클라우드플레어 장애로 전 세계 주요 웹사이트가 다운

Cloudflare, the global infrastructure and cybersecurity leader, caused a bit of a panic this week, after its systems outage made a variety of the world’s most popular websites and apps go down. The incident has shed light on the fragility of the internet and how so many platforms rely on core providers for their essentials. Here’s what you need to know.

클라우드플레어가 하는 일과 많은 사이트가 의존하는 이유

Cloudflare has grown to become an essential part of several key platforms. Its web infrastructure and cybersecurity services enable companies to enhance their performance without sacrificing their security. As such, the company has seen considerable adoption. Cloudflare has become an essential part of the global internet. Today, about 20% of the web runs through its network, giving it an outsized role in ensuring websites load quickly and stay protected from attacks.

클라우드플레어 장애가 전 세계에 미치는 영향

There are many reasons why Cloudflare going down is a huge problem. As mentioned, 20% of all global web traffic relies on its network to ensure its data transmissions remain safe and reliable. Any disruption in service to Cloudflare operations has a ripple effect that can disrupt entire industries.

Remember, today’s workflows are heavily reliant on online services like ChatGPT to assist with writing, coding, research, and problem-solving. Consequently, any slowdown in communications can result in an inability to access vital components, including regulatory databases. It can also prevent real-time filings and delay financial transactions.

장애 동안 다운된 주요 웹사이트

When Cloudflare failed, many of the world’s most visited platforms experienced slowdowns, partial outages, or complete downtime. Reports from users and monitoring platforms confirmed disruptions across:

• X (formerly Twitter) – timelines failed to load, and API calls timed out
• ChatGPT – widespread “network error” and inability to start new conversations
• YouTube – loading failures and broken video playback
• Uber – app connectivity issues for riders and drivers
• Downdetector – the outage-tracking platform itself went offline
• FERC (Federal Energy Regulatory Commission) – regulatory website inaccessible
• New Jersey Transit – trip-planning tools and service updates are unavailable

These platforms rely heavily on Cloudflare’s CDN, DDoS protection, DNS resolution, or traffic optimization layers. When Cloudflare’s network became unstable, each service experienced cascading failures, ranging from slow API responses to complete service interruptions.

2025년 11월 18일 클라우드플레어 장애의 원인

Reviewing the timeline of events from the Cloudflare outage, it’s now possible to fully understand its causes and effects. On November 18th at around 6:20 AM ET, reports started to pour in that users were receiving a Cloudflare Error 500 message when attempting to access certain websites.

Initially, Cloudflare’s internal teams suspected a large-scale DDoS attack because of the spike in HTTP 5xx errors and the way failures fluctuated across the network. The company later confirmed there was no cyberattack; instead, the symptoms were caused by a malformed configuration file used by its Bot Management system.

클라우드플레어 오류 500

After further review, it was determined that the outage wasn’t due to a hacker attempting to breach their systems. The true reason for the outage revolved around the company’s Bot Management System. This AI-powered protocol helps companies determine whether visitors are bots or actual people.

클라우드플레어 오류

The Bot Management System accomplished this task by providing scores to every attempted access by a user. This system integrates several factors regarding the access request, such as time, frequency, and other vital data points. This information is then stored in a feature file.

Notably, this feature file is regenerated every five minutes and propagated across Cloudflare’s global network, enabling the Bot Management system to quickly adapt to changing bot behaviour. After a change to the permissions on the ClickHouse database that generates this file, the query began returning duplicate feature rows, causing the configuration file to exceed the size limit expected by Cloudflare’s core proxy software.

Instead of triggering a graceful safety mode, the oversized configuration file caused the proxy to throw errors and return HTTP 5xx responses for any traffic that depended on the Bot Management module. As a result, many major providers relying on Cloudflare’s CDN and security services saw their sites slow down, become unreliable, or become completely unreachable during the outage window.

클라우드플레어 장애의 영향을 받은 대상

The effects of the outage were felt globally. In less than 15 minutes, reports started to pour in about major outages across the internet. Even Downdetector, the platform dedicated to helping users determine if their favorite platforms are experiencing downtime, was down.

Other major platforms affected included the social media platform X, the world-leading AI provider ChatGPT, YouTube, Uber, and many others. Considering the vital role that social media plays in news today, it’s safe to assume any disruption in service affects people’s ability to stay up to date on the world’s happenings, both locally and internationally.

The outage also affected essential services and government agencies. Specifically, the Federal Energy Regulatory Commission (FERC) site was down. Even commuters suffered as essential service providers like the New Jersey Transit website was down for hours, leaving thousands without any way to plan out their routes.

문제 해결

Once Cloudflare identified the malformed Bot Management configuration file as the root cause, engineers rolled back the feature to the last known-good version. This corrected configuration was then propagated across Cloudflare’s global network to replace the oversized file that had triggered the 5xx errors.

As the updated configuration reached each data center, services that depended on Bot Management began recovering. Core traffic stabilized first, followed by Cloudflare Access and WARP, which took longer because they rely on deeper authentication layers. Most users saw normal functionality return within about three hours, with full restoration completed later in the day.

클라우드플레어에서 반복되는 문제인가?

This incident marks the sixth major outage in recent years that significantly affected global internet services. The company is by no means unreliable, but its crucial placement amongst the leading internet protocols means that any downtime results in global platforms becoming inaccessible. Here’s a look at their past outages and their causes.

2019년 6월 24일

In this incident, Cloudflare went down for hours due to a faulty border gateway protocol (BGP). This essential component helps to direct traffic between networks, providing a smoother surfing experience. The faulty BGP rerouted all of Verizon’s traffic through a single ISP, overloading it and causing several delays.

2020년 7월 17일

Around a year later, Cloudflare had another incident in which a faulty routing resulted in both Discord and Shopify (SHOP ) going down. This outage affected gamers who rely on Discord for communications. It also left the entire e-commerce sector seeking alternatives.

2020년 8월 30일

Only one month later, the company would have another minor incident in which several businesses were down. This scenario was the result of an IP fault. It was later revealed that the fault was due to the internet service provider CenturyLink.

2022년 6월 21일

In this incident, users received 500 errors, leaving their online data completely inaccessible. The outage lasted 1 hour and 30 minutes, affecting several fitness platforms. Specifically, Cloudflare, Fitbit, Peloton, and others were suddenly unable to connect to their users.

2023년 1월 24일

This outage occurred after Cloudflare integrated some new features designed to provide more oversight on token usage. Service tokens are an essential component of internet traffic. They act as credentials enabling access to services, applications, and essential services. The update accidentally wrote over other data, causing the entire system to go down for 121 minutes.

2025년 11월 18일

This latest incident marks the 6th time the service provider has gone down. In an interview, Matthew Prince, Cloudflare co-founder and CEO, apologized to users. He explained the cause of the problem and that this outage was the worst experienced since 2019.스크롤하려면 스와이프 →

클라우드플레어는 AI와 비즈니스에 필수적

One key takeaway from the Cloudflare outage is just how essential this system is to AI protocols. ChatGPT experienced widespread disruption, with many users unable to load conversations or receive responses, though some regions retained partial functionality. The delay and stress caused by the outage could be felt across industries.

클라우드플레어가 미래 장애를 방지하기 위한 계획

Notably, Cloudflare has already begun to implement new safety measures to prevent this incident from occurring again in the future. One key upgrade was a new early warning system that monitors feature file sizes and alerts developers if they become unmanageable. The system can even remove the files to prevent any downtime.

취약한 인터넷

People around the world seem surprised to learn just how fragile the internet is today. There are a handful of companies that remain critical to operations. If they go down, the entire internet suffers. The Amazon AWS outage last month was another reminder that at any moment, internet access could be delayed or completely shut down due to one company experiencing issues.

클라우드플레어 회사 배경

Matthew Prince and Lee Holloway founded Cloudflare in 2009, intending to help track and reduce spam emails. It emerged from the pair’s earlier works, including Project Honey Pot in 2004. This online fraud tracker saw strong support, winning the Harvard Business School Business Plan competition in 2009년 4월.

In 2010, the company hosted a public launch at TechCrunch Disrupt. It was unique at the time for several reasons. For one, it protects from online threats using a free DNS-based content delivery network (CDN) and security services.

Today, Cloudflare is a leading internet infrastructure provider. Since its launch, the company has secured $332M over 7 funding rounds, according to Tracxn reports. Notably, in 2021년 8월, Cloudflare completed a $1.125 billion convertible note offering, its largest capital raise to date.

클라우드플레어 자금 조달 역사

• 2 Early-Stage
• 3 Late-Stage
• 2 Post IPO

클라우드플레어 주식

News of the downtime hit the market immediately and caused the company’s stock to lose 4% in premarket trading. These losses were not permanent as the company has already regained around 2% of the lost value over the last 24 hours.

최신 NET(NET) 주식 뉴스 및 성과

클라우드플레어 경쟁사 및 시장 위치

Cloudflare ranks 1st among 361 active competitors in its sector. Its main competitors include Akamai, Limelight, and Verisign. Cloudflare has 19 institutional investors, including Franklin Templeton Investments, Fidelity Investments, M12, Baidu, Qualcomm (QCOM ) Ventures, and more.

클라우드플레어 장애 | 결론

As business systems become more intertwined with services like ChatGPT, Grammarly, Discord, and others, the likelihood of major outages will continue to increase. Hopefully, this latest incident will help to spur conversations regarding how best to buffer against these incidents moving forward. For now, enjoy your internet access – as it’s a lot more fragile than most will ever know.

Learn about other exciting Cybersecurity Developments Here.