Cloudflare Outage 2025 EXPLAINED

Key Takeaways

What do you think? Is this just another example of why we don't want to centralize our infrastructure in 2025 and going into 2026? We've had multiple examples of this in 2025. We had AWS go down affect many, many websites, thousands of businesses, and now we've got Cloudflare going down affecting again the internet if you like, but basically many many businesses. You may have seen the following when you connected to one of your favorite websites. Internal server error error code 500. Basically an issue with CloudFlare. So this is from Radwware. On the 18th of November 2025, a widespread service disruption impacted CloudFlare's global network, rendering a significant portion of the internet inaccessible for approximately 6 hours. While initial traffic patterns mimicked those of a hypers scale distributed denial of service attack or DDoS attack, forensic analysis has confirmed the root cause was an internal logic failure in the bot management system. So life was good. But at approximately 1120 UTC 500 errors increased dramatically across the Cloudflare network. As you can see here, the issue was not caused directly or indirectly by a cyber attack or malicious activity of any kind. Instead, it was triggered by a change in one of our database systems permissions, which caused the database to output multiple entries into a feature file used by a bot management system. That feature file in turn doubled in size. The larger thanex expected feature file was then propagated to all the machines that make up our network. The software had a limit on the size of the feature file that was below its doubled size that caused the software to fail. Cloudflare wrongly suspected the symptoms. they were seeing were caused by a hypers scale DDoS attack. But then they correctly identified the core issue and were able to stop the propagation of the larger than expected feature file and replace it with an earlier version of the file. Core traffic was flowing largely normally by 14:30 and then they worked over the next few hours to bring the systems back online. Now they say we are very sorry for the impact to our customers and the internet in general. Now as Radwware say here is this a case for diversification relying heavily on a single vendor for critical path services such as DNS W zero trust etc concentrates risk this incident validates the need for architectural redundancy and break glass mechanisms that allow traffic to bypass a failed security layer during vendor specific outages. Now this video is sponsored by Radware but I think it's a valid point. Don't put all your eggs into a single basket. Have a look at diversification. Don't just use one vendor. Have multiple vendors. I think this is a basic premise in all IT systems. have redundant systems like this in case something goes wrong like in this example with Cloudflare or previously with AWS.