Worth Reading: Faster than Dijkstra?
Bruce Davie's article critiques the push for faster-than-Dijkstra algorithms in link-state routing, highlighting that Dijkstra's algorithm efficiently handles large networks, as demonstrated by its performance on a 2000-node network in 2003.
MAIN POINTS
- Dijkstra's algorithm is efficient for large-scale networks.
- A 2003 test showed Dijkstra's algorithm handling 2000 nodes in 100 milliseconds.
- Calls for new routing protocols often stem from poor implementations.
- Enthusiasts sometimes push for new technologies without practical need.
TAKEAWAYS
- Dijkstra's algorithm remains effective for current network sizes.
- Performance tests from 2003 still hold relevance today.
- New routing protocols may not be necessary with proper implementations.
- Technological advancements should be driven by necessity, not novelty.