Cloudflare global outage

18.11.2025 was the day we all got a quick lesson in humility. Familiar services stopped responding, and in more than one office the classic question was asked, “Is it at our place? Or everywhere?”.

The answer was simple – EVERYWHERE.

If anyone was wondering what kind of power this one supplier has, they got a live demonstration.

Cloudflare in a nutshell, what is it?

Cloudflare is like a huge Internet switchbox.

• supports DNS, something like a network phone book,
• CDNs, or servers that deliver content from the nearest possible location,
• WAF, or shield against attacks.

When it works, no one thinks about it. When it goes down, you suddenly find that half the Internet is at a standstill. The domino effect is quick, since either DNS or CDN doesn’t work, the user won’t get to the application, even if the application itself stands steady as a rock.

What has stopped working? Large services such as ChatGPT and X, for example, stopped responding. A bunch of SaaS services hung up. We felt it too – eAuditor cloud® stopped responding due to an interrupted connection to Cloudflare proxies.

It is important to emphasize that this situation did not have any impact on data security. There was no attack, leakage or exploitation of any vulnerability. The problem was solely due to the fact that network traffic could not be properly delivered to the target.

Doesn’t that mean the system is a hole?

No, it means that the Internet works on a network of dependencies. When a provider of this scale has an outage, it’s as if someone cut the power in a large part of the city. You can have your own refrigerator, lights, oven, but if the transformer goes down, you’re sitting in the dark. That doesn’t say anything about your refrigerator, it says everything about the transformer, and that’s how the Internet works.

Systems are connected, optimized, balanced, routed through a network of services, and Cloudflare handles a giant chunk of that traffic.

eAuditor cloud® and what was going on in the background

Here the good news: the system architecture worked as it should. The agent works locally, regardless of what happens with Internet access. It collects data, monitors devices and performs its tasks according to the administrator’s configuration. When communication returned, everything synchronized automatically. The logs showed exactly what we expected, interruption of traffic at the proxy server level, zero risk of data loss.

We could have forced a Cloudflare workaround and manually restored access, but for the sake of security guarantees, we chose not to do so. Sometimes the best solution is simple patience.

What if the failure had lasted longer?

We are also prepared for such scenarios. When the specified time of unavailability would be exceeded, we implement emergency procedures to maintain direct access. However, as long as the failure was on the part of the global provider and falls within the specified time, there is no risk to data or agent operations.

It is worth remembering the basics at this point:

• backups,
• secure data centers,
• high standard of certification,
• multi-component authentication.

There’s a reason our data is stored at OVH in Ozarow, where ISO standards and SOC 1, 2 and 3 compliance are in place.

When it comes to user access, we recommend MFA. It’s available at your fingertips in the eAuditor cloud®.

100% fault tolerance does not exist

100% fault tolerance doesn’t exist, but sound architecture makes a huge difference.

The myth of absolute uptime can be put between fairy tales. But scalability, dispersion, automatic backups, switches and proven procedures make the cloud simply safe.

The 99.9% uptime we offer is a great result. But even with such a high SLA, you have to take into account global events like the one with Cloudflare. The IT world is a web of interconnectedness, when one block goes down, everyone feels it. It’s important that the system can get up, breathe, and get back to work as quickly as possible.