Skip to Content

Cloudflare and the global outage of November 18, 2025...

...and how we mitigated it at QTNOVA.

What happened?

On November 18, 2025, Cloudflare — one of the leading providers of internet services (DNS, CDN, security) — suffered a major global outage. In practice, a huge number of websites and online services that rely on Cloudflare became unreachable or displayed error messages for several hours. Cloudflare itself confirmed that it had experienced a significant problem on its global network, which caused widespread disruptions on major platforms such as X (formerly Twitter), ChatGPT and even corporate websites like McDonald's. In other words, a sizable portion of the Internet slowed down or stopped working because of this incident.

Cloudflare began investigating the problem at around 11:48 UTC (Greenwich time), when it detected an “internal service degradation” affecting various customers (source: cloudflarestatus.com). During the investigation, the company noticed an unusual spike in traffic to one of its services at around 11:20 UTC, which overloaded the network and caused errors for users' traffic. At first, it was unclear whether this spike was due to a cyberattack or to a malfunction. Later, the Cloudflare team identified the cause of the problem at around 13:09 UTC and began working on a solution. Fortunately, it was not a hacker attack but an internal bug: a latent error in one of their services (related to the system that mitigates malicious bots) which, after a routine configuration update, started crashing that component. This malfunction then spread, causing widespread degradation across the Cloudflare network. Put simply, a seemingly harmless change triggered a hidden bug that jammed the Cloudflare “machine” from the inside.

At around 14:42 UTC, Cloudflare announced that it had deployed a fix and considered the incident essentially resolved, starting a monitoring phase to make sure all services returned to normal. From then on, over the course of the afternoon, the situation gradually improved: errors and latency (that is, delays in responses) steadily decreased until they were back to normal levels. Indeed, at around 17:14 UTC Cloudflare reported that the remaining errors had subsided and its systems were operating normally. Overall, the main outage lasted a few hours — roughly from midday until late afternoon (UTC) — with knock-on effects on a huge number of online services around the world.

Chart of Cloudflare outage reports over the past 24 hours (from Downdetector). Note the peak of more than 11,000 reports around the time of the outage, a sign of a very widespread impact.

This episode made it clear just how widespread the dependence on Cloudflare's network is: the repercussions were felt everywhere, from popular social networks to corporate websites and everyday apps. Cloudflare itself acknowledged the seriousness of what happened, apologizing for having “let down” its customers and promising that measures will be taken to prevent a similar episode from happening again. In other words, even a cutting-edge technology company can run into an incident capable of shaking half the Internet, and the priority now is to learn from the mistake and strengthen the infrastructure.

Okay, but... what does Cloudflare do? How does DNS work?

A conceptual illustration of DNS, often called the “phone book” of the Internet, which translates a domain name (e.g., www.example.com) into the corresponding numerical address (IP) of the server.

To understand the context, it's worth briefly explaining what DNS (Domain Name System) is and how it works, in non-technical terms. DNS is the system that translates domain names (like www.website.com) into the IP addresses of the servers where those websites are hosted. People often say that “DNS is like the phone book of the Internet” — and that's exactly what it is. When you type a website address into your browser, for example www.example.com, your computer doesn't know on its own where to go to get that page. It needs the server's numerical address (a bit like the phone number that goes with a name). This is where DNS comes in: it receives your browser's request, looks up the right IP associated with example.com in its “directories” and then returns that information to your device. At that point, the browser can connect to the IP address it found and load the website you wanted.

In practice, without DNS we would have to remember long strings of numbers for every website (for example 192.0.2.1 instead of sitename.com), which would be impossible in everyday Internet use. DNS servers therefore play a fundamental role: they connect the “human” world of readable names (domains) with the “machine” world of IP addresses, letting us browse the web easily. There are various types of DNS servers that work together in this process — from recursive resolvers to root servers and authoritative nameservers — but that level of detail is beyond our scope here. All the reader needs to know is that when you visit a website, several queries are exchanged between DNS servers behind the scenes to find the right path, usually all within a few milliseconds and without you noticing.

Cloudflare is a major player in this system. Many websites use Cloudflare as their authoritative DNS provider, which means that the records saying “domain X corresponds to address Y” are hosted on Cloudflare's servers. Cloudflare also offers other related services, such as content delivery through its global network (CDN) and protection against DDoS attacks. Put simply, Cloudflare routes traffic to its customers' websites using its own DNS servers and network, helping both to distribute the load (to avoid overloading the origin servers) and to protect websites from malicious actors. This is great for improving performance and security... as long as everything works. But if the Cloudflare network has problems (as happened in this case), DNS requests may go unanswered or traffic may not be routed correctly. As a result, users' browsers cannot reach the websites, even though those sites' origin servers would technically be working. It's a bit like having the right phone number while the telephone exchange can't put the call through: the result is that the call (or, in our case, the connection to the website) doesn't go through.

Could it have been avoided?

Faced with a blackout like this, it's natural to ask: could it have been avoided? Unfortunately, events of this scale are often hard to predict or prevent 100%. Cloudflare is known for the reliability of its infrastructure, and yet a single hidden bug caused a massive domino effect. Even the best technology companies can run into unexpected incidents — perfection does not exist in complex systems. From the point of view of users and of Cloudflare's business customers, there was not much they could do to prevent the outage, other than having already prepared some kind of alternative plan in advance (which we'll discuss shortly). Anyone who relies on a single provider for a critical service is inevitably exposed to that provider's single point of failure.

It's worth noting that this Cloudflare episode comes at a time when several major providers have suffered outages in close succession. In the last few months of 2025, there were also major problems at Amazon Web Services and Microsoft Azure. Experts point out that these incidents highlight the fragility of online infrastructure, since so many companies rely on just a few providers to run their services. In other words, the Internet is highly centralized: if one of the main nodes goes down, the knock-on effects can be enormous. One analogy is the power grid: if a single large power plant supplies an entire region, a failure there will leave millions of people in the dark. In the same way, a problem at Cloudflare (or Google, Amazon, etc.) can cause widespread disruption for millions of users.

In retrospect, Cloudflare admitted that “the time to resolution was unacceptable” and committed to making sure it does not happen again. However, while the problem was unfolding, avoiding it entirely was virtually impossible for end users. Those hosting their services behind Cloudflare could only wait for the provider to fix the error. There was no magic button to press to immediately get around the failure — unless a redundant infrastructure with other providers had already been set up in advance (which is uncommon, especially for smaller organizations). In short, looking back, we can say that “it's always easy to be wise after the fact.” At the time, the vast majority of affected websites could do nothing but stay offline until Cloudflare was back up and running.

Could it have been mitigated, and services restored faster?

Mitigating an outage means reducing its impact or duration. Avoiding the incident altogether was not realistic, but in some cases it was possible to limit its effects and recover faster. The key here is preparation, along with system architecture: if a company has planned emergency measures, it can react promptly when a provider like Cloudflare has problems.

In our case, at QTNOVA we were able to mitigate the problem quickly thanks to the flexibility of our infrastructure. Because we manage our entire system configuration as code (infrastructure as code), we had the option to temporarily move the DNS for our services to another provider. Put simply: when we noticed the Cloudflare blackout, we changed a few lines of configuration to point the NS records to an alternative provider (which we had ready), thereby updating our domains' references. This allowed us to make our services reachable again faster (within the time it takes the records to propagate), without having to passively wait for Cloudflare to fully resolve the issue. Like us, other services such as Canva were back online within a short time. Once Cloudflare had resolved its incident and everything was stable again, we moved the DNS back to Cloudflare (again through a versioned configuration change). Thanks to this maneuver, the disruption for our users was minimal. Of course, not every organization has this option: it takes planning and a certain level of technical expertise to prepare a similar plan B. But it is a concrete example of how resilience can make the difference.

In general, relying on a single provider for everything may be convenient in normal times, but it becomes an Achilles' heel in a crisis. A more resilient approach is to adopt redundancy, that is, having backup systems ready to take over if something goes wrong. 

Implementing these strategies can be complex and expensive, but it significantly increases the resilience of online services. In essence, it means not having a single point of failure. Take DNS specifically, a fundamental building block: if your authoritative DNS goes down, your website becomes unreachable even if your servers are working perfectly. That's why some large websites (or organizations that are particularly focused on business continuity) set up secondary DNS systems or emergency fallbacks. However, for many small and medium-sized businesses, this kind of redundancy is not always practical, and they tend to rely on a single provider like Cloudflare, trusting in its reliability.

In conclusion, today's Cloudflare incident reminds us of two important lessons. First: even the giants of the web can fall, and when they do, the impact is felt everywhere. Second: it's worth considering backup solutions and contingency plans, proportionate to your own business, to mitigate the effects of these blackouts. In our own small way, being able to react quickly by switching DNS providers allowed us to avoid the worst. It isn't always “DNS's fault,” as an old IT joke would have it, but when DNS really is to blame (or to thank), making sure this link in the chain is robust can save the day.

Sources:

  • Cloudflare Status – Cloudflare Global Network experiencing issues (official timeline of the incident on November 18, 2025) cloudflarestatus.com
  • Cloudflare communications and real-time updates tomshardware.com
  • CBS News article “Reports of outages surge at X and other apps as Cloudflare says it's having widespread problems” (Nov 18, 2025) cbsnews.com
  • Cloudflare explainer: “What is DNS? | How DNS works”cloudflare.com
  • Post-incident analysis (Medium) of resilience strategies after a DNS outage


Want to know more? Leave us your email below and we'll get back to you as soon as possible.

Send 

Thank you for your message

Our team will get back to you as soon as possible.
In the meantime, we invite you to visit our website.

Guide to choosing business management software for your company