Major breakdown at Cloudflare, which provides services for 20% of the internet / The websites of several publications in Romania could not be accessed / What caused the interruption for hours
Web infrastructure company Cloudflare fully restored its services Tuesday evening after a major outage earlier in the day that prevented thousands of people from accessing major Internet platforms, including X and ChatGPT, Ruters reports.
Numerous websites and online platforms around the world went down temporarily on Tuesday afternoon, including media websites in Romania being affected by technical problems of Cloudflare, the American company used by almost 20% of the Internet for various web services.
Cloudflare announced that the outage, which began around 1:30 p.m. Romanian time, was caused by an automatically generated configuration file designed to handle potential security threats.
The file grew too large and crashed the software that manages traffic for several Cloudflare services, said the company, whose network handles about a fifth of web traffic.
The US company also said it has begun investigating the cause of the problem and has implemented a solution, but some customers may still be affected as service is restored globally. Its shares fell 2.3% in early trading.
Cloudflare added that “there is no evidence that this was the result of an attack or caused by malicious (malicious, no) activity.” The company operates one of the world's largest networks that help websites and applications load faster and stay online, protecting them from traffic spikes and cyber-attacks.
The interruption of the Cloudflare service prevented access to many websites, but also to global platforms such as X, Spotify, OpenAI, Canva, Letterboxd, Grindr or Truth Social – the social network owned by the Trump Organization.
Last month, an outage in Amazon's cloud service caused global unrest as thousands of popular websites and apps, including Snapchat and Reddit, became inaccessible due to AWS service outages.
As of Tuesday evening, all sites that were down due to Cloudflare's difficulties appear to be back to normal. In fact, the website downdetector shows that complaints registered globally against the operation of Cloudflare have dropped to zero.
Cloudflare first explains about problems that have caused numerous sites around the world to go down: “Customers may continue to see higher error rates.”
Reuters notes that the number of complaints of Cloudflare service outages began to decline after a peak in the afternoon around 2:00 p.m.
The US company said in an update that it has already implemented a fix for the technical issues, but that some customers may still be affected as services return to normal.
Including downdetector, the benchmark site that monitors in real-time technical problems encountered by the most popular sites and online platforms, went down briefly on Tuesday before 2:00 p.m.
According to downdetector, among the sites and platforms affected were social networks X and Truth Social, Spotify, Letterboxd, OpenAI or Grindr.
The websites of several publications in Romania have crashed, being affected by this problem. Some have come back online in the meantime. The situation is fluctuating, with some sites going down, coming back online, then becoming inaccessible again.
What is Cloudflare, the US web service provider that caused a major global internet outage on Tuesday, including in Romania?
Why Cloudflare issues affected so many sites
Cloudflare is a California-based internet infrastructure company that provides many of the core technologies that power today's online experiences, as The Independent points out. The company provides tools that protect websites against cyber-attacks and ensure they stay online even under high traffic conditions, for example.
“Cloudflare is aware of an issue that may affect multiple customers and is investigating it,” the company said in a recent update. “Further details will be provided as more information becomes available,” she added.
Affected users saw a message indicating that there was an “internal server error on the Cloudflare network.” The message asked users to “try again in a few minutes.”
The problems started around 13:30 Romanian time, although at that time some sites appeared to load when refreshed. Cloudflare released an update acknowledging the issue about 15 minutes later.
Cloudflare provides internet infrastructure services to companies around the world, usually in a way that is invisible to users. As a result, a wide range of seemingly unrelated sites were affected by the outage, as was the case with a similar technical problem at Amazon Web Services (AWS) last month.
The major internet outage also affected the National Cyber Security Directorate / What DNSC says
How Cloudflare works
To understand how Cloudflare works, it's helpful to consider common mistakes made by websites in the past.
When a user visits a site without Cloudflare or other similar services, the site visitor requests content from a server. However, when there were too many visitors on the server at the same time, it would overload, causing the site to load slowly and crash altogether.
Cloudflare has invented a solution for this problem.
The company has placed its own server – in the form of a huge worldwide network – between the website and the server that hosts it. Website visitors no longer communicate directly with the server, but with the Cloudflare network, which already stores the website content and delivers it through a server, depending on the visitor's location.
For example: you have a website hosted in Romania, and someone from the US tries to connect to that website. The request must travel a long distance (between the site and the web server). Cloudflare solves this problem by offering a vast network of servers around the world.
The American no longer has to connect to the server in Romania, but will connect to the nearest Cloudflare server, somewhere in the United States.
Thus, Cloudflare frees up the servers and ensures that visitors can see the requested web page faster when they are working properly.
