Token Ring

 

UNH Interoperability Lab
Links

Token Ring, although not as widely used as Ethernet, is still a very popular networking technology. This network type was introduced, and heavily promoted, by IBM as its networking standard for many years and was standardized by the IEEE in the 802.5 standard. It is a very robust technology which uses a token-passing Media Access Control protocol to eliminate collisions and to ensure that the network will give excellent performance under conditions of heavy loading. Token Ring originally operated at a speed of 4 Mbps, which was later upgraded to 16 Mbps. Almost all currently produced equipment will support either speed.


Topology

Every Token Ring network is built in a star wired ring topology. The network is physically wired with all nodes connecting to a device called a Multistation Access Unit (MSAU) in a star topology. The MSAUs are connected together in a ring topology. Figure One shows the physical layout of a Token Ring network.

Figure One
Token Ring Physical Topology


Cable Types

Token Ring networks support three types of cable: Unshielded Twisted Pair (UTP), IBM Cabling System Shielded Twisted Pair (STP), and Fiber Optic. This allows the designer of a Token Ring network to use the cable type which will work the best in his or her environment the network. A description of each follows below:

Unshielded Twisted Pair (UTP)
Unshielded Twisted Pair is currently the most popular cable type for new networks. UTP has the ability to support almost any network technology currently available, and the cable is physically small and easy to work with. The disadvantage to using UTP in a Token Ring network is that it drastically limits the size of the ring, especially at 16 Mbps.
IBM Cabling System Shielded Twisted Pair (STP)
IBM has a family of cable types which fit their vision of how to wire a building for data applications. This is called the IBM Cabling System, and consists of seven different types of cable for various applications.
Fiber Optics
Fiber optic cables are used primarily for extending the size of the ring interconnecting MSAUs beyond its normal limitations. It is also occasionally used to connect workstations where security or long distance is needed.

It is acceptable to mix fiber optic cable in Token Ring networks built around either UTP or STP cable, however, do not attempt to use both STP and UTP in one ring. The two cable types have vastly different impedances, and the results of mixing them are often disasterous. If there is a need to expand an existing STP network with a new UTP section, then it is strongly reccomended to build the new UTP portion of the network as a separate ring and connect the existing STP ring and the new UTP ring with a local bridge. This prevents impedance mismatches and other problems from arising which can be difficult or impossible to solve.


Which Cable Type To Use

STP (commonly called Type 1) cable allows a physically larger ring with more nodes on it than UTP does. STP networks are allowed to have up to 250 nodes attached to one ring, and UTP networks may only have a maximum of 72 nodes. However, STP is not very useful for other applications, and is more expensive to purchase and install than UTP. In general, if the network is going to operate at 16 Mbps, then UTP networks are so limited in distance that active MSAUs are virtually required, and it is normally necessary to break the network into several small rings interconnected with local bridges, routers, or a Token Ring switch. This can make the UTP network far more costly than an STP network, even though the cable is less expensive.


The MSAU

The MSAU is the "heart" of every Token Ring network. Every MSAU has two special purpose ports marked Ring In and Ring Out. These ports are only used to interconnect MSAUs, and can not be used to connect to PCs, printers, or any other network device. The MSAU also has at least two Lobe ports. These ports are used to connect to network devices only, and can not be used ot interconnect MSAUs.

All Token Ring networks use a ring topology for the electrical (logical) network, and a combination of star and ring topologies for the actual physical layout of the network. See Figure Two for a graphical depiction of the electrical and physical networks.

Figure Two
Basic Token Ring Network

In Figure Two we can see how Token Ring creates a logical ring network with a physical star. Each cable between the MSAU and the node's Network Interface Card (NIC) has two pairs of wires in it. One pair of wires carries the signal transmitted from the NIC to the MSAU, and the second pair of wires carries the signal recieved from the MSAU to the NIC. In the MSAU, if we do not connect a cable to a Ring In or Ring Out port, that port loops the signal back through the MSAU to maintain the integrity of the ring.

This setup works beautifully for a network which is small enough for all of the nodes to be connected to one MSAU. If we need to enlarge the network by adding another MSAU, then we connect the Ring Out port of our first MSAU to the Ring In port of the new MSAU. We repeat this process for as many MSAUs as we will need to support the number of nodes in the network. When we have interconnected enough MSAUs, then we have two choices to pick from: we can either connect the last MSAUs Ring Out port to the Ring In port of our first MSAU, or we can leave those ports disconnected and rely on the automatic loopback built into them to maintain the integrity of the ring. Figure Three, shows the former option, and Figure Four shows the latter.

Figure Three
Multiple MSAUs & Fully Closed Ring

Figure Four
Multiple MSAUs & Ring Not Fully Closed

One may wonder why to bother closing the ring with a cable between the last MSAU's Ring Out and the First MSAU's Ring In when the MSAUs will automatically close the ring without it. There is a very good reason for closing the ring, however, and it has to do with reliability of the network. In Figure Two the reader can see a second ring between the two MSAUs. This ring is not normally used, however it is there and can be used if the main ring happens to be broken at any point. This ensures that the network will stay up even if a cable run should happen to fail. Without the backup ring, any cable fault will most likely cause the entire network to fail.


Active & Passive MSAU Considerations

Every Token Ring NIC serves the purpose of repeating any signal it recieves to the next node on the ring. In theory, this allows the network to operate with a MSAU which only functions as a wiring device to connect one node to the next. Such a MSAU is called Passive, and it merely passes through whatever signal it recieves to the next port without changing the signal in any way. Such a MSAU requires no electrical power and is the least expensive type available, making them appear very attractive from a cost standpoint. However, they have a major disadvantage in that the cable length supported from MSAU to node will vary depending on the overall size of the ring, number of nodes, and other factors.

Token Ring also allows the use of Active MSAUs. These units require electrical power, and amplify any signal recieved back to full strength before sending it to the next node. This results in a MSAU to node distance which does not vary with the overall network. It also increases the reliability of the network to some extent. Active MSAUs are strongly reccomended, especially in UTP networks.


How Does Token Passing Work?

Earlier it was mentioned that Token Ring uses a token passing Media Access Control (MAC) protocol. Now, it is time to examine token passing in more detail.

Token passing is a Media Access Control, or MAC, protocol which determines how stations transmit data to the network and when they can do so. It is a very "low level" protocol built in to every Token Ring device and operates automatically, with no user setup or intervention required.

A token passing network operates in the following manner. One station on the network generates a special frame called a token and transmits it to the network. The network is an electrical ring and the next node recieves the token. If that station has any data to transmit, it does not repeat the token, but begins sending its data instead. The station may continue to transmit data until a timer called a Token Holding Timer expires. This timer begins when the token is captured, and when it expires, the station must stop sending data and send a new token to the network. The next station in the ring will see the token, and either capture it and send its own data or it may simply repeat the token. The process continues until the token has made a complete trip around the ring.

It is important to keep in mind that every node on the ring which is not transmitting its own data acts like a repeater and sends anything it recieves to the next node on the network. If a station happens to recieve a frame addressed to it, it copies the frame to an internal buffer, marks the frame as being recieved, and repeats it to the next node on the network, which in turn repeats the frame. This process continues until the frame reaches the station which originally sent it. That station checks the frame to make sure it was recieved and also checks to make sure the data it got back from the network is the same as what it originally transmitted. If any errors are recieved, software on the transmitting node retransmits the frame.

Token passing is sometimes difficult to understand. An analogy to a real-world situation can help to clear up the confusion surrounding it. Let's assume we have ten people standing around in a circle. Each person may speak only to the person on his or her right. Any time a person hears something from the left, he or she must immediately repeat it to the person on his or her right, and the person who originated a message expects to hear what he or she originally said to the person on his or her right be eventually repeated by the person on his or her left. That way, it is certain that the message got through the group without any errors.

Before anyone in our group can begin speaking, he or she must first have permission. The permission scheme we use is very simple; we pass a coin around the circle. If someone has something to say, he or she waits until the coin is handed to him or her, then holds the coin, says what he or she wishes, and then passes the coin to the next person in the ring. Sometimes, a message will be very long and may take a long time to send, which can deprive everyone else in the circle of a chance to say what he or she wants to say. Therefore, let's add a rule that the coin can only be held for ten seconds at a time. If a message will take more than ten seconds, then the sender must stop speaking, pass the coin to the next person, and wait for the coin to return before sending the next ten seconds of his or her message.

The above is a greatly simplified version of how token passing works. It may appear to be cumbersome and inefficient, however it is not. There is no need for the users of the computers on the network to do anything to make it function, and it actually results in a network which can handle a very heavy load better than a contention based network such as Ethernet.


Advantages & Disadvantages

Every technology has advantages and disadvantages, and Token Ring is not an exception. These advantages and disadvantages need to be weighed against the needs of an organization when the decision of which technology is made. Some of the advantages and disadvantages of Token Ring are listed below:

Advantages

"Deterministic" Nature
Token Ring networks are deterministic in nature. In English, this means that the nodes may only transmit at certain, well defined times. This eliminates incidents of two or more nodes transmitting at the same time and corrupting each other's transmission (collisions). The elimination of collisions results in a network where data almost always gets to its destination on the first attempt, greatly reducing retransmissions which waste bandwidth. The end result is a higher effective bandwidth at the same wire speed.
 

Disadvantages

Network Planning Required
Token Ring networks, especially those using passive MSAUs, must be properly preplanned before installation. This is due to limitations in the maximum length of the ring which can be handled. If a Token Ring network is simply "thrown together" without following the rules on our Design Rules page, it may not work.
 
Difficult To Physically Install
Token Ring networks using STP cable can be very difficult to physically install due to the size, weight, and stiffness of the STP cable. Networks using UTP cable avoid these difficulties, but are very limited in distance and number of nodes which can be put on one ring.
 
Difficult Troubleshooting
All ring networks share a somewhat difficult troubleshooting procedure. If any node or cable run fails, there is not usually an easy way to determine where the fault has occured. This results in a long and tedious process of elimination where the network must be broken into several small pieces and each piece troubleshot individually until the problem is found and corrected. If an intermittant problem is encountered, it may be nearly impossible to find and correct it.
 
Cost
Token Ring network equipment costs several times as much as Ethernet hardware. This is due to the complex token passing protocol which increases the difficulty of manufacturing such equipment. Networks built around STP cable require the use of a specialized, expensive cable which is not very useful for other applications, and networks built around UTP are so limited in size and maximum number of nodes that they generally require many small rings to be interconnected with costly bridges and routers.

Token Ring networks give the benefit of a very fast. deterministic network, but require a major commitment to properly plan and maintain. Before choosing Token Ring as the technology for a particular project, it is very important to ensure that the organization has the ability to commit the time and resources necessary. Not doing so is basically a recipe for disaster.