Mitigation vs Remediation
Working around the problems we cannot solve
Network Operations teams make a clear distinction between mitigation and remediation. They must. Fixing a problem is not the most urgent goal. Returning a service or a network experience to what is expected, is the most urgent goal. You might even go so far as to say mitigation is the raison d'etre of dynamic IP routing.
Once an incident is observed, Network Operations teams often act first to implement a mitigation, for example route around a problem in the network, changing weights / costs if need be. That is not the same as solving / repairing the problem in the network. Repair may require more investigation and more time.
It is in this sense that I wrote about Routing not being a solved problem. The inclusion of a TTL is not solving the problem of loops, it is merely a mitigation of the effects of them, endlessly looping packets. If loops were a solved problem a TTL would not be required.
There are many information challenges and many information concepts which apply to a holistic view of networking, for example:
However, in the article Routing Will Never Be A Solved Problem, there was a focus on the “physics” of information transfer, simply because they are the most widely understood and least refutable.
For example, latency is such a well known, experienced, and accepted idea by networking professionals, it is relatively easy to make robust arguments using it, including:
There is no way to ensure all routers, have the same information, at all times, because information will take longer to travel to some routers than to others.
We can imagine various mitigations for this problem, but we can never truly remediate / solve it, in this universe, with the laws of physics that we currently understand.
So yes, routing can be a discipline with our best current mitigations, but some things can never fundamentally change or be “solved”. Generally, our mitigations will also have imperfections, like a packet looping until it hits TTL, and never actually making it to the desired destination. While TCP might kick in to ensure the packet does make it eventually to the destination, that does not mean the IP layer problem was solved, nor does it mean the packet will make it to the destination by the desired time.
Mitigation is a beautiful thing, especially if you work in network operations. It is also a beautiful thing in engineering, because it allows us to move on to the next problem.
However mitigation is not remediation, repair, or solving an underlying problem.