Previous Table of Contents Next

With all of these approaches, the goal is to obtain the largest amount of funding for the network (within the constraints of needs) and then to stretch that budget accordingly. There will likely be points in the design that have longer amortization schedules than others, and this will help to make the budget go further. For example, many corporations plan for the cable plant to last over fifteen years (an optimistic figure in some cases), so you shouldn’t skimp on cabling materials or installation. Such expenses can be amortized over a number of years, thus making them appear more cost effective. Plus, a few pennies saved here will likely cost a great deal more in the long run. Ultimately, it’s best to try and work with the business and the corporate culture to establish a fair method for dealing with the cost factors.

Network Design in the Real World: Cabling

A network designer installed three live Category 5 wires to each desktop along with a six-pair Category 3 for voice services in a campus installation that I eventually took over. A live connection meant that it was terminated to a shared media hub or switch. Cross-connects were accomplished virtually, using VLANs (virtual LANs). This design cost a great deal to implement, but saved thousands of dollars in cabling and cross-connects. MAC (move, add, and change) costs were greatly reduced and theoretically could have been eliminated with dynamic VLAN assignments. By the way, this particular shop had three different platforms—Macintosh, Windows, and Unix—on almost every desktop, lending itself to the three-drop design.

This is a great demonstration of the importance of considering corporate needs and, to a certain degree, culture. Various efforts to remove even some of the machines from each desktop were largely unsuccessful, prima-rily because of the corporate culture at the time. IT was unable to resolve this conflict, which resulted in spending a great deal on network, workstation, and software equipment and licenses. While the network designer should be able to work with other IT groups and management to prevent such waste, a good designer should also be able to accommodate their demands. We’ll come back to this network when discussing broadcasts and other constraints. For now, just note that multiple networks were desirable for each desktop—Macintosh and Windows on one and Unix on the other—adding another expense to the design criteria.

The Bottom Line

It helps to have a bit of accounting experience or at least a relationship with the Accounting department when calculating network design costs. Forgoing options such as leasing, there are a couple of ways to assess the cost of a network design.

Basically, costs will appear in two general categories. The first is initial costs—those costs that appear once, typically at the beginning of the purchasing process. For example, the acquisition of a router or switch would likely be an initial cost. Initial costs are important for a number of reasons. However, these costs can be a bit misleading. Larger corporations will incorporate an amortization on equipment based on the projected lifespan of the device. Thus, a router may actually be entered as a cost over 30 months instead of just one month. This variance can greatly impact the budgets of both the network and the corporation. It’s important to consult with the Accounting group in your organization so that you understand how such costs are treated.

The second category is recurring costs. These costs frequently relate to circuits and maintenance contracts and are typically paid on a monthly or annual basis. These costs can frequently overshadow the initial costs—a $100,000 router is cheap compared to a monthly $50,000 telecommunications bill. Consider that the monthly cost for a $100,000 router is only 18 percent of the cost for a $50,000-a-month circuit after the first year—and that router will have residual value for years beyond.

A significant amount of this material is written in the context of large corporations and enterprise-class businesses. In reality, the concepts hold true for even the smallest companies.

Additional Design Goals

While Cisco typically refers to the three goals of network design, our discussion would be incomplete if the list was not augmented. In addition to scalability, adaptability, and cost control, designers must be familiar with predictability, ease of implementation, manageability, and troubleshooting. These goals integrate well with the three-tier model and will be presented in greater detail in the section, “The Three-Tier (Hierarchical) Network Model,” later in this chapter.

Scalability refers to the ability to add additional nodes and bandwidth to the network, and its characteristics typically interrelate with those of predictability. Predictable networks provide the administrator with a clear traffic flow for data and, combined with baselining and monitoring, solid capacity-planning information.

A well-designed network is easily implemented. This characteristic also applies to modular designs, but it does not have to. Implementations typically work best when the developer draws upon prior experience and introduces the design in phases. Prior to deploying any new design, the developer should test it in a lab or discuss the installation with others in the field. The adage “Why reinvent the wheel?” is particularly valuable here.

The last network design goal encompasses the recurrent demand for diagnostics. Unfortunately, even the best designs fail, and sometimes these failures are the result of the design itself. A good design should focus on solid documentation and be as straightforward as possible. For example, a design that uses network address translation (NAT) when it is not required would likely be more difficult to fix in a crisis than one without NAT. Designers should refrain from adding features just because they are available and focus on simplicity of design.

Troubleshooting capabilities can be enhanced by placing monitoring tools in the network. Protocol analyzers and remote monitoring (RMON) probes should be available for rapid dispatch if permanent installations are not an option at critical points in the network, including the core and distribution layers. This chapter will later define the core and distribution layers, in addition to the hierarchical model. For now, simply consider the core and distribution layers as the backbone of the network.

Previous Table of Contents Next