Zero Trust Architecture in Kubernetes

Chapter 6. Monitor All Access

We mentioned that authenticating users and machines at as many points as possible, regardless of whether they originated internal or external to your network, is key to a zero trust environment. Zero trust means nothing is trusted by default. Today’s cloud and hybrid cloud networks have no easily definable security perimeter, and it’s better for security in general to authenticate accounts as frequently as possible.

Now that it’s common wisdom in the tech industry (based on my peer interactions) that zero trust, when implemented properly, is a superior security model, it makes me imagine the havoc that the traditional perimeter security model facilitated for decades. “Ah ha, I breached through the DMZ and network perimeter with this decryption key some developer left stored in cleartext in a database! That’s right. I am the company’s network administrator. Let’s make some interesting changes to the .htaccess file on this web app’s Apache server...” Often, just one successful authentication exploit was all it took. Actually, let’s shift to the present tense. There are still lots of traditional perimeter security networks running today. Legacy tech can often be difficult to replace. That’s why Fortran programmers are paid so well.

One authentication vector at a perimeter is woefully insufficient. But in zero trust, we get rid of the perimeter concept and challenge users to authenticate (or present proof of authentication) every time they access anything. ...