EXPERT RESPONSE
Security researchers generally view these as a bad idea, although there has been some exciting research from the University of Washington centered on a project called Phalanx (pdf). The idea is that any server requests would have to be processed through the "good" botnet, which is geographically dispersed. Because a large number of servers are implemented as intermediaries, it becomes difficult to overwhelm one specific network link.
Still, as I stated earlier, I believe that this is a bad idea, for two reasons. First, think about how hard it is to secure existing systems. Now, expand that by a few thousand systems directly accessible from the Internet. This scenario leads directly to my second fear: control. Imagine the public relations nightmare should your good botnet be taken over and used to DoS someone else's network.
I propose that instead of building counter-botnets, security professionals could better spend their time tracking the patch-installation success rate for the systems they currently have. Leave the bot-herding to the bad guys.
More information:
|
