FDDI stands for Fiber Distributed Data Interface. In a nutshell, FDDI is a 100 Mbps LAN technology which can run over fiber optic or copper cable. It is the oldest 100 Mbps network type commonly available, and is widely used as a backbone technology to interconnect several smaller Ethernet or Token Ring networks. It is also used whenever high reliability and/or high speed are required for a specific application.
FDDI uses three basic topologies: Ring, Star, and Tree. These topologies can be combined to build large networks (up to 500 nodes) which use the advantages of each while avoiding each topology's drawbacks.
A definition of each topology is:
Figure One (below) graphically illustrates the different topologies.
As mentioned earlier, it is possible to combine these topologies in one network. For example, assume we have three buildings. In each building we install an FDDI network wired in a tree topology. We can interconnect the buildings in a ring topology, thereby creating a "ring of trees" network.
There are four cable types which can be used with FDDI. They are:
The variety of cable types allows one to design an FDDI network which takes advantage of the strengths of each type in different parts of the network. For example, in areas where long distances are needed, fiber optic cable is commonly used. For areas where distances are fairly short, less expensive Category 5 cable can be used. This gives a lot of flexibility in designing an FDDI network.
Regardless of cable type, the maximum overall logical ring length of an FDDI network can not exceed 200,000 meters (660,000 feet). It is strongly reccomended to keep the actual ring length below 100,000 meters to allow for situations where the primary ring is "wrapped" around a break. Wrapping basically doubles the length of the ring. Also, there may be no more than 500 nodes on one ring.
At the lowest layer, FDDI creates a network comprised of two rings interconnecting all of the nodes on the network. Each ring transmits data in a direction opposite to the other one. These rings are logical in nature, and exist regardless of how the network is physically connected together.
The reason for having two rings is fault tolerance. Most of the time, the primary ring carries the data and the secondary ring is idle. In the event of a break in the ring, the nodes nearest the break will loop the primary ring to the secondary ring, which bypasses the fault and results in an unbroken ring.
FDDI also offers the ability to use both rings for data transmission at the same time. This feature boosts the network speed to 200 Mbps. In the event of a fault, the secondary ring will revert to its previous function, and the overall network speed will drop to 100 Mbps.
FDDI uses a token passing protocol which is similar, but not identical to, Token Ring. In such an arrangement a special type of packet called a token is sent around the network. Any node which wishes to transmit data to the network first captures the token, sends a packet of data to the network, then it releases the token. Every station on the network will recieve the transmission and repeat it. If a station recieves a transmission addressed to it, it will mark the transmission as recieved and repeat it to the network. The transmission will travel around the ring until it is recieved by the station which originally sent it, which removes it from the ring. If a station does not recieve its transmission back, it assumes that an error occured somewhere.
There are two main ways of interconnecting nodes in an FDDI network. The first way is calles Single Attachment Station, or SAS for short. The other is called Dual Attached Station, or DAS.
In a DAS network, it is not necessary to use concentrators, although it is a good idea to do so. Concentrators provide a measure of protection by providing a wrapping function. A DAS network without concentrators can survive one break in ring integrity, with a second break isolating at least one and possibly many nodes. A Dual Homed connected DAS network has a considerable amount of redundancy built in, and is very unlikely to fail.
Many FDDI networks are used with a combination of SAS and DAS. For example, a network designer may use SAS for connecting individual workstations to the FDDI concentrators, as SAS is considerably less expensive to implement, and the loss of a workstation will not normally be a significant problem for the overall network. However, the file servers and inter-concentrator links are extremely critical. The loss of one of these could result in a large group of users being isolated from critical data they need to do their jobs. Therefore, these links require the utmost reliability the designer can provide regardless of cost, so dual-homing DAS would be used here.
Like any technology, there are FDDI has strengths and weaknesses. The major ones are shown below:
Most network users do not require the high speed (and associated high cost) of a full-blown FDDI network. However, sometimes an organization will have such a large number of users that the aggregate bandwidth needed is far more than a 10 Mbps Ethernet or 16 Mbps Token Ring network can provide. One solution for this type of problem is to employ a combination of Switching and FDDI technologies to create a high speed backbone interconnecting a large number of small Ethernet or Token Ring networks. Figure Two shows this type of a network.
"Dual Homed" FDDI Backbone With Ethernet Switching
In Figure Two, we have two file servers connected via FDDI to two FDDI concentrators. Note that each server has a DAS FDDI card installed in it, and each port on each card is attached to a different concentrator. This provides fault tolerance, as neither server can be knocked off the network by a failure of either concentrator. There is also a 10/100 Ethernet switch connected to the concentrators via a Dual Homed DAS connection. The switch will feed multiple small 10 Base-T networks.
A network of this type allows taking advantage of the speed and reliability of FDDI where it is needed (at the servers) while protecting an existing investment in 10 Base-T technology. None of the workstations on the network need to be upgraded at all to take advantage of the increased aggregate bandwidth available. We simply split them in to smaller groups, with each group having 10 Mbps of speed dedicated for its exclusive use. If we have five groups, then there will be an aggregate of roughly 50 Mbps of bandwidth available where there was previously only 10 Mbps.
Note that this type of solution will eliminate network slowness due to congestion, or too many nodes attempting to use 10 Mbps of bandwidth at once. If the application demands more than 10 Mbps at any workstation (for example, huge CAD files), then simply put an FDDI card in that workstation and connect it directly to one or both FDDI concentrators. This feature allows balancing cost with the actual need for speed and security of each station on the network.
A very in-depth FDDI tutorial can be found at the University Of Newhampshire IOL's FDDI Tutorial page.