Token Ring FAQ

How does token ring work?

A token ring network uses a special frame called a token that rotates around the ring when no stations are actively sending information. When a station wants to transmit on the ring, it must capture this token frame. The owner of the token is the only station that can transmit on the ring, unlike the Ethernet topology where any station can transmit at any time. Once a station captures the token, it changes the token into a frame format so data can be sent.

As the data traverses the ring, it passes through each station on the way to the destination station. Each station receives the frame and regenerates and repeats the frame onto the ring. As each station repeats the frame, it performs error checks on the information within the frame. If an error is found, a special bit in the frame called the Error Detection bit is set so other stations will not report the same error.

Once the data arrives at the destination station, the frame is copied to the destination's token ring card buffer memory. The destination station repeats the frame onto the ring, changing two series of bits on the frame. These bits, called the Address Recognized Indicator (ARI) and the Frame Copied Indicator (FCI), determines if the destination station had seen the frame and has had ample buffer space available to copy the frame into memory. If the frame is not copied into memory, it is the responsibility of the sending station to retransmit the frame.

The frame continues around the ring, arriving back at the source station who recognizes the sending address as it's own. The frame is then stripped from the ring, and the source station sends a free token downstream.

What is used to convert between Ethernet and Token Ring?

There is no 'converter' that allows an Ethernet network and Token Ring network to communicate between each other. A conversion process must occur between the two topologies, since they both use different signaling types, frame structures, and frame sizes.

There are two methods to accomplish this 'conversion'; bridging, and routing.

• Bridging

Bridging is a method of communicating between devices at OSI layer 2, the data link layer. A bridge connects two networks together and acts as a traffic director. If traffic is destined to the other network, the bridge allows the traffic to pass. If the traffic is local to a single network, the bridge does not pass the traffic unnecessarily to the other connected network.

The bridge makes this determination based on the Media Access Control (MAC) address of the workstations on the network. The bridge keeps an updated list of everyone active on the network, and uses this list to direct traffic from one network to another.

This method of operation makes the network appear as a single logical network, since the only separation of traffic from one network to another is done at the MAC address level.

There are many bridge manufacturers and bridge types on the market. The newest version of this bridging technology is called a DLC Switch or LAN Switch. These switches have a much higher port density than the older two or three port bridges, allowing for much more flexibility and network segmentation.

• Routing

The second method of 'converting' from Ethernet to Token Ring is called routing. Routing occurs at OSI layer 3, and separates physical networks into separate logical networks. This differentiates routing from bridging, since bridging maintains a single logical network.

In a routed network, the sending workstation determines if outgoing traffic is local or remote. If the traffic belongs to another network, the originating station sends the frame directly to the router for further processing.

Upon receiving the frame from the source workstation, the router examines the frame for the destination address. The router maintains a routing table which is used to determine the final destination of the data packet through the router.

Routing is the most common method of connecting Ethernet networks to Token Ring networks in most organizations. Most network operating systems have routing capabilities built into the servers. By placing a token ring and Ethernet card into a Novell NetWare 3.x/4.x or Windows NT v4.x server, the two topologies can communicate between each other.

One caveat; some protocols are not routeable. A good example is Microsoft's NetBEUI, which has no OSI layer 3 network address and therefore cannot be routed. Protocols which cannot be routed must be bridged between physical networks.

What physical devices are required for a token ring network?

Token ring connectivity requires three separate physical entities; a Multistation Access Unit (MAU), a token ring lobe cable, and a token ring adapter card.

A Multistation Access Unit (MAU or MSAU) is a hub-like device that connects to all token ring stations. Although the token ring stations are attached to the MAU in a physical star configuration, a true ring is maintained inside the MAU.

Unlike an Ethernet hub, a MAU consists of physical or electronic relays which keep each station in a loopback state until a voltage is sent from the station to the MAU. Since this voltage does not affect data communications, it is referred to as a 'phantom' voltage. Once this phantom voltage is received by the MAU, a relay is activated that inserts the token ring station onto the ring.

MAUs are connected together with Ring In/Ring Out (RI/RO) cables. To maintain a true ring, both the RI and the RO ports must be connected from one MAU to the other.

A token ring lobe cable connects the token ring station to the MAU. This cable communicates over four wires; two for transmit and two for receive. The cable can be Shielded Twisted Pair (STP) or Unshielded Twisted Pair (UTP).

A token ring adapter card is the physical interface that a station uses to connect to a token ring network. There are token ring adapter cards for almost every computer bus type.,

What types of cables are used for token ring?

There are three major physical token ring cabling systems; Shielded Twisted Pair (STP), Unshielded Twisted Pair (UTP), and optic fiber.

What pin assignments are used in token ring cabling?

An IBM-type Data Connector or Universal Data Connector (IDC or UDC), is a hermaphroditic connector (neither male nor female). These connectors attach to each other without having a specified male or female connector type on each end. These connectors are commonly found on IBM Type 1 cabling, a two-pair shielded cable.

The UDC connector has the following cabling requirements:

• Red - Receive +
• Green - Receive -
• Orange - Transmit +
• Black - Transmit -

A DB-9 connector uses four wires (two pairs) for token ring networking:

• Pin 1 - Red - Receive +
• Pin 5 - Black - Transmit -
• Pin 6 - Green - Receive -
• Pin 9 - Orange - Transmit +

A RJ-45 connector is an eight wire twisted pair cable:

• Pin 3 - Blue/White - Transmit -
• Pin 4 - White/Orange - Receive +
• Pin 5 - Orange/White - Receive -
• Pin 6 - White/Blue - Transmit +

RJ-11 connectors are rarely used:

• Pin 2 - Blue/White - Transmit -
• Pin 3 - White/Orange - Receive +
• Pin 4 - Orange/White - Receive -
• Pin 5 - White/Blue - Transmit +

What is the difference between a MAU, a CAU, and a LAM?

A MAU is a 8228 Multistation Access Unit. This unit provides eight workstation connectors and 2 MAU ports (also called Ring In/Ring Out ports).

A CAU is a 8230 Controlled Access Unit (Basically a MAU with intelligence). A CAU supports up to four LAMs. The Ring In/Ring Out ports of a CAU are copper, but can replaced with fiber connectors.

A LAM is a Lobe Attachment Module for the 8230. Each LAM supports 20 workstations.

Can two token ring stations be directly attached?

Unlike Ethernet stations, token ring stations _cannot_ be directly attached with a cross-over cable. Because of the process required for inserting into a ring, a loopback process must complete and phantom voltage must exist on a wire for a relay to open. A MAU must be used to directly connect two workstations.

However, some token ring switches allow a station to directly connect to a _switch_. This Direct Token Ring (DTR) connection is a non-standard method of connecting a switch and a workstation onto a single ring. This non-standard DTR connectivity does _not_ allow for two workstations to be directly connected.

What is the maximum distance between a MAU and a token ring station, or between two token ring stations?

In token ring networking, distance requirements are different from vendor to vendor. In general terms, the recommended standard distance between stations for Type 1 cabling is approximately 300 meters, and the recommended standard distance between stations for UTP cabling is about 150 meters.

Token ring distances are computed as the distance between repeaters. IN a token ring network, each Network Interface Card (NIC) is a repeater. Therefore, the length between stations cannot exceed the cable lengths listed above.

Some manufacturers use 'active' MAUs which can regenerate the token ring signal and act as a repeater. In these cases, the distances between the token ring workstations and the MAUs can be much larger than many 'passive' MAUs. Many active MAUs have other network management features such as SNMP capabilities and auto-station removal for stations inserting at the incorrect speeds.

What is the formula for computing adjusted ring length (ARL)?

The adjusted ring length of a token ring network is the sum of all cable lengths between wiring closets, minus the shortest cable between wiring closets. The ARL is used to determine the total length of the ring, and the maximum lobe distances (see section [4.8]).

This calculation determines the ring length if part of the ring is removed for troubleshooting. When a cable is removed from a Ring In/Ring Out port, the loop-back creates a much larger ring than normal. The ARL calculation defines the largest ring size that can occur, based on the shortest cable between wiring closets.

Why is ring length important?

The design of any network is dependent on limits. In token ring networks, ring length is a large factor in the physical design of an error-free networ. If the ring is too long, timing and attenuation issues can create physical-layer errors, disrupting communication over the entire ring.

In the design of a token ring network, total ring length dictates the maximum length of cable between the workstation and the MAU. This value, called the lobe length, can be computed with a series of tables. These tables are computed for passive MAU networks. Active MAUs provide capabilities that deviate greatly from the values in these tables. Consult the manufacturer of the active MAUs for values that are appropriate for that equipment.

At what speeds does token ring run?

Token ring runs at speeds of 4 megabits per second (500,000 bytes per second) and 16 megabits per second (2,000,000 bytes per second). Some token ring switches support a non-standard referred to as Direct Token Ring (DTR), or full-duplex token ring. This allows for 16 megabit speeds in the sending and receiving directions simultaneously, for a maximum of 32 megabits per second (4,000,000 bytes per second).

How many stations are supported by a single token ring network?

Again, this number is dependent on the token ring equipment that is used in the network. Current standards list a maximum of 72 stations on a UTP ring, and approximately 250 to 260 on a Type 1 network.

What is High Speed Token Ring?

High Speed Token Ring, or HSTR, is a new token ring standard that promises to push token ring speeds to 100 Mbps and 1 Gbps. The High Speed Token Ring Alliance consists of 3Com, Bay Networks, IBM, Madge, Olicom, UNH Interoperability Lab, and Xylan.

The first HSTR specification will allow for 100 Mbps token ring speeds over both Type 1 and UTP copper cabling. Further specifications will tackle 100 Mbps token ring over fiber. These standards are due for completion in June or July of 1998. Another HSTR specification will allow for 1 Gbps HSTR over fiber, and this standard is due to be completed at the end of 1998.