is prohibited.
A better solution would be selecting evenly (fairly) the segments to be dropped from both
contending data bursts. Likewise, the truncated burst size should be monitored at the core
nodes, and guaranteed to be larger than the Minimum Burst Length (MBL), which is the
minimum length allowed into the network to avoid congestion in the control channels.
Besides fairness, data burst size, and implementation simplicity, any proposed technique
should be designed to deal with the switching time, which is the time needed to configure
the switching fabric (i.e., to switch an output port from one DB to another). To understand
what follows, the following definitions are provided and illustrated in Figure 6.
??? DBO: Original Data Burst with Arrival time TOA and Leaving time TOL ,
??? DBC : Contending Data Burst with Arrival time TCA and Leaving time TCL ,
??? TDB: Truncated Data Burst (i.e., a DB with dropped segments),
??? N, M: Respectively, the number of segments in DBO, DBC,
??? DS: Data Segment with Length DSL,
??? R: the expected number of segments to be dropped from each data burst,
??????
???
??????
??® ??’
=
DSL
T T
R OL CA
2
.
When data bursts arrive to a core node, the technique performs three functions arranged in
three main events.
Pages:
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792