Thanks to several of you for asking for my opinion of the article Carpet bombing in cyberspace: Why America needs a military botnet by Col. Charles W. Williamson III. I'd like to cite a few excerpts and comment directly.
The world has abandoned a fortress mentality in the real world, and we need to move beyond it in cyberspace. America needs a network that can project power by building an af.mil robot network (botnet) that can direct such massive amounts of traffic to target computers that they can no longer communicate and become no more useful to our adversaries than hunks of metal and plastic. America needs the ability to carpet bomb in cyberspace to create the deterrent we lack...
This is interesting. Why do we need to project force in cyberspace to deter our enemies? Cyberwar is usually cited as a means of conducting asymmetric warfare, meaning one side is much weaker than other in conventional means. Cyberwar is expected to be conducted against US assets (critical infrastructure) because the enemy lacks the capability to destroy or degrade that asset using kinetic weapons. If we can deter enemies using our existing, overwhelming kinetic force, why possess an ability to "carpet bomb in cyberspace?"
Today’s air base defense concept still uses a layered defense in depth, but it starts as far as possible from the air bases, then relies on close-in defense only as a last resort. That capability in cyberspace can exist in an af.mil botnet...
The U.S. would not, and need not, infect unwitting computers as zombies. We can build enough power over time from our own resources.
Rob Kaufman, of the Air Force Information Operations Center, suggests mounting botnet code on the Air Force’s high-speed intrusion-detection systems. Defensively, that allows a quick response by directly linking our counterattack to the system that detects an incoming attack. The systems also have enough processing speed and communication capacity to handle large amounts of traffic.
Oh, that's a great idea. Let's tie up the really only useful element of the Air Force's defense -- that which provides some degree of situational awareness -- with the task of packeting someone.
Next, in what is truly the most inventive part of this concept, Lt. Chris Tollinger of the Air Force Intelligence, Surveillance and Reconnaissance Agency envisions continually capturing the thousands of computers the Air Force would normally discard every year for technology refresh, removing the power-hungry and heat-inducing hard drives, replacing them with low-power flash drives, then installing them in any available space every Air Force base can find. Even though those computers may no longer be sufficiently powerful to work for our people, individual machines need not be cutting-edge because the network as a whole can create massive power.
I see... so the very network that is important enough to be deemed a "weapons system," thanks to the logistics, communication, and related traffic it carries, is going to be filled with tons of DDoS traffic from recycled PCs? Do you think QoS is supposed to take care of this problem?
After that, the Air Force could add botnet code to all its desktop computers attached to the Nonsecret Internet Protocol Network (NIPRNet). Once the system reaches a level of maturity, it can add other .mil computers, then .gov machines.
To generate the right amount of power for offense, all the available computers must be under the control of a single commander, even if he provides the capability for multiple theaters. While it cannot be segmented like an orange for individual theater commanders, it can certainly be placed under their tactical control.
I am sure the botnet software installed would be super secure. Can you say "biggest latent botnet" in history? Every single .mil and .gov computer under the control of a single commander -- probably a Russian or Chinese infiltrator? Just who is Col. Williamson working for, anyway?
This is a really dumb idea, at least as presented. I'm all for Taking the Fight to the Enemy, but building a botnet on operational networks, especially on operational defensive systems and even production equipment, is just wrong. If we want to remove someone from the network, it's far simpler to disable the right cable using conventional means.
Let's assume the Air Force did build a botnet, on separate, non-production computers, on dark space, ready to point towards an enemy. Where would that target be? No single, or handful, of computers DDoS'd Estonian infrastructure. (Every military planner loves to cite Estonia these days.) If someone decided to DDoS one or more US computers or routers, where would we point our botnet? Where would Estonia have pointed any botnet it owned -- Russia? Back at the computers DDoSing Estonian assets? How is this supposed to work? "Don't DDoS us or we'll DDoS you?" Mutually Assured DDoS?
There are smarter ways to conduct operations in cyberspace, and this is not one of them. Back to the drawing board, sir.