Security Manager's Journal: Handling zero-days with zero staff

Computerworld - Today's security headlines are buzzing with news about a zero-day Java exploit. This newest threat takes advantage of vulnerabilities in Oracle's Java platform, and because of this, it affects a number of operating systems. The exploit is a highly sophisticated, cleverly written piece of code that appears to have been produced by talented programmers. Remember the days when malware was written by troubled teens, and inherent bugs in the code would burn out most of them before they could spread very far and do much damage? Now, professional programmers are being hired by governments intent on waging cyberwarfare or committing industrial espionage in order to gain competitive advantage.

This new exploit provides attackers with drive-by infection capability. End users don't even need to click on a link or execute a program. Simply visiting an infested website can infect a computer. Last May, I wrote about an earlier form of this kind of infection, a fake antivirus program that took over my own computer after I opened an infected Web page from a Google search. Malware like this is becoming more insidious, and more efficient. And it looks like it's going to continue to improve, as professional programmers continue to apply their skills to the black-hat world. So it's almost inevitable that some computers on my network will get infected and compromised. But how will I know when that happens?

One thing I've done to combat the threat of advanced malware on my company's network is to install devices on the network that detect the behavior of malware. I have a top-rated antivirus and security software suite on my endpoint computers, but it's signature-based -- and that's not enough, given the increasing numbers of zero-day threats we've seen. Signatures take a day or two to be released, and in the meantime, my network is at risk. The behavior-based network threat detection doesn't rely on signatures. It analyzes the content of network traffic in real time. When it sees patterns corresponding to malware or attempts to connect to command-and-control servers that are used by attackers to manage compromised computers, it blocks the traffic and sends an alert.

The question is, what to do with all those alerts? I have them going to my security information and event management (SIEM), which I talked about recently, along with alerts from my intrusion-detection systems and various other logs and data sources. I also have the more reliable malware alerts going directly to the desktop support team for cleanup. But the key to getting value from all this data ultimately results from human interaction. Somebody needs to look at the alerts, interpret them and decide what action to take. This requires skills beyond those of my company's network operations center (NOC), which is really only capable of assigning alerts to different people and making a determination about their urgency. Security alerts are not that easy to interpret, and it's hard to weed out the false positives.