A Wide Area Network (WAN) is any network which covers a large geographic area. Many companies have multiple offices in different cities, and have a need to connect these offices together to share information. This is where WANs come in to play.
Local Area Network (LAN) technologies, such as Ethernet, Token Ring, Fast Ethernet, FDDI, or 100VG-AnyLAN are great for providing very high speed connectivity within a building. However, these technologies are limited in terms of the distance at which they can span. When a network must support connections over a larger area, WAN technologies must be used.
Figure One shows a basic Wide Area Network. In this drawing, assume that a company has offices in New York, Los Angeles, and Miami. Let us also assume that they have standardized on 10 Base-T as their LAN technology. The New York office is the corporate headquarters, and the other two are sales offices. The remote sales offices need to access information (sales orders, current inventory, prices, etc.) which is stored at the headquarters site. Since it is technically not possible to create an Ethernet network which connects all of these offices, we need to get a communication circuit between each sales office to the headquarters. These communication circuits are connected to a device at each site called a Bridge or Router which allows them to be interfaced to a LAN at each location.
Basic Wide Area Network
It is important to note that the communication circuits used in most WAN implementations are a bottleneck in the overall system. The speed degradation associated with a WAN is a direct result of the speed of the communication circuits used. For example, if the communication circuits are 56 kbps DDS lines, then they can only carry about 1/178 the amount of data per second that an Ethernet LAN is able to carry. An operation which would take one second to complete over a 10 Mbps LAN will take 178 seconds to complete over a 56 kbps WAN. However, a full T-1 circuit running at 1.544 Mbps is 1/6 as fast as the LAN and an operation which takes one second to complete over a LAN will take roughly six seconds to complete over a T-1 based WAN. This is a 27-fold speed up from a 56 kbps line.
Here is an analogy which may help to clarify this concept. Let's assume that we have a 5-gallon container of water and we want to pour that water into another 5-gallon container. If we pour the water directly from one container to the other, then it will flow as quickly as the opening in the containers will allow. This can be equated with the speed of an application over a LAN. Now, let's put a funnel into the container our water is destined for. The opening in the container is 3" in diameter, but the neck of the funnel is only 1/2" in diameter. Therefore, the neck of the funnel limits how fast the water can flow from the first container into the second one. This can be equated with the speed of a WAN. Figure Two shows a graphical example of this. If we need to pour more water per second than a certain funnel can support, then we need to replace the funnel with one having a larger neck diameter. Simalarly, if we need to transfer more data per second than a communication line can support, then we need to use a faster line.
Funnel Limiting Flow Rate
There are many different types of communication circuits available. They range in speed from a dial-up analog line using modems at 1200 bps to a dedicated OC-3 circuit running at 155,000,000 bps (155 Mbps). Like most things in life, in communication circuits you generally "get what you pay for". For example, an analog line between two points may only cost $50.00 per month, whereas an OC-3 circuit between the same points may cost $20,000.00 per month. Circuit cost is an issue with any WAN setup, and needs to be balanced with the required performance to arrive at the optimum choice of line type to use. Generally, line costs limits the speed of WAN links to the range of 56 kB/s to 1.544 Mbps.