This can, in many bridges, take
some time to accomplish. The net effect is that in large networks, especially those with
ring topologies, the Spanning Tree Protocols can often take more than 1 sec to converge
after a failure or a recovery. This is adequate for most customers, but not for all.
The most serious problem with the spanning tree algorithms as currently standardized
and implemented is the potential for the creation of permanent forwarding loops.
When loops occur, bridged frames can circulate forever; each multicast frame, in particular,
spews out a copy of itself each time it traverses the loop. If the Spanning Tree
Protocol is executed correctly, this is an impossible occurrence. However, if a software
bug or hardware failure results in a bridge that sinks, but does not source, the bridge
protocol data units (BPDUs) that carry the Spanning Tree Protocol, while allowing that
bridge to continue forwarding data frames, also create a permanent loop.
These drawbacks to the spanning tree algorithms are well-known in the networking
community, and as a result, the spanning tree algorithms have acquired a bad reputation.
This is compounded by the lack of good information on these protocols in the available
literature (a lack that this chapter is intended to address).
Pages:
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891