FDDI Architecture and Operation
Introduction
Token Ring (that is, IBM and IEEE 802.5) had a period of success and popularity, followed by a fall. FDDI (Fiber Distributed Data Inteface) had a period of success far more modest than that of Token Ring, followed by practical oblivion. If there was ever a proof that the "best man" doesn't always win, FDDI would be a good part of it. Not that FDDI is necessarily "the best man" but it sure is a clever technology, improving on the already clever concept of a token-passing ring.
So, just to remind you of why I think that it is worth using a few cyberbits to talk a bit about FDDI, even though it fell into the black hole of underused technologies, it is because the concepts are useful for understanding network technologies, and because sometimes people still come across it and they need some way to find out what it is.
I'll reiterate what I stated in the introduction to the token ring technologies portion of this portal. FDDI came on the scene several years after Token Ring, and it pushed the concept of a LAN as a local network by offering a network that could be as large as 200 kilometers (124 miles) in diameter and could connect thousands of stations. It was sometimes referred to by the term MAN (Metropolitan Area Network), and could clearly be used on large campuses and beyond. It also could serve as a backbone network to connect LANs. Developed by the ANSI X3T9.5 committee in the mid-1980s, it offered far greater transmission speed (100 Mbps) than was offered by either Ethernet or Token Ring at that time. Further, FDDI offered impressive architectural resiliency and reliability. The ANSI standard was later internationalized as ISO 9314.
You learned that in a Token Ring, each station serves as a repeater. Further, you saw that to prevent a failed station from "breaking" the ring, an electro-mechanical switch in the MSAU bypasses the port of the failed (or inactive) station. Further, you saw that the IBM Token Ring provides "dual ring redundancy". However, this is an IBM feature, but not part of the IEEE standard.
In an FDDI ring, two motivating factors resulted in resiliency being built into the standard. First of all, FDDI was intended to span much greater distances than a Token Ring. Further, as we shall see below, a station in the ring is less likely to be merely an end station, which if it fails but the ring is closed affects no one else. Rather, stations are typically concentrators or critical nodes whose failure not only breaks the ring but disconnects a whole group of stations that are served by it. The resiliency offered by the FDDI topology provides solutions, similar to those you saw with Token Ring, but there are added ones, and, unlike Token Ring (where IBM offered resiliency features that were not part of the IEEE standard), they are part of the standard.
Topology
FDDI uses a Dual-Ring topology. That means that it is comprised of two counter-rotating rings, which means that traffic on each ring flows in the opposite direction of the other, depicted in Figure 36. The A and B in each station represent two ports, which will be explained in the section about station types. The rings are referred to as Primary and Secondary.
Figure 36: FDDI Counter-Rotating Rings
During normal operation, the transmission of data is over the Primary ring, while the Secondary ring remains idle. The Secondary ring's role is to serve as backup, in case the Primary ring fails or in case a station fails (or is removed from the ring/turned off). However, a station on the ring could be connected to the ring via an Optical Bypass Switch (OBS), so that even if the station is turned off or fails, the Primary ring is left intact. An OBS can detect if the attached station is not functioning, in which case the OBS diverts the optical path away from the device. (The Secondary ring potentially can be used as an additional transmission path.)
The way this works is demonstrated by a terrific applet (though with several minor bugs), by Stephan Willmanns, that I found on the Internet. The first little problem with the applet is that the "help" button doesn't work, but actually the applet is pretty self-explanatory, so that is really a minor problem. However, I will explain how to work it, anyway. And, if you read German, then you can look at Stephan's thesis, for which this applet was written some time ago. (If you are interested in other information about this applet, such as the Java documentation, go to this page, where you will find some other interesting applets, as well. They are listed alphabetically, so just look for "FDDI" to find this one.)
Here's how it works. Figure 37 shows what it looks like when you start it by clicking on the Start animation button.
Figure 37: Start FDDI Applet
Oh yeah, another tiny problem I found - the Options menu offers the opportunity to change the speed of the simulation, but it seems you have to be able to operate in nanoseconds to be able to actually set this field. I didn't succeed. But I said it is a tiny problem, because the applet works nicely just as is.
The Add button allows you to add more stations to the ring. If you try it, you'll see that you get another station each time you click on it, and a message in the status box.
The Remove button allows you to remove stations, but you can't have fewer than two (and a message in the status box will let you know).
If you click on the Start button, the token will start circulating on the Primary ring. The Pause button will stop the simulation, and the Continue will resume it.
That's simple enough, but you don't just want to watch a token circulate on the ring. Notice that the dropdown menu at the bottom left of the applet says "send". If you click on one of the stations, it will be selected as the station that will send a message, and if you click on another station, it will be selected as the receiving station, as shown in Figure 38.
Figure 38: Message Sent in FDDI Applet
The sending station is indicated with a rectangle around it; the receiving station has an oval around it. Now you should notice another little bug, which is that the stations are supposed to be numbered (as you can see if you look at the figures in Stephan's thesis), but the applet displays no contrast in color and you can't see the numbers. You'll have to use your imagination (or pay attention as the status box states the station number, and then play a memory game with yourself).
The data is represented by solid (filled-in) circles (different from the color of the token), and when the data is received by the destination, the circles have no fill. Once the already read data returns to the sender, it is stripped by the sender, and a token continues to rotate on the ring. You can actually select multiple pairs of send/receive stations, since Early Token Release is a feature of FDDI. I won't clutter this page with screen captures of these events - try it for yourself. Make sure to see how multiple stations send and receive.
Now we get to the more interesting part. The drop-down menu in the bottom left of the applet has two more options: off/on and broken ring. The off/on selection allows you to turn a station off by selecting the station you want to turn off. The turned-off station will be indicated (and the status box will state which station it is, and you can add to the station numbers you've been keeping in your head). But the ring operation is not affected in the animation - we can assume that this must be because the connected stations all have Optical Bypass Units, but I do think this should been shown in the figure, because OBUs are not obligatory, and if a station is connected without one, and that station fails or is turned off, then we would see the same reaction as we do with a cable break.
.So, go take a look what happens if you choose the broken ring option in the drop-down menu on the bottom left, as captured in Figure 39.
Figure 39: Broken Ring in FDDI Applet
You can only see how the data path changes if you actually try the applet yourself. So go ahead, it is cool! Once a cable (or device) is repaired, FDDI automatically reverts to operating on the Primary ring, with the Secondary ring serving as a backup (and you can remove the cable break in the animation and see that the ring operates normally, again).
Station Types
Several ways were defined to connect devices to an FDDI network. The way in which a station is attached defines the station type. There are four such types:
Dual-attached station (DAS) (Figure 40) - connected to both the primary and the secondary rings. Dual-attached stations have two ports: A (primary-in, secondary-out) and B (primary-out, secondary-in). (Now you understand the A and B in Figure 36, above.) Unless the station is connected with an OBS, disconnection of the station (or power off) breaks the ring, triggering the transmission path to wrap around and use the secondary ring.
Figure 40: Dual-Attached Station
Dual-attached concentrator (DAC) (Figure 41) - a dual-attached station that is used to connect additional stations. This device can be considered as a root of a tree, since it provides connection for additional stations and concentrators. The ports used to connect additional stations are called M ports (a station with one or more M port is called a concentrator).
Figure 41: Dual-Attached Concentrator
Single-attached station (SAS) (Figure 42) - connected only to the primary ring, through a concentrator. Only a single port is needed. When the device is disconnected or powered off it does not have a negative effect on the ring.
Figure 42: Single-Attached Station
Single-attached concentrator (SAC) - a concentrator that is connected to the primary ring only (through a tree).
The MAC that you see in the diagrams is the same acronym Medium Access Control that you are familiar with from the IEEE protocol layers. It represents the implementation in the equipment that is responsible for handling medium access, frame format, token handling, addressing, and CRC algorithms.
With the different station types, we have the building blocks of an FDDI networks, which can be described as a "ring of trees", as depicted in Figure 43.
Figure 43: Ring of Trees
The use of concentrators permits a station to be connected using only a single pair of fibers, resulting in lower connection costs. Further, connecting one DAC with multiple M ports for SASs assures a smaller chance of network failure due to the malfunction of a station, or frequent switching on and off of a station, as would be the case with PCs.
A Word about the Transmission Media
The primary medium for FDDI is Optical Fiber (see the last paragraph in the page called "optic" to find out about FDDI's fiber specifications). A version that operates over copper wires was defined, and was called CDDI.
