Botnets, Part 1: Why They Strike and How to Defend Against Them
Botnets have been overhyped yet underreported. Meanwhile, increasingly sophisticated and hard-to-detect botnet attacks have lulled many organizations into complacency.
Could cyber attacks really black out cities, shut down phone services, and halt trains? Have 150 million computers already become botnet zombies? These warnings, some experts say, have been overhyped. Meanwhile, today botnets are running phishing scams; hosting porn and pharmacy websites; pumping out spam; and committing click fraud, extortion, and economic espionage. They also are becoming harder to detect. According to a June 13, 2007 FBI warning, botnets pose "a growing threat to national security, the national information infrastructure, and the economy."
Defenders are rising to the challenge. For example, two companies, FireEye and Damballa, now offer user-friendly defenses targeted specifically against botnets and give free trials to qualifying organizations. I'll cover the details later in this article.
The Botnet Threat
Botnets consist of software stealthily installed on many computers and remotely controlled by a central authority. Originally, botnets combined the technologies of worms to recruit "zombie" computers, Trojan backdoors for remote control, and Internet Relay Chat (IRC) bots for command and control (C&C).
Attacks by these early forms of botnets first became common in 1999. Most notoriously, in February 2000, an underage Canadian boy used a distributed denial of service (DDoS) botnet to shut down the websites of Amazon.com, Dell, eBay, Yahoo!, and CNN. Soon thereafter, President Clinton's cyber czar, Richard Clarke, warned that cyber attacks could become "as bad as being attacked by bombs." He suggested a harrowing scenario: "A President goes forth and orders troops to move. The lights go out, the phones don't ring, the trains don't move. That's what we mean by an electronic Pearl Harbor."
Network engineers quickly learned to filter out script-kiddie DDoS attacks. Even when attacks succeeded, warnings of dire consequences turned out to be overblown. In July 2001, the Code Red worm combined with a Baltimore tunnel fire to trigger Internet outages. Despite this, an "electronic Pearl Harbor" failed to materialize.
Botnet attacks on Estonia in May 2007, and against Georgia during the August 2008 Russian invasion, suggest that such attacks are not decisive during time of war—so far. However, the botnet scene is evolving. Bot herders increasingly use more sophisticated C&C; for example, web drop boxes. To defeat black holing, they compromise reputable servers and protect their C&C nodes with "fast-flux" domain name servers (DNS) that update IP addresses so fast that they can't easily be identified, much less blocked. To prevent hijacking or analysis of their botnets, they encrypt their C&C.
Unlike old-fashioned worms, the vectors that recruit zombie computers often are tightly targeted, keeping them under the radar of antivirus and intrusion-detection systems. Some botnets use social engineering, a technique that's nearly impossible to combat. Others compromise legitimate websites; for example, with SQL injection attacks, so that merely browsing turns the user's computer into a zombie.
Today's botnets often are designed to detect the defenders' sensors and hide from them. Some disable antivirus and intrusion-detection systems in ways that ordinary users are unlikely to notice.
Driving this evolution is the fact that cybercrime organizations rarely use the easily detectable tools created in the virus and hacker undergrounds. Instead, they employ professional programmers. At the November 2006 Adaptive and Resilient Computing Security (ARCS) Workshop in Santa Fe, NM, Dr. Andrew Jones, head of security technology research at British Telecom, commented, "In some parts of the world, a gentleman will persuade you to work for him if you are good enough."
Another factor causing the increasing danger of botnets, says FireEye's Phillip Lin, is that many IT managers "think botnets just mean DDoS attacks." In a phone interview, he suggested that when such attacks fail to materialize, organizations feel a false sense of safety.
These seemingly harmless results may lure victims into failing to act even when they discover botnet infections. Rick Wesson, CEO of Support Intelligence, stated in the New York Times, "Many corporations seem to think it's O.K. to be infected several times a month" with botnets.
"But really," says Lin, botnets now are "platforms to distribute malware." For example, Damballa's Bill Guerry reported in a phone interview that Damballa researchers have been tracking a bot army that has been employing some 10,000 different varieties of malware.
Thanks to these factors, botnets can cause enormous harm before victims realize that they're under attack.
The Coreflood/AFcore botnet is an example that has lulled sysadmins into complacency while enriching its crime masters. According to Joe Stewart, director of malware research at SecureWorks, it has been active for more than six years. He recently obtained one of its C&C servers, and found a literal treasure trove of financial fraud data. Of its contents, he reports, "8,485 passwords are for banks and credit unions in the U.S. and overseas. Of the 29 banks, 12 are foreign banks and the rest are based out of the U.S. The majority of them are banks and not credit unions.... [The controller] had been in continuous operation since 2005. It is still in operation even today, on another server, only interrupted for a few days. [Fifty] gigabytes of data stolen from infected users was left behind on the machine... from a six-month time period...in some cases thousands of computers on corporate and government networks were infected."
Even when the victim organization recognizes botnet compromise, it may choose not to fight back. At the same ARCS workshop mentioned earlier, Richard Power of GS3 Intelligence noted, "An entity with tens of billions in revenue launched a serious internal security program" in response to suspected industrial espionage. They discovered a "sustained information elicitation attack." However, an executive with the victim company brought the investigation to a standstill when he told them "the target issue was secret." Power says this sort of balking is typical because such organizations "don't want to risk their secrets to security professionals."