Information-security managers always have to deal with trade-offs in securing their company's data, systems, and networks. Dedicating too much time and budget to the wrong areas could leave their business vulnerable.

Attackers have trade-offs, as well: A common one pits the stealthiness of an attack versus how well it can withstand the countermeasures used by defenders.

At the recent USENIX Security Conference, for example, a group of researchers from the Georgia Institute of Technology and security firm Damballa showed that domain-generation algorithms (DGA) -- a technique used by attackers to foil takedown attempts -- create an easily detectable fingerprint in the domain-name system (DNS) traffic that emanates from an infected system.

"You can use a fairly lightweight technique to determine if DGA botnets are inside your network by observing two fundamental things -- how nonexistent (NX) domains are on the rise and how herds of servers are looking up similar NX domains," says Manos Antonakakis, principal scientist at Damballa, who presented the work.

Attackers' calculus on whether to harden their botnets and compromised systems from defenders' countermeasures can help investigators glean clues as to the motives of the intruders. Different attackers will weigh the decision of stealth versus resilience differently. Opportunistic attackers, such as cybercriminals, will generally choose to harden their systems, while attackers that want to stay persistent will likely attempt to avoid actions that might draw attention, says Brett Stone-Gross, senior security researchers at managed-security service provider Dell Secureworks.

Attackers focused on cyberespionage and covert operations -- known in the defense industry as the "advanced persistent threat," or APT -- create stealthy malware that focuses on deniability over dollars.

"APT malware is generally throwaway malware," says Stone-Gross, pointing out that many of the recent spying attacks are less sophisticated than their cybercriminal counterparts. "They pump out a lot of samples because they are not that difficult to program."

The Gameover Zeus botnet used domain-generation algorithms as a fallback mechanism to re-establish control over the botnet, if defenders somehow were able to sever the attacker's control of the botnet.

[ Many companies do not scrutinize their domain-name service traffic, leaving an opening for malware to communicate using the protocol. See Malware To Increasingly Abuse DNS?. ]

Another technology that has the same trade-off is peer-to-peer networking, which many malicious programs use to foil takedown by removing easily identifiable command-and-control nodes. While peer-to-peer networking hardens the attackers' infrastructure, a reasonably aware monitoring solution can easily detect it.

"Peer-to-peer botnets are even more noisy and the traffic is very distinctive," Stone-Gross says. "They have to constantly refresh their peer list, and so they have to frequently connect to peers and update the peer list."

Cybercriminals are more interested in the quick score, rather than waiting patiently for the right bit of information, says Vikram Thakur, principal security response manager for security firm Symantec. Hardening their infrastructure delivers valuable time to siphon on more money from victims. And, in many cases, the chance of detection is so low -- the response to any positive detection so slow -- that the criminals are convinced that stealth is not necessary.

"Sifting through the data is difficult," Thakur says. "And the malware authors are saying, 'I will take my chances.' It is a reasonable risk for them to take to protect their resilience of their botnet."

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.