Connecting devices are used to connect the segments of a network together or to connect networks to create an internetwork. These devices are classified into five categories: switches, repeaters, bridges, routers and gateways.

Each of these devices except the first one (switches) interacts with protocols at different layers of the OSI model. Repeaters forward all electrical signals and are active only at the physical layer. Bridges store and forward complete packets and affect the flow control of a single LAN.

Bridges are active at the physical and data link layers. Routers provide links between two separate LANs and are active in the physical, data link and network layers. Finally, gateways provide translation services between incompatible LANs or applications, and are active in all layers.

• Switches

A switched network consists of a series of interlinked switches. Switches are hardware/ software devices capable of creating temporary connections between two or more devices to the switch but not to each other. Switching mechanisms are generally classified into three methods: circuit switching, packet switching and message switching.

• Circuit switching creates a direct physical connection between two devices such as telephones or computers. Once a connection is made between two systems, circuit switching creates a dedicated path between two end users. The end users can use the path for as long as they want.

• Packet switching is one way to provide a reasonable solution for data transmission. In a packet-switched network, data are transmitted in discrete units of variable-length blocks called packets. Each packet contains not only data, but also a header with control information.

The packets are sent over the network node to node. At each node, the packet is stored briefly before being routed according to the information in its header. In the datagram approach to packet switching, each packet is treated independently of all others as though it exists alone.

In the virtual circuit approach to packet switching, if a single route is chosen between sender and receiver at the beginning of the session, all packets travel one after another along that route. Although these two approaches seem the same, there exists a fundamental difference between them.

In circuit switching, the path between the two end users consists of only one channel. In the virtual circuit, the line is not dedicated to two users. The line is divided into channels and each channel can use one of the channels in a link.

• Message switching is known as the store and forwarding method. In this approach, a computer (or a node) receives a message, stores it until the appropriate route is free, then sends it out. This method has now been phased out.

• Repeaters

A repeater is an electronic device that operates on the physical layer only of the OSI model. A repeater boosts the transmission signal from one segment and continues the signal to another segment. Thus, a repeater allows us to extend the physical length of a network.

Signals that carry information can travel a limited distance within a network before degradation of the data integrity due to noise. A repeater receives the signal before attenuation, regenerates the original bit pattern and puts the restored copy back on to the link.

• Bridges

Bridges operate in both the physical and the data link layers of the OSI model. A single bridge connects different types of networks together and promotes interconnectivity between networks. Bridges divide a large network into smaller segments.

Unlike repeaters, bridges contain logic that allows them to keep separate the traffic for each segment. Bridges are smart enough to relay a frame towards the intended recipient so that traffic can be filtered.

In fact, this filtering operation makes bridges useful for controlling congestion, isolating problem links and promoting security through this partitioning of traffic. A bridge can access the physical addresses of all stations connected to it.

When a frame enters a bridge, the bridge not only regenerates the signal but also checks the address of the destination and forwards the new copy to the segment to which the address belongs.

When a bridge encounters a packet, it reads the address contained in the frame and compares that address with a table of all the stations on both segments. When it finds a match, it discovers to which segment the station belongs and relays the packet to that segment only.

• Routers

Routers operate in the physical, data link and network layers of the OSI model. The Internet is a combination of networks connected by routers. When a datagram goes from a source to a destination, it will probably pass through many routers until it reaches the router attached to the destination network.

Routers determine the path a packet should take. Routers relay packets among multiple interconnected networks. In particular, an IP router forwards IP datagrams among the networks to which it connects. A router uses the destination address on a datagram to choose a next-hop to which it forwards the datagram.

A packet sent from a station on one network to a station on a neighbouring network goes first to a jointly held router, which switches it over the destination network. In fact, the easiest way to build the Internet is to connect two or more networks with a router.

Routers provide connections to many different types of physical networks: Ethernet, token ring, point-to-point links, FDDI and so on.

• The routing module receives an IP packet from the processing module. If the packet is to be forwarded, it should be passed to the routing module. It finds the IP address of the next station along with the interface number from which the packet should be sent.

It then sends the packet with information to the fragmentation module. The fragmentation module consults the MTU table to find the maximum transfer unit (MTU) for the specific interface number.

• The routing table is used by the routing module to determine the next-hop address of the packet. Every router keeps a routing table that has one entry for each destination network.

The entry consists of the destination network IP address, the shortest distance to reach the destination in hop count, and the next router (next hop) to which the packet should be delivered to reach its final destination.

The hop count is the number of networks a packet enters to reach its final destination. A router should have a routing table to consult when a packet is ready to be forwarded. The routing table should specify the optimum path for the packet.

The table can be either static or dynamic. A static table is one that is not changed frequently, but a dynamic table is one that is updated automatically when there is a change somewhere in the Internet. Today, the Internet needs dynamic routing tables.

• A metric is a cost assigned for passing through a network. The total metric of a particular router is equal to the sum of the metrics of networks that comprise the route. A router chooses the route with the shortest (smallest value) metric.

The metric assigned to each network depends on the type of protocol. The Routing Information Protocol (RIP) treats each network as one hop count. So if a packet passes through 10 networks to reach the destination, the total cost is 10 hop counts.

The Open Shortest Path First protocol (OSPF) allows the administrator to assign a cost for passing through a network based on the type of service required. A route through a network can have different metrics (costs).

OSPF allows each router to have several routing tables based on the required type of service. The Border Gateway Protocol (BGP) defines the metric totally differently. The policy criterion in BGP is set by the administrator. The policy defines the paths that should be chosen.

• Gateways

Gateways operate over the entire range in all seven layers of the OSI model. Internet routing devices have traditionally been called gateways. A gateway is a protocol converter which connects two or more heterogeneous systems and translates among them.

The gateway thus refers to a device that performs protocol translation between devices. A gateway can accept a packet formatted for one protocol and convert it to a packet formatted for another protocol before forwarding it. The gateway understands the protocol used by each network linked into the router and is therefore able to translate from one to another.