
With the recent news of the Conficker worm and that it could lead to the world's largest botnet, and the fact that the topic comes up from time to time at the Institute for Cyber Security, I thought it was worth exploring what a botnet is and what it's not.
A botnet is not exactly what its portmanteau suggests, which is a bot network. If you've ever visited a chatroom, you know what a bot is--its a program that run very repetitive tasks; in the case of a chatroom, they might cycle through eight different phrases to get you to click on a link to their site. Since they are almost always very primitive and easy to identify, they easily fail the Turing test.
A botnet is better described as a "zombie network". A zombie is a computer that has been infiltrated by a malicious attacker (usually by proxy via another zombie) and now hosts a small program (a bot, let's say) that runs in the background that performs tasks on behalf of the attacker. To escape notice, zombie computers tend to function like they always did, but may run slower depending on how much of the machine the bot is using. Usually these bots' main purpose is to multiply, to remain hidden, and possibly scan for private data. Some are more sophisticated than others.
Since many computers are insecure (or not secure enough, let's say) there are plenty of botnets. In fact, a zombie machine may have multiple bots running in it (a zombie with multiple zombie masters!).
Thus, when an article suggests a botnet has affected a thousand machines, what it's saying is that there are one thousand computers (desktop, laptop, enterprise server, doesn't matter--all are susceptible) that have the same program running on them on behalf of the malicious attacker. The Conficker worm may have created upwards of nine million zombies.
One last bit--not every botnet is malicious. Some large-scale deistributed systems can make use of "spare cycles" on a machine (that is, the computer is on but idle) to perform a complex task. The SETI@Home project, which has over three million users, is a famous example. The project analyzes lare amounts of radiowave data, and more computers that help allow the project to cover greater frequency ranges with more sensitivity.
Erhan J. Kartaltepe,
erhan.kartaltepe-at-utsa.edu