For maximum scalability, the B-components know nothing of individual EVCs. The
B-components distribute frames only based on the outer MAC addresses and B-tags; they
do not forward frames based on the I-tags. The I-tags are used only by the I-components.
Hiding per-EVC information allows a network of backbone bridges to support a very
large number of EVCs with a minimum of internal state.
Using Spanning Tree Effectively
Although the spanning tree protocols have deficiencies, as addressed in the following
section, many of the complaints about spanning tree protocols stem from ignorance of
its current development state and/or poor implementations, rather than from its real
flaws. Most commonly, the complaint is either, ???There is only one path??”everything has
to go the long way around, through the root of the tree???; ???It takes forever to converge
after a failure???; or ???It floods frames with unknown unicast addresses everywhere in the
network.??? All such problems are solvable by current bridge technology (and/or are, in
fact, not problems).
400 Chapter 13
A Single Path for the Customer??™s Data Imagine for a moment an EVC connecting eight
sites, spread across a continent, providing the equivalent of a single shared medium
to the customer.
Pages:
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901