Implement Port-Knocking Security with knockd

When dealing with computer security, you should assume that hackers will be trying to get in through any available doors your system may have, no matter how many precautions you might have taken. The method of allowing entrance depending on a password is a classic one and is widely used. In order to “open a door” (meaning, connect to a port on your computer), you first have to specify the correct password. This can work (provided the password is tough enough to crack, and you don't fall prey to many hacking attacks that might reveal your password), but it still presents a problem. The mere fact of knowing a door exists is enough to tempt would-be intruders.

So, an open port can be thought of as a door with (possibly) a lock, where the password works as the key. If you are running some kind of public service (for example, a Web server), it's pretty obvious that you can't go overboard with protection; otherwise, no one will be able to use your service. However, if you want to allow access only to a few people, you can hide the fact that there actually is a door to the system from the rest of the world. You can “knock intruders away”, by not only putting a lock on the door, but also by hiding the lock itself! Port knocking is a simple method for protecting your ports, keeping them closed and invisible to the world until users provide a secret knock, which will then (and only then) open the port so they can enter the password and gain entrance.

Port knocking is appropriate for users who require access to servers that are not publicly available. The server can keep all its ports closed, but open them on demand as soon as users have authenticated themselves by providing a specific knock sequence (a sort of password). After the port is opened, usual security mechanisms (passwords, certificates and so on) apply. This is an extra advantage; the protected services won't require any modification, because the extra security is provided at the firewall level. Finally, port knocking is easy to implement and quite modest as far as resources, so it won't cause any overloads on the server.

In this article, I explain how to implement port knocking in order to add yet another layer to your system security.

Would-be hackers cannot attack your system unless they know which port to try. Plenty of port-scanning tools are available. A simple way to check your machine's security level is by running an on-line test, such as GRC's ShieldsUp (Figure 1). The test results in Figure 1 show that attackers wouldn't even know a machine is available to attack, because all the port queries were ignored and went unanswered.

Figure 1. A completely locked-up site, in “stealth” mode, doesn't give any information to attackers, who couldn't even learn that the site actually exists.

Another common tool is nmap, which is a veritable Swiss Army knife of scanning and inspection options. A simple nmap -v your.site.url command will try to find any open ports. Note that by default, nmap checks only the 1–1000 range, which comprises all the “usual” ports, but you could do a more thorough test by adding a -p1-65535 parameter. Listing 1 shows how you can rest assured that your site is closed to the world. So, now that you know you are safe, how do you go about opening a port, but keep it obscured from view?