The Xbox technology director, Scott Van Vliet, classified an outage lasting about 20 hours in Xbox services as 'unacceptable.' The incident occurred on Monday, July 27th, and was caused by an external licensing service that impacted several platform functionalities.
Impact of the Instability
Due to this failure, many players faced difficulties logging into their accounts, accessing their game libraries, and even playing titles already installed on consoles. Van Vliet detailed that the persistence of the problem for almost a full day was the factor leading to its classification as an 'unacceptable situation,' given that short instabilities are considered tolerable.
Cause and Initial Response
According to the executive, the origin of the problem lay in an external licensing service, essential to the Xbox ecosystem, which also harmed company partners dependent on that same service. The issue was detected early on Monday through an automated monitoring system. The Xbox team responded by redirecting traffic to the parts of the system that were still operational while investigating the cause.
Recovery Strategy and Improvements
This mitigation attempt resulted in the irregular return of the service in different regions, which explains why users perceived the restoration at different times. Regarding future actions, Van Vliet stated that comprehensive post-incident analyses are being conducted. He emphasized that his focus is not just on the root cause, but on understanding why a failure in one service generated such instability and why recovery took so long.
The promised improvements include strengthening critical dependencies for login and game startup, enhancing methods for detecting and controlling this type of failure, and increasing transparency with users during issues. When questioned about the licensing system, Van Vliet clarified that it is a Microsoft service itself, shared across various company products. In response to a user comment on X, he reinforced that ideal systems must have resilient solutions for dependency failures, concluding that 'this is where we will improve.'
