IBM Token-Ring was developed by IBM in the middle 1980s with an interest in supplying a fast and reliable alternative to Ethernet. Although IBM Token-Ring (or Token-Ring as we will refer to it in this chapter) networks are wired in a star configuration, Token-Ring actually operates in a logical ring, meaning the central device that connects the computers (a Multistation Access Unit , or MAU ) hosts an internal ring (more about these devices in a moment), where access to the network media is handled by possession of a token that is passed from computer to computer on the ring.
Token-Ring specifications are found under the IEEE 802.5 specifications. Token-Ring operates at the Data Link layer of the OSI model ( specifically the MAC sublayer of the Data Link layer). Token-Ring implementations often use Type 1 shielded cabling (limiting ring membership to 250 devices) or Category 5 twisted pair cabling (limited ring membership to 72 devices).
Token-Ring hardware (network cards and MAUs) is generally more expensive than Ethernet hardware, but Token-Ring networks are considered more reliable in high-traffic situations because the way that computers gain access to the network medium does not produce the collisions that can take place in the Ethernet environment. Let's take a close look at the Token-Ring media-access strategy.
Token-Ring Network Access Strategy
As already mentioned, Token-Ring networks are wired in a star configuration, with a Multistation Access Unit (MAU) providing the central connection for the devices on the LAN. A MAU on a Token-Ring network is the equivalent of a hub on an Ethernet network, but the MAU is a much more sophisticated (and expensive) device. Token-Ring uses a ring topology, where data travels in only one direction. However, the actual ring on which the token is circulated is a logical ring inside of the MAU. Figure 4.2 provides a diagram of a Token-Ring network and the logical ring provided by the MAU.
Figure 4.2. Token-Ring LANs are wired in a star topology but operate as a ring.
Access to the network media is handled using a token. The token is passed around the ring until a computer wishing to send information out onto the network takes possession of the token.
A computer that passes the token to the next computer on the logical ring would be called the nearest active upstream neighbor (NAUN) . The computer receiving the token is the nearest active downstream neighbor (NADN) . Once a computer takes possession of the token and transmits data, it then creates a new token and passes it to its NADN. The token makes its way around the ring until a node on the network takes possession of it to transmit.
Token-Ring IEEE and Cabling Standards
The specifications for running IBM Token-Ring architecture have been defined by the IEEE and are designated as IEEE 802.5. Token-Ring, like Ethernet and the other network architectures (such as FDDI, which we discuss in the next section), operates at the OSI model's Data Link layer (covered in detail in the next chapter, "Network Protocols: Real and Imagined").
Because using a token-passing strategy means that computers can only broadcast data when they possess the token, Token-Ring is characterized by no collisions. Another plus of the Token-Ring strategy is that more-equal access to the network media is provided when compared to the strategy used by Ethernet (a device on an Ethernet network can actually dominate the network by flooding it with data). The fact that Token-Ring networks provide equal access for sending data has typically made them popular in certain industries, such as banking, because they provide a more guaranteed real-time delivery of important information (such as bank deposits and withdrawals).
Typical Token-Ring networks are slower than Ethernet (particularly Fast Ethernet). Token-Ring comes in two different flavors: a 4Mbps type and a 16Mbps type.
However, faster Token-Ring implementations are becoming a reality. IBM and other Token-Ring hardware vendors , such as Olicom Inc., are now making 100Mbps Token-Ring NICs. This new 100Mbps flavor of Token-Ring is referred to as High Speed Token-Ring or HSTR. Olicom also has developed a high-speed switch to complement the use of HSTR NICs on a Token-Ring network. The high-speed NICs and switches can be used on existing Token-Ring infrastructures without a change of LAN cabling.
Gigabit Token-Ring is now also available and is being used for high-speed backbones to connect Token-Ring LANs. The 1000Mbps (1000Mbps is a gigabit) throughput on the network is possible because of Gigabit Token-Ring switches that have been developed by companies such as Madge. These switches basically allow an uplink to be created between two Token-Ring LANs using fiber- optic cabling as the link backbone.
As far as cabling specifications go, Token-Ring uses an entirely different system for numbering and categorizing the types of cables used for Token-Ring networks. The two most common cabling types used are referred to as Type 1 and Type 3 . Type 1 is a twisted-pair cable that is encased in a braided shielding (this type of cable is known as shielded twisted pair, or STP ). Type 3 cable is an unshielded twisted-pair cable and is less expensive than the Type 1 shielded cable. It is limited for use on 4Mbps networks, however.
Fiber-optic cable can also be used to link Token-Ring MAUs together. Because fiber-optic cable provides for greater cable run lengths (which we will discuss shortly), connecting MAUs allows the possibility to extend the LAN over a greater physical distance. This type of cable is referred to as Type 5 .