Spanning Tree Protocol Entity (STPE)
Suppose all bridges come up at the same time. Initially, at t=0, all bridges think they are the root and they issue the messages shown on the t=0 line. For simplicity, the bridge ID is a two digit number and the couplet such as 01,3561 should be interpreted as: root_ID, cost_to_root. The messages sent to the LAN attached to B1 and B4 are irrelevant to this discussion and are not shown.
At the end of this first iteration, after the bridges have compared messages sent and received on the various links, they conclude the following:
•B1 will continue to see itself as the root bridge. Its link 5 and link 1 (to the LAN, not shown) will be set as designated links.
•B2 will continue to see itself as the root bridge temporarily. Its link 1 and link 6 will be set as designated links.
•B3 will determine B1 as the root bridge. Its link 5 is its root link and its link 1 is a designated link.
•B4 will temporarily see B2 as the root bridge. Its link 6 is its root link and its link 1 (not shown) is a designated link.
At t = Hello Time, the root bridges issue the spanning tree hello message. After the bridges have compared the new messages, they form the following conclusion:
•B1 will continue to see itself as the root bridge. Its link 5 and link 1 will be set as designated links.
•B2 will determine B1 as the root bridge. Its link 1 will be the root link and link 6 will be set as a designated link.
•B3 will determine B1 as the root bridge. Its link 5 is its root link and its link 1 is a designated link.
•B4 will temporarily still see B2 as the root bridge. Its link 6 is its root link and its link 1 is a designated link.
At t = 2 x Hello Time, there is only one root bridge B1 to issue the spanning tree hello message. This message is passed along the tree and, at this iteration, B4 finally sees B1 as the root bridge, and the spanning tree has converged.
From this example, it is seen that the convergence time is dependent on how many link hops the farthest bridge is from the root bridge and how frequently the hello message is sent. Until the bridges converge, no bridge should learn station locations (nor forward data frames); therefore, the forward delay (the time the bridge should wait before learning), should be set to a value that might be at least as high as the hello time multiplied by the network diameter. With special consideration this time might be considerably reduced. For example, in the preceding network, if B2 and B3 are interchanged, the spanning tree will converge at t = Hello Time (half the time of the original network).
Note
The Forward Delay time is used a second time to allow the bridge to learn station locations before allowing the bridge to forward frames.
When the spanning tree is configured for manual operation, no spanning tree protocol needs to converge, but the station locations still must be learned.
For manual operation, a different timer is used only for setting the Learn Only Period. This timer is used by the bridge to set the time it will learn only after the bridge has booted.
98 | Bridging |
