What's the difference between Ethernet and Token Ring in terms of performance at various traffic levels?

Answer 1

The following criteria are used to determine the Pros and Cons of Ethernet and Token Ring:-
Ability for trouble shooting
Token Ring is better than Ethernet. Test procedures are invoked every time a systems first accesses the ring. If one of the procedures fails, messages are generated to indicate the possible causes of failure, including suspicious addresses. In ethernet, however, problems are identified through a series of isolated procedures.
Easier Installation
Ethernet is easier to install because of the flexibility to mix and match different kinds of topologies ( hub or star, bus, and tree ) makes it extremely easy to implement an Ethernet network of any size. On the other hand, Token Ring requires a hub where connections are made to the main Multi-station Access Unit with any other system or hub.
Multi-vendor Environment
Ethernet is the choice because most non-IBM systems support it as the primary LAN. Examples are Digital Euqipment, Hewlett Packard, and Sun Microsystems.
Ability to support more users
The upper limit for an isolated LAN is around 200 systems. Small LANS of both Ethernet and Token Ring can be established and then interconnected.
When there are frequent accesses to the network
Ethernet is less suitable when many users have to access information through the network. An example would be a travel reservation company in which the agents need to retrieve data across a LAN. They may check whether flights and hotel rooms are available at first, then proceed to make bookings. These "frequent calls" to the network will lead to fluctuation of levels of performance. To remedy this problem, the Ethernet can be divided into segments.

Answer 2

Depends on different factors. First off Token Ring is more or less dead from a market perspective. During TR's hay day, Ethernet maxed out at 10Mbps (10Base?) and used only shared access, i.e. hubs. When used in the shared CDMA environment Ethernet is not efficient at all, the best you can hope for is 30% utilization. Once the traffic level gets to high there are so many collisions that it become unusable. Also at this time, Ethernet was half duplex, meaning you could only send or receive not both at the same time. (T1 circuits - for example, both send and receive at the same time 1.5Mb up, 1.5Mb down). Ethernet has since evolved to use switching technology where only the two devices trying to communicate with each other are "connected" to each other. (somewhat meaningless if all workstations are trying to get to the same server/router), and switched Ethernet also supports full duplex, meaning you can receive and transmit at the same time. In this type of topology the utilization is closer to 100%. Ethernet has also evolved to 100Mbps, 1Gbps (current commonly found port) and 10Gbps (commonly used for trunks). Token Ring, topped out at 16Mbps. TR is highly efficient, because every station needs permission to send data on the ring. They can't arbitrarily send packets like Ethernet. So if you go back to 1995, TR was the more efficient. If you get a switched Ethernet network running full duplex they are equally efficient. Ethernet is by far and away the cheapest and easiest to set up.