5.1 Sizing Methods

5.1 Sizing Methods

There are many devices and line facilities that can be employed in a data communications network whose configuration or sizing problems are similar. Examples of line facilities include the number of dial-in local business and WATS lines required to be connected to telephone company rotaries, while examples of communications equipment sizing includes determining the number of channels on LAN access controllers, multiplexers, data concentrators , and port selectors.

5.1.1 Experimental Modeling

Basically, two methods can be used to configure the size of communications network devices. The first method, commonly known as experimental modeling , involves the selection of the device configuration based on a mixture of previous experience and intuition. Normally, the configuration selected is less than the base capacity plus expansion capacity of the device. This enables the size of the device to be adjusted or upgraded without a major equipment modification, if the initial sizing proved inaccurate. An example of experimental modeling is shown in Table 5.1.

 Experimental modeling results in the adjustment of a network configuration based upon previous experience and gut intuition: (upper) initial configuration and (lower) adjusted configuration. Power Supply Central Logic 1 3 5 7 9 Empty Slots 2 4 6 8 10 a Power Supply Central Logic 1 3 5 7 9 11 13 15 2 4 6 8 10 12 14 16 b

A rack-mounted LAN access controller is shown in Table 5.1a. Initially, the controller was obtained with five dual-port adapters to support ten ports of simultaneous operation. Assuming the base unit can support eight dual-port adapters, if the network manager's previous experience or gut intuition proves wrong, the controller can be upgraded easily. This is shown in Table 5.1b where the addition of three dual-port adapters permits the controller to support 16 ports in its adjusted configuration.

Assuming each controller port is connected to a modem and business line, experimental modeling, while often representing a practical solution to equipment sizing, can also be expensive. For example, if your organization began with the configuration shown in Table 5.1b and adjusted the number of controller ports downward to that shown in Table 5.1a, you would incur some cost for the extra modems and business lines, even if the manufacturer of the LAN access controller was willing to take back the port adapters you did not actually need. Thus, while experimental modeling is better than simply guessing, it can also result in the expenditure of funds for unnecessary hardware and communications facilities.

5.1.2 The Scientific Approach

The second method that can be employed to size network components ignores experience and intuition. This method is based on acknowledgment of data traffic and the scientific application of mathematical formulas to traffic data. Hence, it is known as the scientific approach method of equipment sizing. While some of the mathematics involved in determining equipment sizing can become quite complex, a series of tables generated by the development of appropriate computer programs can be employed to reduce many sizing problems to one of a single table lookup process.

Although there are advantages and disadvantages to each method, the application of a scientific methodology to equipment sizing is a rigorously defined approach. Thus, there should be a much higher degree of confi-dence and accuracy of the configuration selected when this method is used. On the negative side, the use of a scientific method requires a firm knowledge or accurate estimate of the data traffic. Unfortunately, for some organizations, this may be difficult to obtain. In many cases, a combination of two techniques will provide an optimum situation. For such situations, sizing can be conducted using the scientific method, with the understanding that the configuration selected may require adjustment under the experimental modeling concept. In the remainder of this chapter, we focus attention on the application of the scientific methodology to equipment sizing problems.