Agencies feel botnets' light footprint
"If I'm the Chinese trying to take over the State Department's computers, I don't use thousands of bots ... I use one or two." Alan Paller, SANS Institute
Henrik G. de Gyor
In the vast world of the Internet beyond the federal government's borders, millions upon millions of invisible, automated soldiers are laying siege to the computers of companies and citizens alike.
These 'soldiers' are bots'computers that have been taken over by worms, Trojans or other malware, knit together into vast networks and directed by bot-herders, the creators and controllers of these networks, to spew out spam and other useless or damaging content.
Government computers, too, are being corrupted and used as parts of a botnet, but the real threat is more insidious.
Hackers, particularly nation states, are using stealth botnets'they might be termed spearbots'to steal information from federal systems, experts say. This is different from the typical, wide-ranging cyberattack which looks for any hole in the defense.
'Botnets are going to be much more stealthy [in government computers] because of what they're trying to get,' said Buck French, CEO of Securify Inc., a security monitoring and control company based in Cupertino, Calif. 'They're not trying to get compute power, they're trying to steal sensitive data. ... They'll have a stealthy, light footprint, because with all the effort it takes to get it there, they don't want it detected.'
Alan Paller, research director of the SANS Institute in Bethesda, Md., said, 'If I'm the Chinese trying to take over the State Department's computers, I don't use thousands of bots, because it won't do any good. I use one or two. ... It's a targeted attack; that's what's been happening in the past year.'
Researchers are looking for ways to detect the existence of a bot after it has gotten onto a computer, Paller said.
'That's the Manhattan Project [of security research], the biggest, hardest project, to get technology to help you find these things,' he said. 'They're also working on 'fingerprinting' all files, so they can tell whether the files on a machine match the fingerprints of questionable software. Both approaches have the same goal, to find the bad stuff. One is more computer behavior-based, the other is looking at the software itself.'
Government systems are safer than private-sector computers because many, if not most, botherders design their malware to avoid government computers, said Richard Barger, senior principal consultant in the federal consulting services division at Symantec Corp. of Cupertino, Calif.
'Traditionally, botnets are a low threat to [Defense Department] networks because they're loud,' creating high-profile traffic, Barger said. 'When we do reverse code engineering, we see they actually stay away from .mil or .gov e-mail addresses.'
There are good reasons for bot-herders to stay away from government systems, French said.
'There are easier systems to put bots on than government systems,' he said. 'They're more likely to go after universities [and] home systems ... than federal systems, where it might be a federal felony.'
Still, the U.S. Computer Emergency Readiness Team in the Homeland Security Department sees an increased number of incidents in the private and public sectors (see chart, below).
And because of the growth of these threats, DHS has been running a pilot program to detect bots trying to remove information from government systems, said Jerry Dixon, director of DHS' National Cyber Security Division.
The Einstein Program, as it's called, is a combination of government and commercial software and hardware, Dixon said. It sits outside the firewall so it sees everything interacting through the Internet with federal systems. It does not look at the payload, he said.
'We did an 18-month [privacy impact assessment] to ensure we were not looking at any private information,' Dixon said. 'We look at the network address and source destination, so we know where the traffic is going to and from. If we see traffic going to a command-and-control computer, we know there might be a problem.'
If there is a problem, DHS notifies the IP provider and asks them to take it offline.
If the botherder is outside the United States, DHS works with partners in the United Kingdom, Australia, South Korea and India to shut it down, Dixon said. If the source is in a nation the U.S. doesn't have good relations with, Dixon said, 'We rely on Korea-CERT and India-CERT to work with countries like North Korea or China or others that they might have better ties to.'
Using Einstein, seven departments currently monitor outbound Internet traffic, Dixon said. When U.S.-CERT becomes aware of botnet activities, the organization can use Einstein to notify agencies, which then investigate to see if they have a problem. If so, NCSD will evaluate whether it's a new vulnerability or a new type of bot.
Spread the Word
If it is a new vulnerability, NCSD tries to validate it and works with vendors to get antivirus software or firewalls fixed. But if it's a new kind of bot, 'we write up a paper and send it out to all agencies through the [Government Forum of Incident Responders and Security Teams] network,' Dixon said. 'We give them a heads-up about the virus and new definitions coming, and remind them to check gateway logs for command-and-control channels.'
Einstein has shown real value, he said, and NCSD is now moving it out of the pilot phase.
'We had to get Einstein certified and accredited,' he said. 'We have quite a few agencies lined up to deploy the program.'
Dixon added that defending against botnets is like playing a game of whack-a-mole because 'they are so prevalent,' and they keep popping up everywhere you look. 'It comes back down to machines being vulnerable and not being properly protected.'