Comparing Contention and Token Passing

Comparing Contention and Token Passing

As an access control mechanism, token passing appears to be clearly superior to contention. You find, however, that Ethernet, by far the dominant LAN standard, has achieved its prominence while firmly wedded to contention access control.

Token passing requires a variety of complex control mechanisms for it to work well. The necessary hardware is considerably more expensive than the hardware required to implement the much simpler contention mechanisms. The higher cost of token passing networks is difficult to justify unless the special features are required.

Because token-passing networks are designed for high reliability, building network diagnostic and troubleshooting capabilities into the network hardware is common. These capabilities increase the cost of token-passing networks. Organizations must decide whether this additional reliability is worth the extra cost.

Conversely, although token-passing networks perform better than contention-based networks when traffic levels are high, contention networks exhibit superior performance under lighter loading conditions. Passing the token around (and other maintenance operations) eats into the available bandwidth. As a result, a 10 Mbps Ethernet and a 16 Mbps Token Ring perform comparably well under light loading conditions, but the Ethernet costs considerably less.

Figure 4.3 illustrates the performance characteristics you can expect from each access control method. (This figure implies that token-passing throughput eventually reaches a zero level, which cannot happen, regardless of the loading conditions. Although a station’s access to the network might be limited, access is guaranteed with each circuit of the token.)