Token-Ring and IEEE 802.5
Although IBM is usually considered to be the founder of the Token-Ring LAN standard, it was actually patented by Dr. Olaf Solderblum in Sweden in 1967. IBM obtained the technology from Dr. Solderblum and, with the assistance of Texas Instruments, developed the chipset technology and guidelines.
IBM released the technology to the IEEE, whose 802.5 subcommittee developed and released the 4Mbps Token-Ring standard in 1985. The IEEE 802.5 specification defines the MAC sublayer and the Physical layer specification, using the 802.2 specification at the LLC layer for protocol identification.
To learn more about the LLC and MAC sublayers, refer to Chapter 2.
In 1989, the IEEE released an enhancement to the 802.5 standard that defines 16Mbps Token-Ring operations.
General Token-Ring Operation
Token-Ring transmission is unidirectional, with each device always receiving from its upstream neighbor and sending to its downstream neighbor. It is a token-passing ring topology that passes frames with no collision risk because only one device can transmit at a time. Devices can, however, access the medium and transmit upon reception of a free token, which is a 3-byte signal that propagates around the ring.
The following are Token-Ring's most important characteristics:
All devices connect serially, transmitting a signal in one direction.
Each device's transmit pair connects through to its downstream neighbor's receive pair.
Signal transmission is unidirectional.
Each device directly connects in a physical star formation through central hubs known as Multi-Station Access Units (MSAUs) (see Figure 3.15). The MSAU's objective is to keep the ring functional by electrically bypassing a non-functional device or port when end devices are either turned off or fail.
Each device's network card operates as a fully functional unidirectional repeater, completely regenerating the signal and bit repeating it on.
It can operate at either 4Mbps or 16Mbps, but not both, as determined by the configuration of the network card.
All devices must agree on the speed of the ring.
Token-Ring technology employs one device as an Active Monitor, watching every other device and keeping the ring operational. All other devices are Standby Monitors, waiting for the opportunity to assume the Active Monitor's role if it is either gone or fails in its duties. You can think of the operation as an extremely paranoid society with Big Brother making sure everything functions correctlyuntil Big Brother himself malfunctions.
Responsibilities of the Active Monitor include
Guaranteeing a good frame or token exists on the ring at all times
Maintaining the master clock for the ring
Removing endlessly circulating frames
Kicking off a ring poll process every seven seconds for neighbor identification
Channel Access Method
Token-Ring devices access the channel using a token-passing method. When a device has information to transmit, it must wait for a free token, a 3-byte frame that traverses the ring and provides access to the medium. When it receives a token, it can convert it to a frame.
Stations send their frames around the ring hoping to find the destination host. All other devices on the ring check the destination address of the frame to determine whether it is for them, and then bit repeat the signal on. Each network interface card acts as a repeater, amplifying, retiming, and bit repeating the signal. The responsibility of stripping its frame and releasing a new token belongs to the sending device.
Two types of Token-Ring frames exist: token and data/command (see the following two sections for more information on these types of frames). The 802.5 specification does not define a minimum frame size for either; however, it does say that a minimum 24-bit constant signal must exist on the wire at all times.
Maximum frame size is based on two things: ring speed and Token Holding Timer (THT).
The THT is 8.9ms regardless of the ring speed. This timer mandates how long a device can transmit after obtaining the token before it must release the channel to others. The timer, coupled with the ring speed, determines the amount of data a station can transmit while it has access to the channel.
For example, on a 4Mbps ring speed and a THT of 8.9ms, a station could theoretically send out a maximum frame size of 4,450 bytes, whereas on a 16Mbps ring, the maximum frame size could be as high as 17,800 bytes.
Token frames are three bytes in length, consisting of the following:
A start delimiter (SD)
An access control byte (AC)
An end delimiter (ED)
This three-byte signal circles the ring providing attached devices with access to the channel (see Figure 3.16). Stations identify the signal as a free token by looking at the status of the token bit within the AC field. If the bit is a 0, this is a token frame. SDs and EDs simply mark the beginning and end of a frame.
All other frames carry either upper-layer user data or MAC layer management and control information. User data frames are LLC frames; control or command frames are MAC frames.
IEEE 802.5 implements two different data frames: Token-Ring and Token-Ring_SNAP. Both include the IEEE 802.2 LLC header; the only difference between the two frames is that the Token-Ring_SNAP frame has an additional 5-byte SNAP header following the LLC header. Figure 3.17 illustrates a Token-Ring data frame.
MAC layer command frames do not carry upper-layer data and, therefore, do not implement the LLC. MAC layer frames exist at the Data Link layer for the purpose of ring maintenance and management functions, and they never propagate beyond their local ring.
Ring Poll Process
Ring recovery restores the ring to operational status every seven seconds to facilitate ring recovery after new stations have inserted into or left the ring. Then, a process known as Ring Poll or Neighbor Notification begins. For communication on the ring to occur, devices must be capable of identifying and synchronizing to their upstream neighbor's signal.
The ring poll process is begun by the Active Monitor, which sends out a frame identifying its presence and its upstream neighbor's address. All other stations participate until all devices on the ring have identified themselves and their Nearest Active Upstream Neighbor (NAUN).
Beaconing is a problem-signaling process that occurs as the result of a failure or other problem on the ring. Failures can range from those that can be solved automatically at the chipset level through auto-configuration and recovery methods to those so severe that they require human intervention, such as a cable break. When a ring is in beacon mode, devices are incapable of sending data until the ring has resolved the condition. Beacon frames identify the problem area, known as the fault domain (see Figure 3.18).
Beaconing automatically occurs and takes place when
A device detects a serious problem on the ring.
The device sends out a Beacon MAC layer frame identifying its address and the address of its upstream neighbor.
The beacon frame alerts the downstream neighbor of a broken line or signaling problem on the ring that originates at the upstream neighbor.
The beaconing device identifies the fault domain as being somewhere between its receive port and its upstream neighbor's transmit port, including all cabling and devices in between.
After a network administrator determines the fault domain through beaconing, troubleshooting the problem is relatively easy. Because the fault domain exists only between the upstream and downstream neighbor, the troubleshooter can isolate the failure to one of these devices or intervening hardware components.