Let??™s suppose that four customers, all with this same SLA, are being carried
over a single 100 Mbps LAN. The 60 Mbps total of guaranteed traffic is perhaps
sustainable. Suppose, however, three of the customers are blasting 50 Mbps, and the
fourth only the guaranteed 15 Mbps. The offered load is then 165 Mbps on a 100 Mbps
LAN. Certainly, the provider does not want the excess traffic from the three customers
to cause the remaining customer to lose frames.
If the bridge transmitting to this LAN simply marked, for each customer, all frames
in excess of 15 Mbps to a lower priority (say, from 5 to 4), and if those two priority levels
went into different queues (so the frames conformant to the SLA could be transmitted
in preference to those in the 15??“50 Mbps range), there would be a problem. Suppose a
customer transmits first frame A and then frame B. Suppose the bridge decides that
frame A is in excess of the 15 Mbps limit and marks it with priority 4, but frame B is
OK and marks it with priority 5. Then, frame B could be delivered to its destination
before frame A. Using this example, this would happen very frequently.
Provider bridges, therefore, have a Drop Eligible Indicator (DEI) in the S-tag and a
definition of rate policing that supports the DEI.
Pages:
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875