Update: Attacks on Google may have been work of amateurs

Security firm says the attacks may not have been nearly as sophisticated as some thought

Contrary to popular assumptions, the recent cyberattacks against Google and more than 30 other high-tech companies were carried out by relatively unsophisticated attackers using outdated botnet tools, according to Damballa Inc., an Atlanta-based security firm.

The company, which offers a range of botnet protection services, released a report Tuesday based on what it said was a detailed analysis of the data surrounding the attacks, the malware that was used and the command-and-control topologies used by the perpetrators.

It reveals that the threat can "best be classified as just another common botnet attack and one that is more amateur than average," the Damballa report noted. "The attack is most notable not for its advanced use of an Internet Explorer 6 Zero-Day exploit, but rather for its unsophisticated design and a pedigree that points to a fast-learning but nevertheless amateur criminal botnet team," the report said.

Google in December disclosed that it had been the victim of a targeted cyberattack designed to steal intellectual property, e-mail and other data from the company. It said the attacks appeared to originate from China and also affected many other high-tech companies.

In an e-mailed comment, a Google spokesman said that Damballa does not have any firsthand knowledge of Google's investigation of the attacks. "Beyond that, we are not going to comment on our ongoing investigation. We stand behind our original statement," he said.

The attacks, dubbed Operation Aurora after the botnet that was used to launch them, were described by Google and many in the industry as being particularly sophisticated and state-sponsored. Some called it a classic example of a new category of Advanced Persistent Threats (APT) facing many commercial enterprises.

According to Damballa, however, the actual attack data suggests otherwise. The Dynamic DNS (DDNS) command-and-control system that was used to control the Aurora botnet for instance, is "old school" and is rarely used today by professional criminal botnet operators. The reliance on a DDNS command-and-control infrastructure suggests that those behind the Google attacks were "new and amateur botnet operators," Damballa said.

The malware tool most commonly associated with the Aurora botnet is also a relatively unsophisticated Trojan Horse program called Trojan.Hydraq, said Gunter Ollmann, vice president of research at Damballa. In addition to Hydraq, the attackers also employed a variety of other malware tools on the Aurora botnet, many of which were unsophisticated and based on old obfuscation and evasion techniques, Ollmann said.

Compared to malware tools such as Conficker, the malware that was used in the Google attacks was "orders of magnitude" less potent, he said.

What the data shows is that the Google attacks were most likely carried out by relatively inexperienced attackers who were experimenting with different attack tools and techniques, Olmmann said. The fact that they were relying on DDNS itself is a major giveaway, he said. Botnets that use DDNS services are relatively easily for law enforcement authorities to shut down, which is why most professional botnet operators don't use them anymore.

"At the end of the day, what is pretty important to understand is, if this type of an unsophisticated attack was successful against these types of organizations, professional botnet operators have a much better chance of breaking in," he said.

The data also suggests that the attacks may not have been as targeted as Google and others might have assumed them to be, according to Damballa. The Aurora attacks appear to have first originated in July 2009 from mainland China. According to Damballa, a university in China and a Chinese collocation facility were critical "early incubators" of the infection.

By the time Google first discovered it had been attacked, computers in at least seven other countries had been similarly affected by the same botnet. And by Jan. 12, when Google publicly disclosed the attacks, computers in as many as 22 other countries had also been compromised.

One security researcher at the RSA Security Conference said that the sophisticated part of the attack was not really the botnet itself or the malware -- it was the social engineering used to target victims and the way criminals traversed a victim's networks.

Alex Stamos, a partner with Isec Partners Inc., said that the cybercriminals escalated privileges on the Windows network, accessed Active Directory servers and cracked database passwords -- then using that information to steal data. Because all antivirus companies are now detecting the Aurora code, "the malware is gone," Stamos said. "These guys are never going to use that malware again."

But one security researcher at the RSA Security conference said that the sophisticated part of the attack was not the botnet or the malware. According to Alex Stamos a partner with Isec Partner, the sophisticated part was the social engineering involved in targeting victims and the way the criminals moved around once they were inside victim's networks, escalating privileges on the Windows network, accessing Active Directory servers and cracking passwords in its database, and then using that information to steal data from targets.

Because antivirus companies now detect the Aurora code, "the malware is gone. These guys are never going to use that malware again," he said.

Robert McMillan of the IDG News Service contributed to this report.

Jaikumar Vijayan covers data security and privacy issues, financial services security and e-voting for Computerworld. Follow Jaikumar on Twitter at @jaivijayan or subscribe to Jaikumar's RSS feed . His e-mail address is jvijayan@computerworld.com.

From CIO: 8 Free Online Courses to Grow Your Tech Skills
Join the discussion
Be the first to comment on this article. Our Commenting Policies