STP / Spanning Tree Basics
Definition
The Spanning Tree Protocol (STP, IEEE 802.1D) builds a loop-free logical topology over a physically redundant switched network by blocking backup ports until an active link fails. Classic STP relies on timers, by default a maximum age of 20 s and two forward delay periods of 15 s, and takes 30 to 50 seconds to converge, while RSTP (IEEE 802.1w) usually converges within a few seconds or less. Without spanning tree, one cable connected between two ports of the same network forms a loop in which broadcast frames circulate indefinitely, and the resulting broadcast storm can saturate the network and stop audio, video and control together.
In the field
A frequent cause of show outages is a spare cable patched between two live switch ports, which closes a loop. With RSTP enabled on every managed switch, the loop is blocked; without it, the broadcast storm stops the network within seconds. Many switches also offer loop detection or BPDU protection on edge ports, and spare cables are labelled and kept unplugged.
Sources
- Spanning Tree Protocol (Wikipedia, pointeur secondaire) : STP jusqu'à 30 à 50 s pour réagir à un changement de topologie, RSTP en quelques secondes ou moins · en.wikipedia.org
- Cisco, Understand and Tune Spanning Tree Protocol Timers, valeurs par défaut : hello time 2 s, max age 20 s, forward delay 15 s ; temporisations imposées par le pont racine · cisco.com
- NetworkAcademy.IO, Introduction to Rapid PVST+ : STP 30 à 50 s contre quelques secondes ou moins pour le RSTP · networkacademy.io
- Jochen Schulz, Dante: Network Topologies : anneau seulement avec spanning tree activé · jochenschulz.me