Complexity with a clear purpose.

A proposal at work called for multiple routing profiles. Looking through the implications, I realised that the design would require every border router to support a very large number of routes.

I was also concerned about the reliability of sharing routes between separate routing instances (VRFs) at that scale. Together, those requirements would limit our equipment choices to models with very high routing capacity and add complexity to the network.

I recommended that we not proceed, and we didn’t. It’s a decision I’m still glad we made, because of the complexity and maintenance burden the design would have introduced.

One thing work has taught me beyond the textbook is to treat routers as hardware with limits. A design can make sense on paper and still run into those limits if they aren’t considered early enough.

That shapes how I approach network design: I want to understand what the equipment can actually support before relying on it.