The mechanisms and procedures that determine and select paths through a network so that packets or packets-equivalents are forwarded from a source toward a destination, typically by computing next hops, maintaining reachability information, and installing forwarding state in network elements.

Principle

Routing distributes and aggregates reachability information and applies path-selection rules so that local forwarding decisions collectively produce end-to-end packet delivery subject to trade-offs among optimality, scalability and convergence.

Demonstration

Illustrative scenario: A set of routers exchange reachability updates, each computes preferred next hops using a metric (e.g., cost), installs forwarding entries, and thereafter forwards packets toward the chosen next hop until the destination is reached.

Misapplication

Assuming routing guarantees in-order delivery, zero loss, or global optimality. The error is treating path selection as a promise of perfect delivery rather than as a best-effort decision subject to topology, policy, and transient failure.

Consequence

Correct routing computation and distribution enable reachability and efficient use of links; incorrect or inconsistent routing can create loops, blackholes, suboptimal paths, increased latency, or route oscillation requiring mitigation or reconvergence.

Reversal

In networks where endpoints perform source routing, overlay routing, or where circuits are pre-established, traditional distributed routing may be bypassed or its role reduced, shifting path selection to endpoints or control-plane provisioning.

Boundary

Clearly within: distributed algorithms that compute next hops (e.g., link-state or distance-vector approaches). Boundary case: traffic engineering systems that compute constrained paths but rely on routing to carry reachability. Clearly outside: per-hop forwarding hardware that applies installed forwarding state without computing paths.

Semantic Tension

Optimality ↔ Scalability — striving for globally optimal routes increases state and computation, while limiting state improves scalability but may accept suboptimal paths.

Synthesis

Routing is the distributed decision process that trades global path optimality and responsiveness against scalability and policy: it produces forwarding state so many autonomous devices cooperatively deliver packets despite topology changes and administrative constraints.