouted to a different edge location. There are more than 110 such clusters that are organized into two sets: an inner ring and an outer ring. The outer ring is configured to take over for the inner ring in the event of technical issues.

According to Microsoft, the outage began around noon EDT. It took down not only Azure Front Door but also more than a dozen other Azure services that depend on the CDN to work. The list of affected services included several cybersecurity products that customers rely on to protect their cloud environments.

Shortly after the outage began, Microsoft determined that it was caused by an “inadvertent configuration change.” The company’s engineers subsequently launched an effort to restore the last configuration known not to contain any errors. The most recent update on Microsoft’s status page states that it has completed the rollback and is now “observing recovery.”

The company is currently rerouting requests from malfunctioning Azure Front Door nodes to servers that have been successfully recovered. However, some requests are still finding their way to the malfunctioning nodes, which is causing “intermittent failures or reduced availability for a subset of customers.” Microsoft added that the service is “operating above 98% availability.”

The company has temporarily blocked customers from changing the configuration of their Azure Front Door deployments to prevent potential technical issues. Earlier today, some customers also had difficulties making configuration changes in Azure’s management console. Microsoft fixed the issue by moving the console off Azure Front Door.

Today’s service disruption comes a few days after an hours-long outage in Amazon Web Services Inc.’s busy US-EAST-1 data center cluster. The malfunction impacted ChatGPT, Disney+, Snapchat and numerous other popular services. In a blog post published after the outage, AWS explained that it was caused by the DNS system of its Amazon DynamoDB managed database service.

A DNS issue was also to blame for Azure’s previous large-scale outage in 2021. The error took several Azure and Microsoft 365 services offline for several hours.

Photo: efes/Pixabay