1.7 Other LANs

---

There are other LANs that once showed promise, but have fallen out of favor for one reason or another. These include ARCnet, StarLAN, and the FDDI. Other LAN technologies came and went so quickly in the early 1990s that they are not worth further discussion. These include IBM’s 25-Mbps LAN technology based on ATM and Hewlett-Packard’s 100VG-AnyLAN, both of which were intended to support Ethernet and token-ring frames. An intriguing idea from National Semiconductor resulted in isochronous Ethernet, which essentially provided conventional Ethernet LANs with quality of service for multimedia applications via Integrated Services Digital Network (ISDN) to the desktop. All of these technologies elicited a big yawn from the marketplace.

1.7.1 ARCnet

The Attached Resource Computer network (ARCnet), introduced by Datapoint Corp. in 1977, is the first LAN technology. Over the years, ARCnet has been overshadowed by higher-speed LAN technologies, notably Ethernet and token ring. Today, ARCnet products are still available and embedded in many companies’ products, but they may not be advertised as ARCnet. Unlike other network technologies, there is no upward migration path for ARCnet to higher speeds that approach the Ethernet standard at 10 Gbps.

Currently, ARCnet products for the LAN are available that operate at 2.5 Mbps. Higher-speed ARCnet products are available at 5 Mbps and 10 Mbps, but these are used as general-purpose communications controllers for networking microcontrollers and intelligent peripherals in industrial, automotive, and embedded control environments using an ARCnet protocol engine. An ARCnet protocol engine is the ideal solution for embedded control applications because it provides a deterministic token-passing protocol, a highly reliable and proven networking scheme.

Most ARCnet products operate at 2.5 Mbps. When an ARCnet node receives the token, it is permitted to send packets of data to other stations. While the tokenring protocol passes its token around a physical cable ring, ARCnet passes its token from node to node in order of each node’s address.

Nodes could be located up to 2,000 feet from an ARCnet hub through the use of RG-62 coaxial cabling. The total end-to-end length of the network could be 20,000 feet—nearly 4 miles. With twisted-pair wiring, each node could be located up to 400 feet from an ARCnet hub. As many as 254 connections may be supported on a single ARCnet LAN via interconnected active hubs.

ARCnet was originally designed to operate in a distributed star configuration, which entailed each node being directly connected to a hub (see Figure 1.5), with several hubs connected to each other. This design suited organizations that terminated cables in centrally located wiring closets. Later, an Ethernet-like bus topology was introduced for ARCnet. This allowed nodes to be interconnected via a single run of cable. The two topologies could even be combined for maximum configuration flexibility. For example, instead of connecting a single ARCnet node to a port on an ARCnet hub, a bus cable with a maximum of eight nodes attached could be connected to the hub.

Figure 1.5: ARCnet configuration incorporating both passive and active hubs.

ARCnet makes use of two types of hubs: passive and active. Passive hubs are small, four-port devices (nonpowered) that support workstations at distances up to 100 feet using coaxial cabling. Active hubs are eight-port units that support workstations at distances up to 200 feet using coaxial cabling and up to 400 feet with twisted-pair wiring. By attaching passive hubs to each of an active hub’s eight ports, a single active hub can support 24 workstations.

The primary advantages of this distributed star arrangement include cost savings on cable installation and hub ports. A central active hub that uses twisted-pair wiring offers the best protection against network failure, since each station has its own dedicated connection to the active hub. Furthermore, the central hub approach makes all the wiring accessible at one point, which simplifies troubleshooting, fault isolation, and network expansion.

From the beginning, ARCnet used internal transceivers in its hubs. Under Datapoint’s concept of “conjoint networks,” there was no need for bridges and routers to interconnect multiple LANs. Selected workstations and/or file servers could be configured to participate directly in up to six LANs at the same time. Access to each LAN or group of LANs was effectively controlled through hardware.

Although ARCnet is rarely considered today by companies seeking LAN solutions, the technology is still employed for such niche applications as data acquisition, plant monitoring and control, closed-circuit cameras, commercial heating and air conditioning, waterway controls, automotive navigation systems, building automation, and motor drive communications. The vendors of these applications use ARCnet as the network for linking various system components.

1.7.2 StarLAN

AT&T developed StarLAN in the 1980s to satisfy the need for a low-cost, easy-to-install LAN that would offer more configuration flexibility than token ring and higher availability than Ethernet. The hub-based StarLAN was offered in two versions—1 Mbps and 10 Mbps. Since it was based on IEEE 802.3 standards, StarLAN offered interoperability with Ethernet and token ring through driver software.

The StarLAN architecture was based on the use of one or more hubs. Connectivity between PCs and hubs was achieved through the use of UTP wiring. In multi-hub networks, up to five levels of hubs could be cascaded, with one hub designated as the header to which one or more intermediate hubs were connected. The maximum distance between adjacent hubs was 250 meters. The maximum span of a five-level network was 2,500 meters.

In 1991, AT&T and NCR merged under the name AT&T Global Information Solutions, where responsibility for StarLAN resided until 1996. That year, AT&T Global Information Solutions changed its name back to NCR Corp. in anticipation of being spun off to AT&T shareholders as an independent, publicly traded company. Around that time, NCR discontinued the StarLAN product line.

1.7.3 Fiber Distributed Data Interface

Fiber distributed data interface (FDDI) is a 100-Mbps token-passing network that employs a dual counter-rotating ring topology for fault tolerance. Originally conceived to operate over multimode fiber-optic cable, the standard has evolved to embrace single-mode fiber-optic cable, STP copper, and even UTP copper wiring. It is designed to provide high-bandwidth, general-purpose interconnection between computers and peripherals, including the interconnection of LANs (see Figure 1.6) and other networks, within a building or campus environment.

Figure 1.6: Via encapsulation, FDDI can carry Ethernet and token-ring frames as data, providing a multiprotocol backbone network.

Operation

A timed token-passing access protocol is used to pass frames of up to 4,500 bytes in size, supporting up to 1,000 connections over a maximum multimode fiber path of 200 km (124 miles) in length. Each station along the path serves as the means for attaching and identifying devices on the network, regenerating and repeating frames sent to it. Unlike other types of LANs, FDDI allows both asynchronous (time-insensitive) and synchronous (time-sensitive) devices to share the network.

Synchronous services (e.g., voice and video) are intolerant of delays and must be guaranteed a fixed bandwidth or time slot. Synchronous traffic is therefore given priority over asynchronous traffic, which is better able to withstand delay. FDDI stresses reliability, and its architecture includes integral management capabilities, including automatic failure detection and network reconfiguration.

Any change in the network status—such as power-up or the addition of a new station—leads to a “claim” process during which all stations on the network bid for the right to initialize the network. Every station indicates how often it must see the token to support its synchronous service. The lowest bid represents the station that must see the token most frequently. That request is stored as the target token rotation time (TTRT). Every station is guaranteed to see the token within 2 × TTRT seconds of its last appearance.

This process is completed when a station receives its own claim token. The winning station issues the first unrestricted token, initializing the network on the first rotation. On the second rotation, synchronous devices may start transmitting. On the third and subsequent rotations, asynchronous devices may transmit, if there is available bandwidth. Errors are corrected automatically via a beacon-and-recovery process during which the individual stations seek to correct the situation.

Architecture

These processes are defined in a set of standards sanctioned by the American National Standards Institute (ANSI). The standards address the following four functional areas of the FDDI architecture (see Figure 1.7):

• Physical media dependent (PMD): Data is transmitted between stations after converting the data bits into a series of optical pulses. The pulses are then transmitted over the cable linking the various stations. The PMD sublayer describes the optical transceivers—specifically, the minimum optical power and sensitivity levels over the optical data link. This layer also defines the connectors and media characteristics for point-to-point communications between stations on the FDDI network. The PMD sublayer is a subset of the physical layer of the OSI reference model, defining all of the services needed to transport a bit stream from station to station. It also specifies the cabling requirements for FDDI-compliant cable plant, including worst-case jitter and variations in cable attenuation.

• Physical layer (PHY): The PHY protocol defines those portions of the physical layer that are media independent, describing data encoding/decoding, establishing clock synchronization, and defining the handshaking sequence used between adjacent stations to test link integrity. It also provides the synchronization of incoming and outgoing code-bit clocks and delineates octet boundaries as required for the transmission of information to or from higher layers. These processes allow the receiving station to synchronize its clock to the transmitting station.

• MAC: FDDI’s data link layer is divided into two sublayers. The MAC sublayer governs access to the medium. It describes the frame format, interpretation of frame content, generates and repeats frames, issues and captures tokens, controls timers, monitors the ring, and interfaces with station management. The LLC sublayer is required for proper ring operation and is part of the IEEE 802.2 standard.

  
  Figure 1.7: FDDI layers and their relationship to the seven-layer OSI reference model.

In keeping with the IEEE model, the FDDI MAC is fully compatible with the IEEE 802.2 LLC standard. Applications that interface to the LLC and operate over existing LANs, such as IEEE 802.3 CSMA/CD or 802.5 token ring, are able to operate over an FDDI network. The FDDI MAC, like the 802.5 token-ring MAC, has two types of protocol data units, a frame and a token. Frames are used to carry data (such as LLC frames), while tokens are used to control a station’s access to the network. At the MAC layer, data is transmitted in 4-bit blocks called 4B/5B symbols. The symbol coding is such that four bits of data are converted to a 5-bit pattern; thus, the 100-Mbps FDDI rate is provided at 125 million signals per second on the medium. This signaling type is employed to maintain signal synchronization on the fiber.

• Station management (SMT): The SMT facility provides the system management services for the FDDI protocol suite, detailing control requirements for the proper operation and interoperability of stations on the FDDI ring. It acts in concert with the PMD, PHY, and MAC layers. The SMT facility is used to manage connections, configurations, and interfaces. It defines such services as ring and station initialization, fault isolation and recovery, and error control. SMT is also used for statistics gathering, address administration, and ring partitioning.

Topology

FDDI is a token-passing ring network. Like all rings, it consists of a set of stations connected by point-to-point links to form a closed loop. Each station receives signals on its input side and regenerates them for transmission on the output side. Any number of stations, theoretically, can be attached to the network, although default values in the FDDI standard assume no more than 1,000 physical attachments and a 200-km path.

FDDI uses two counter-rotating rings: a primary ring and a secondary ring. Data traffic usually travels on the primary ring. The secondary ring operates in the opposite direction and is available for fault tolerance. If appropriately configured, stations may transmit simultaneously on both rings, thereby doubling the bandwidth of the network to 200 Mbps.

Three classes of equipment are used in the FDDI environment: single-attached stations (SASs), dual-attached stations (DASs), and concentrators (CONs).

A DAS physically connects to both rings, while a SAS connects only to the primary ring via a wiring concentrator. In the case of a link failure, the internal circuitry of a DAS can heal the network using a combination of the primary and secondary rings. If a link failure occurs between a concentrator and a SAS, the SAS becomes isolated from the network.

These equipment types may be arranged in any of three topologies: dual ring, tree, and dual ring of trees (see Figure 1.8). In the dual-ring topology, DASs form a physical loop, in which case all the stations are dual attached. In a tree topology, remote SASs are linked to a concentrator, which is connected to another concentrator on the main ring.

Figure 1.8: FDDI dual-ring topology with three types of interconnecting devices.

Any DAS connected to a concentrator performs as a SAS. Concentrators may be used to create a network hierarchy, which is known as a dual ring of trees. This topology offers a flexible hierarchical system design that is efficient and economical. Devices requiring highly reliable communications attach to the main ring, while those less critical attach to branches off the main ring. Thus, SAS devices can communicate with the main ring, but without the added cost of equipping them with a dual-ring interface or a loop-around capability that would otherwise be required to ensure the reliability of the ring in the event of a station failure.

Failure Protection

FDDI provides an optional bypass switch at each node to overcome a failure anywhere on the network. In the event of a node failure, it is bypassed optically, removing it from the network. Up to three nodes in sequence may be bypassed; enough optical power will remain to support the operable portions of the network.

In the event of a cable break, the dual counter-rotating ring topology of FDDI allows use of the redundant cable to handle normal 100 Mbps traffic. If both the primary and secondary cables fail, the stations adjacent to the failures automatically loop the data around and between rings (see Figure 1.9), thus forming a new C-shaped ring from the operational portions of the original two rings. When the fault is healed, the network will reconfigure itself again.

Figure 1.9: Self-healing capability of FDDI’s dual-ring topology.

Normally, FDDI concentrators offer two buses, which correspond to the two FDDI backbone rings. Fault tolerance is also provided for stations that are connected to the ring via a concentrator because the concentrator provides the looparound function for attached stations.

---