Expected CCR Ethernet Performance

Hughes Aircraft Company; Aurora, CO

This modeling effort suggests that the local availability of basic validated simulation models of the components of distributed systems, such as Ethernet LANs, token ring LANs, file servers, workstations, and host computers, would allow the building of application specific simulation models, much more quickly than is the norm, by modifying and extending these basic models. This is due to the fact that the central functions need not be changed once the basic model has been developed, verified, and validated.

The CCR Ethernet Model is used to simulate the behavior of the CCR Ethernet under three different traffic scenarios. As a result of this "traffic tuning", the average and the minimum channel utilization each dropped by about 59%. While the peak utilization remained about constant, the period of peak utilization dropped by about 77%. On the other hand, the average frame delay dropped by about 79% while maintaining a constant relative statistical uncertainty (coefficient of variation). The peak frame delay only dropped by about 14%, but again the period of high frame delay (200 to 300 ms.) dropped by about 77%. In addition, the number of collisions and the number of "dropped" frames (dropped by Ethernet for exceeding 16 consecutive collisions) decreased by about 90%. At the same time, the peak occupancy of the defer queue (those frames waiting for a channel to become idle) dropped by about 79% and the average occupancy dropped by at least 50%.

Alternatives Considered: C, SLAM II

Benefits of SIMSCRIPT II.5: SIMSCRIPT II.5 is a free-form English-like general-purpose simulation language which supports both the "event view" and the "process view" of discrete event simulation.