Yesterday, the internet faced a rare but major disruption. Cloudflare, the company that powers millions of websites worldwide, went down for more than two hours. Millions of users trying to access sites like news portals, e-commerce stores, and apps saw error pages instead of content. Now, Cloudflare has officially explained what went wrong. And the reason is both technical and surprising.
Cloudflare says the outage was not caused by a cyber attack. Instead, it started with a small change in the permissions of one of their database systems. The outage began at 11:20 UTC and was mostly resolved by 14:30 UTC, with all systems fully restored by 17:06. Services affected included core CDN and security functions, Cloudflare Turnstile (login service), Workers KV (key-value storage), Access authentication, and the Cloudflare Dashboard.
Cloudflare is one of the backbones of the modern internet. Its systems handle traffic for millions of websites, large and small. When Cloudflare goes down, it is not just a single site that fails—it’s a ripple effect across the web. Even some systems designed to be independent, like their status page, briefly failed, which confused engineers initially and made it look like a potential DDoS attack.
The outage started because a small database change created a bigger problem. The Bot Management system uses a file with “features” that help identify bots on websites. The file suddenly became too large, exceeding what the system could handle. This caused websites to show 5xx error pages.
The file is updated every five minutes, and sometimes it was fine and sometimes it was bad. This caused the network to fail and recover repeatedly, making it harder to diagnose.
Several services depended on the core system, so when it failed, multiple parts of Cloudflare went down. Even their key-value storage and login system were affected temporarily.
Cloudflare says it is taking several steps to prevent future outages. It will:
- Harden how configuration files are handled
- Add more global kill switches for features
- Prevent error reports from overwhelming system resources
- Review failure modes across all core modules
These steps should reduce the risk of similar outages, but they also remind everyone that the internet is not infallible.
Yesterday’s outage is a reminder of how fragile the internet can be. When a network as central as Cloudflare fails, the impact is far-reaching. Businesses that rely on websites and online services lost revenue, some potentially millions of dollars, during the downtime. Publishers, e-commerce platforms, and app-based services were also affected, leaving users frustrated and operations stalled.
While companies like Cloudflare can issue apologies, the financial and reputational damage cannot be undone. Outages like these show that even brief disruptions can ripple across the digital economy.
