Security researchers have looked at ways to abuse the domain-name service (DNS) for years. From denial-of-service attacks aimed at domain-name servers to various forms of domain hijacking, the DNS system is a perennial target of researchers and attackers alike.

Now some researchers are warning that the protocol may increasingly be used to help criminals communicate with compromised systems. At last month's RSA Conference, Ed Skoudis, a senior security consultant with InGuardians, predicted that more malware would hide its commands and exfiltrated data in DNS packets. The advantage for malware writers is that even if a company bars a potentially infected computer from contacting the Internet, malware could send DNS requests to a local server, which would then act as a proxy, bypassing defenses.


"With DNS as a command-and-control channel, as long as the internal machine can resolve names on the Internet, then there is command and control," Skoudis told attendees. "A machine could be blocked by sending outbound connections, so, instead, the machine sends a request to its internal DNS server, and that DNS server forwards onto the Internet, ultimately getting to the bad guy's server."

To date, the tactic has been relatively rare: Perhaps a dozen malware variants have used the domain-name system to send commands and updates to botnets. Skoudis cited two malware attacks that used a DNS channel as part of major breaches. Two relatively minor attacks -- PowPow and Wibimo -- have used the technique as well, according to Dell SecureWorks.

In all, around 5 percent of attacks contain some sort of DNS signaling, according to researchers at Internet Identity, a firm that specializes in securing Internet infrastructure and information. "It's the kind of communication that is not detected because most people are not watching DNS," says Lars Harvey, the firm's CEO.

The covert or malicious use of DNS for communications typically falls into two categories, says Gunter Ollmann, vice president of research at threat-protection firm Damballa. Tunneling uses the DNS port 53 to bypass firewalls that typically leave the port open so as not to interfere with domain lookups. The second technique hides data in DNS packets and uses the domain-name system infrastructure to transport the information to the destination chosen by the attacker.

It's the latter method that will be used more often, Ollmann says. Today, most command-and-control traffic is carried over the Hypertext Transfer Protocol (HTTP), but as defenders find ways of blocking the traffic, attackers will look for other possibilities.

"The bad guys are becoming more aware of how DNS is a weak spot in our defenses," he says. "And as other defensive technologies have improved inside enterprise networks, that particular door -- port 53 -- looks more attractive."

Detecting DNS is not that difficult, however. Sending communications over the domain-name system typically results in anomalous volumes of information, which is fairly easy to detect if a company is watching DNS. Ironically, the domain-name service security extensions, or DNSSEC, could make shipping data over DNS easier to hide. Normal DNS traffic has a limit of 512 bytes, while DNSSEC packets have no limit, so the protocol could be used to exfiltrate data from an unwary company's servers.

"If the volume of traffic is anomalous, any system that detects anomalies will detect it," Ollmann says. "If you do have deep-packet inspection capability, then you will detect the low and slow, once-a-week transmission of data."

At least one researcher argues that the difficulty in hiding covert DNS communications makes it unlikely that the use of the protocol will take off in the criminal underground. Communications are hard to hide inside DNS packets if a company is looking out for odd traffic patterns, says Joe Stewart, director of malware research at Dell SecureWorks.

"There is nothing incredible stealthy about it," he says. "It is hard to blend in with DNS traffic and pass a reasonable amount of data inside the packets."

Companies need to be watching their DNS traffic to make sure online thieves are not communicating with compromised systems or stealing data from the corporate network, researchers say. While logging all traffic is too onerous, companies should watch out for DNS traffic issued to parts of the world where they do no business and use traffic monitoring to look for anomalies.

Have a comment on this story? Please click "Add Your Comment" below. If you'd like to contact Dark Reading's editors directly, send us a message.