Packet-Switching Networks

In WAN connections that use packet switching , the data is divided into packets. The small packet size used in packet switching provides fast and efficient delivery of data.

Each packet has its own control information and is switched through the network independently. This means that data packets can follow different routes through the WAN cloud and reach the destination out of sequence. Sequencing information included with the packet (it is placed in the packet's header) is used by the receiving device to reassemble the data in the appropriate order.

Packet-switching networks take advantage of virtual circuits when transferring data. A virtual circuit establishes a defined route across the service provider's network so that all the data packets move to the destination along the same route (remember that this route is shared by packets from a lot of other users because switched networks use shared lines). The use of virtual circuits in packet-switching networks can improve the overall performance of your data transfers.

A number of packet-switching technologies exist, such as X.25, Frame Relay, and ATM. The X.25 standard was actually originally developed for connecting mainframe computers at different sites. By today's standards, X.25 (which became available in the late 1970s) is considered slow because it does a great deal of error checking (due to the low quality of phone lines available at the time it was developed). Despite the fact that there are faster technologies such as Frame Relay and ATM (two packet-switching technologies that we discuss in a moment), X.25 is still used by some companies.

Tip

Packet-switching networks have been available since the late 1970s when X.25 became available. The lower cost of packet-switching networks (when compared to dedicated leased lines) led to a fairly rapid evolution of packet-switching protocols such as Frame Relay and ATM. Although the X.25 protocol seems older than the hills now (when compared to Frame Relay and ATM), it is still used by some companies and institutions for WAN connections. It is, however, rapidly being replaced by other packet-switching alternatives.

Frame Relay

Frame Relay is the successor to the X.25 protocol and is faster because it has shed some of the error-checking functions that slowed the packet-switching capabilities of X.25. Frame Relay operates at the Data Link layer of the OSI model (look back at Chapter 5, "Network Protocols: Real and Imagined," for more about the OSI model) and uses permanent virtual circuits for communication sessions between different points on the WAN.

A permanent virtual circuit (PVC) is actually created between two points on the WAN by configuring devices such as routers (that terminate a LAN that is connected to the WAN) with a number that is provided by the service provider. This number, the Data Link Connection Identifier (DLCI), designates the PVC that will connect the two LANs via the WAN. Logical addresses such as IP addresses are then mapped to the PVC, providing a path through the WAN for the movement of data between the connected LANs.

The odd thing about Frame Relay is that it places the data in packets of varying sizes (this is like moving envelopes of different sizes through the same mail slot). This does not significantly impede data flow, but it does introduce some latency into overall throughput. Latency is a fancy word for lag time; the different size Frame Relay packets take some additional time to process by WAN switches, thus the latency. ATM, which we discuss in the next section, actually uses fixed-size packets to increase the potential throughput.

The Frame Relay protocol was originally developed for use over ISDN connections. It is now used over T-Carrier and Fractional-T connections.

Asynchronous Transfer Mode (ATM)

Asynchronous Transfer Mode (ATM) is a WAN packet-switching technology that uses a 53-byte packet called a cell to move data such as voice and video across public networks (such as the POTS) and private WANs. The ATM protocol stack consists of protocols that reside at the Data Link and Physical layers of the OSI reference model. ATM networks support the multiplexing of information, meaning that several channels of information can be contained in one data stream.

ATM uses a virtual channel as the connection between a sending and a receiving device. A virtual channel is basically equivalent to a virtual circuit (refer to the preceding section on Frame Relay). If a company wants redundant virtual channels connecting sites, several channels can be bundled together. This bundle of virtual channels is referred to as a virtual path . Virtual channel numbers and virtual path numbers are supplied to the customer by the service provider. These numbers are then used to configure devices such as routers that sit between the LAN and the provider's switching station (connected by the local loop).

ATM is typically run over high-speed fiber- optic networks. For example, Synchronous Optical Network/Synchronous Digital Hierarchy (or SONET , as it is typically known) is one of the Physical layer specifications for ATM. SONET can provide a throughput of 155Mbps over fiber-optic cabling. ATM backbones are widely used in the telecommunications industry. Many providers of DSL actually connect their DSL customers to the WAN using a high-speed fiber-optic backbone that uses the ATM WAN protocol.

Tip

To make things a little confusing, ATM is now being used as a LAN protocol. ATM switches that are used on a LAN use LAN Emulation (LANE) to emulate a LAN on top of the ATM network. This is accomplished by the LANE protocol, which can emulate either an Ethernet or token-ring network. In simple terms, this means that ATM is able to provide a high-speed environment over twisted pair wire by pretending to be either an Ethernet or token ring LAN. LANE is an advanced topic related to switching devices (switches are discussed briefly in Chapter 3, "Networking Hardware"), which we won't get to in this introductory book.