Final Thoughts on Hiding Information
Hiding information has several positive effects:
It breaks a network into failure domains by limiting the scope of devices that must react to any particular change in topology or reachability.
It reduces the velocity and scope of control plane state, allowing network to scale to larger sizes while retaining network stability.
It is a “hook” through which to implement policy, specifically in relation to network security.
It might seem hiding more state is always better, based on these advantages. However, as with all things in network engineering, the truth is closer to a tradeoff. If you have not found the tradeoff, you have not looked hard enough. In the case of information hiding, refer back to Chapter 1, “Fundamental Concepts,” specifically the section on complexity, and the example given concerning stretch and route aggregation. A second instance of hiding state can be found in relation to micro-loops, which are explained in Chapter 13, “Unicast Loop-Free Paths (2).” The more you slow down the velocity of state, the longer such microloops will exist in the network.
Hiding state is, then, a useful tool in the hands of good designers, but it can also cause many problems by negatively impacting network performance.