Fast Ethernet vs FDDI

Abstract:

In the following, we compare the two competing LAN technologies, Fast Ethernet and FDDI [1][2]. Both of them are originated from the two longtime rival technologies, Ethernet and Token Ring respectively [1][2][3]. In addition, both approaches are technically sound and being used successfully in the real world with its technical advantages and disadvantages. But it seems to us that Fast Ethernet probably is the better one because it is technologically as competitive, relatively simpler, easier to maintain and more cost effective.

Background on Ethernet and Token Ring

Ethernet (IEEE 802.3) stands for a 1-persistent CSMA/CD LAN. The basic idea here is that a host listens to the cable first when it wants to transmit. If the cable is busy, the host waits until it goes idle; otherwise it transmits immediately. If two or more hosts simultaneously begin transmitting on an idle cable, they will collide. All colliding hosts then terminate their transmission, wait a random time, and repeat the whole process all over again. The protocol is simple and hosts can be installed on the fly without taking the whole network down. On the other hand, to know for sure that the frame it just sent did not collide with any other frame, the transmitter need to send a minimum of 64 bytes in the frame. This could represent a substantial overhead. Other disadvantages of Ethernet include nondeterministic, priorityless, and less efficient as the speed increases.

Token Ring (IEEE 802.5) is deterministic star-wired ring architecture, and the sequence in which users gain access is predetermined. The controlling station, called the active monitor, generates a special signaling sequence called a Token that controls the right to transmit. This Token is continually passed around the network from one node to the next. When a host has something to send, it captures the Token, and changes it to a frame, setting its status to busy. It then adds the frame header, information and trailer fields. The header includes the address of the hosts that will copy the frame. All nodes read the frame as it is passed around the ring to determine if they are the recipient of a message. If they are, they extract the data, retransmitting the frame to the next host on the ring. When the frame returns to the originating station, it removes the frame and reissues a free token which can then be used by another host. The token-access control scheme thus allows all hosts to share the network bandwidth in an orderly and efficient manner. The advantages of Token Ring, in addition to deterministic, are excellent throughput and efficiency at high load. The major minus point is the presence of a centralized monitor function, which includes a critical component. Another disadvantage is that the ring is broken whenever one host is down or the cable breaks.

Fast Ethernet and FDDI

Fast Ethernet

Fast Ethernet (100Base-T) is an extension of IEEE802.3 CSMA/CD. Its LAN has a star topology connection for up to 210 meters in diameter using either UTP or fiber. Fast Ethernet use the same MAC layer as 10Base-T Ethernet, so Ethernet users can easily migrate to Fast Ethernet. Applications and higher level protocols developed on 10Base-T Ethernet will run on 100Base-T Fast Ethernet without modification, and 100Base-T adaptors are available to switch between the wide used 10 Mbps ethernet and the 100 Mbps standard. In addition to having the advantages of Ethernet, Fast Ethernet has a 100 Mbps throughput make it competitive with FDDI networks. The limit for a Ethernet is 2500m and 1024 hosts with with no more than 4 repeaters between each host pair. The maximum segment length for Fast Ethernet is only 100m, unless Fiber optics is used which could be as far as 2000m.

FDDI

FDDI (Fiber Distributed Data Interface) is an advance type of Token Ring. An FDDI network consists two independent rings that transmit data in the opposite direction, so that it is able to tolerate a single break in the cable. Each host implements a small elasticity buffer that temporarily holds the bits of a frame as they pass through the host, so that frames are able to be sent over the ring without all the hosts having to be synchronized. FDDI also allows SAS (single attachment stations) in the configuration of the ring. These improvements for FDDI give it a 100 Mbps throughput. The limit for a FDDI is 100km and 1000 hosts. FDDI can transmit a frame up to 4500 bytes, which about 3 times as much as Ethernet can do.

Comparing Fast Ethernet and FDDI

Both Fast Ethernet and FDDI are high speed LAN technologies, growing from Ethernet and Token Ring respectively.

In the following, we will compare the two technologies in terms of throughput, latency, deterministic, configuration, maintenance, compatibility, reliability, cost, user community, and interconnectivity.

throughput - On shared media LANs such as Fast Ethernet and FDDI, line speed decreases in proportion to the number of nodes contending for a portion of the total available bandwidth. Now assume that same amount of hosts on the LAN, even thought Fast Ethernet and FDDI have the same 100 Mbps throughput, FDDI will ran faster than Fast Ethernet because there collision detection could reduce the bandwidth by 30-50%. Result from DATA COMM Test Lab [4] shows that FDDI runs at 96 Mbps (96% efficiency).

latency - FDDI has much higher latency since data frames has to be passed by many hosts in between. For Ethernet, the latency is relative lower, although it is nondeterministic based on random wait after detected collision. The test results [5] show that the latency for FDDI products is about twice as much as their Fast Ethernet comparable counterpart.

deterministic - FDDI is deterministic. But Ethernet is not, because of the wait for a random time after detecting a collision.

configuration and maintenance - It is less complicated with Ethernet LAN, because of its simple structure. Adding one more host on Ethernet is much simpler than do it on FDDI. Ethernet protocol is simple and hosts can be installed on the fly without taking the whole network down.

distance - Because of the distance limit, (100m for twisted pair and 2000m for fiber optics), on Fast Ethernet segment, FDDI is clearly the better choice with a distance up to 100km.

host load - Both Fast Ethernet and FDDI can accept up to about 1000 hosts. For FDDI, more hosts means potential longer latency. For Fast Ethernet, more hosts mean more collision and more congestion. Even worse for higher than 60% loading, the overall throughput of Ethernet could be stalled, because hosts will be busy with detecting collision/waiting, and thus are not able to transmit.

compatibility - Ethernet users can easily migrate to Fast Ethernet. Applications and higher level protocols developed on 10Base-T Ethernet will run on 100Base-T Fast Ethernet without modification, and 100Base-T adaptors are available to switch between the wide used 10 Mbps ethernet and the 100 Mbps standard.

reliability - Ethernet is simple, has less things to break, and thus is pretty reliable. For FDDI, the whole LAN could be brought down if one or more hosts in the ring break. The dual ring is only able to tolerate a single break in the cable.

cost - Fast Ethernet although does not have the technical edge over FDDI in terms of speed, it is much more easier to configurate and maintain than FDDI. In addition to that, Fast Ethernet products cost a fraction to what would be for its FDDI counterparts.

users - Ethernet has more users. One of the most important obstacles to the installation of high-performance networks are users failure to accept the new technology. In the case of Ethernet a large community of users were convinced to install a commonly agreed type of high performance network because manufacturers are able to provide standard add on features, which as they are produced in bulk, have been offer cheaper.

interconnectivity - European Research Lab [6] found out that when FDDI and Fast Ethernet are bridged, only 50 Mbps can be handled before exhibit heavy packets loss in their Client-Server experiment. In contrast, Ethernet switches deliver dedicated bandwidth.

Applications

At current 100 Mbps level, FDDI is more efficient with its high bandwidth, so it is better suited for applications requiring a higher bandwidth. In the other hand, FDDI has a longer latency than Ethernet. Therefore, for latency sensitive applications, Fast Ethernet is the better choice.

Conclusion

In the last section, we compare Fast Ethernet and FDDI in details for both technical and non-technical issues. Here is the break-down summary for the two competing technologies, where a `X' indicates that the corresponding technology is a relative winner in that category.

Fast Ethernet improves a lot from 10 Mbps Ethernet. Now 100 Mbps Fast Ethernet is able to be competitive to FDDI in throughput. Newer Fast Ethernet at a speed of 1000 Mbps (1 Gbps) is coming available. In addition to that, Fast Ethernet is more cost effective and already a larger user community. So we believe that Fast Ethernet is the wining technology over FDDI.

Future - Will FDDI be killed by Fast Ethernet?

The simpleness and cost effective has been a great market drive for Ethernet and recently Fast Ethernet. One of the most important obstacles to the installation of high-performance networks are users failure to accept the new technology. In the case of Ethernet a large community of users were convinced to install a commonly agreed type of high performance network because manufacturers are able to provide standard add on features, which as they are produced in bulk, have been offer cheaper. For FDDI, there is no such a big luck. The FDDI adaptor units are more expensive and the FDDI networks are somehow in danger of dying away if better performing and cheaper network adaptors for other technologies, like Fast Ethernet, are offered in the future.

Reference

[1]. Computer Network, Larry Peterson and Bruce Davis, Morgan Kaufmann Publishers, 1996.
[2]. Computer Network, Andrew Tanenbaum, Prentice Hall PTR, 1996.
[3]. Emerging high speed LANs: a process control perspective, Dennis Tkacs, ISA Transactions 35, pp 121-125, 1996.
[4]. Choosing High-Speed LANs, Stephen Saunders, Data Communications, volume 35, no. 2 pp 121-125, 1996.
[5]. Canned Head - Ethernet, FDDI and fast Ethernet switch evaluation, Robert Mandeville and David Newman, volume 25, no. 2, pp 80-98, Feb., 1996.
[6]. Client-Server Challenge: FDDI or Ethernet Switches?, Robert Mandeville, Data Communications, volume 23, no. 7, pp 107-113, May 1994.