Researchers at Google have used the company’s Safe Browsing infrastructure to analyze the distributed denial-of-service (DDoS) attacks launched recently against the anti-censorship organization GreatFire and GitHub.
Google’s Safe Browsing is designed to analyze web pages and determine if they host malicious content. In the case of the DDoS attacks targeting GitHub and GreatFire, Google used the system to observe the attacks at HTTP protocol level.
The attacks were launched with what experts have dubbed the “Great Cannon,” a tool co-located with the Chinese government’s Great Firewall censorship system. The Great Cannon is a man-in-the-middle (MitM) tool that injects malicious packets into unencrypted traffic. In the attacks against GitHub and GreatFire, the attackers replaced HTML and JavaScript elements hosted on Baidu servers with JavaScript code designed to make repeated requests to resources on the targeted domains.
Google’s Safe Browsing showed that the attackers started injecting content on March 3, when the first phase of the attack started. This stage, which lasted until March 6, was used for testing purposes. On the first day, the number of requests was limited, but the limitations were removed between March 4 and March 6.
According to Google, the second phase started on March 10 and ended on March 13. During this phase, a single IP address associated with the sinajs.cn domain was targeted. Over the next days, the attackers started targeting several cloudfront.net hosts with requests served both via HTTP and HTTPS.
The attacks against cloudfront.net hosts, in which greatfire.org was targeted, ceased on March 25. That’s when the Great Cannon was aimed at the GitHub pages of GreatFire and a repository hosting mirrors of the Chinese-language New York Times website.
The DDoS against GitHub, which is said to be the largest attack of this kind in the website’s history, stopped on April 7, Google said.
The search giant noted that the Safe Browsing data doesn’t allow its researchers to confidently attribute the attack.
The company also pointed out that such attacks would not be possible on an encrypted Internet.
“Had the entire web already moved to encrypted traffic via TLS, such an injection attack would not have been possible,” said Niels Provos, distinguished engineer in Google’s Security Team. “This provides further motivation for transitioning the web to encrypted and integrity-protected communication.”
