AWS Outage

As the massive outage of AWS on 20th of October, 2025 enters its 12th hour there are a couple of thoughts coming to mind.

First is that AWS fucked up, badly. No, not because it's a 12 hour outage, AWS as a provider messed up badly because they let cascading failures happen: an initial DNS outage resulted in a trigger of a failure which successively collapsed other systems that weren't able to recover even once the initial DNS failure was fixed.

This is a systemic design failure in AWS, people who build IT infra account for this when designing infrastructure and yes, not everything can be accounted for but issues like this do not reach twelve hours! AWS did not design their systems with resiliency in mind. This is a basic system design issue.

Second, there are a suspiciously large amount of european companies having issues right now, even though the failure should only impact us-east-1 as an AWS region. Data privacy and GDPR compliance much?

Third, some people who shall remain nameless were dragging Signal for being impacted by the AWS issues. The AWS outage is a valuable hilight of services that we rely on but take for granted, like Signal. A miniscule amount of the millions of Signal users donate or support the Signal Technology Foundation, yet some expected levels of uptime that even mission-critical services that people directly pay for do not always manage.

It takes tens of millions of dollars per year to offer Signal for free (the last tax return data from 2023 shows 11 million USD for hosting expenses), without selling any user data or violating your privacy in any way. Please consider donating to or supporting Signal.