Redundancy and Multiple Paths

The internet is not one long wire. It's a mesh of routers, and between any two points there are usually several possible routes. Redundancy is the deliberate choice to build more connections than the bare minimum, so that if one link goes down, data can reroute through another. This costs more money and equipment up front, extra cables, extra routers, extra maintenance. The payoff is that a single cut cable, a single failed router, or a single overloaded link doesn't take down the whole conversation. Data packets get rerouted automatically, and the two devices keep talking.

Picture two offices connected by exactly one cable. If a backhoe cuts it, every message between the offices stops cold until someone fixes it. Now add a second, independent cable running a different physical route. Cut one, and traffic just flows over the other with maybe a brief delay. That second path is the redundancy. Add a third and fourth spare path and the network can survive multiple simultaneous failures, not just one, though each extra path adds cost for a shrinking amount of extra protection, since all the spares failing at once gets less and less likely.

This sits next to fault tolerance as cause and effect: redundancy is the design choice, fault tolerance is the property that results from it. It's easy to confuse redundancy with backing up data, but they guard against completely different disasters. A backup is a copy of information, stored somewhere else, so you don't lose the data itself. Path redundancy is about the connection between two points staying open, so traffic keeps moving even when a link breaks. You can lose data with perfect path redundancy, and you can have perfect backups while your network is completely cut off.