The old way of modeling networks—that networks are living organisms that can't be controlled—is based on thinking that doesn't take into account basic science and engineering principles. Unfortunately for us, this kind of thinking has controlled our network designs and management techniques for too long.
We Need a Better Idea
The fact that we're not gaining any ground in the security world tells us that we need a better idea. We need to understand how our systems interact and how the information that they produce can be used to our advantage. We need to apply basic engineering control principles to our network to control how risk is introduced so that we can predict how our networks are going to react when they are attacked.
Trust vs. Risk
Although risk is an important factor in determining the overall state of security for a network, perhaps a better way of looking at individual endpoints should be based on trust. By setting a minimum level of configuration for each endpoint, you begin to build the element of trust into each system. You trust that a system will behave in a predescribed manner when faced with security stress.
Process Control Helps Build a Model
The basic engineering principle that can help us is based on process control technology. Process control technology uses the PID algorithm to ensure that a predetermined set-point is achieved and maintained within identified limits.
All the devices on the network have a role, and that role can be associated with some form of control. By using this new model, we can more accurately set an acceptable limit of risk, build trust, and thus protect our networks more effectively. In addition, we can identify those elements of technology that are wasting our time and budget, because we will more easily understand their role and contribution in the solution.
Business Processes Cannot Be Ignored
As you map control processes to control modes and feedback paths, you must remember to look at the business process that they affect (and vice versa). The human element plays a large role in those processes and in many cases is the most unreliable or variable element within it.
We Need a Common Language
Like all other engineering disciplines, the information security discipline needs a universal set of icons and nomenclature that allows security professionals to exchange information effectively and reliably. Our present system of scratching out bricks with fire on them and clouds representing networks and the associated endpoints isn't working.
The set of icons presented in this chapter form the foundation for a schematic representation of our security elements and their associated control processes.