Document Links

Contents Introduction Key Planning Considerations Design Objectives Getting Started General Design Connectivity Issues for Ethernet Servers and Services Conclusion Appendix A: Standards Bodies Appendix B: Tools and Equipment Appendix C: Network Monitoring Tools References/Credits Illustrations Figure 1: Network Layers Figure 2: Network Topologies Figure 3: Hierarchical Star Network Tables Table 1: UTP Wires and Colors Printer Friendly Version Best Practices in Network Design (PDF file, 238 KB) Get Acrobat Reader

Best Practices in Network Design

Design Objectives

Performance

Performance is an end-to-end issue. Comments like "the network is slow" and "my computer is slow" are statements that users make as blanket statements of dissatisfaction. As non-technical people, they are expressing that they are trying to work and it's taking a long time or the process is abnormal in some way. Computer and network support staff hear these comments, but without a scalable, manageable network, find it difficult or impossible to solve the problems in a timely and economical fashion.

A properly designed and implemented network can provide not only a valuable tool to enhancing administrative and academic functions, but can also be easily and quickly repaired and expanded. A properly designed network can free up technical staff to assist in helping the users of the system to make better use of the resources.

So why start talking about performance? Performance is the one thing your users (customers) see and react to. Prior to networks, when one's PC was slow, no one else knew. PCs were, after all, "personal" computers; speed was measured by the CPU clock speed. Today, these computers are no longer "personal." Computers are networked into a larger community, and the demands and complexity placed on the system have increased dramatically.

The rapid evolution of the Internet has made computing a network-centric system instead of a computer-centric (or person-centric) system. In a network-centric environment, performance is affected not only by the computer, but also by other users utilizing services provided by servers, the LAN or the WAN.

In a properly designed network, each component or section of the network can be measured, monitored and tracked to help diagnose problems. Administrators can then be alerted to any needed upgrades, failures or bandwidth needs that should be addressed.

Network Layers

Networks can be thought of as layers of equipment and software. These layers include cabling, network, computers and a WAN (see Figure 1). Each location can be viewed as having these three different layers or systems and each needs the appropriate attention to build a scalable and flexible system.

Adding a lab with 30 computers can significantly impact traffic flow on a network. By having a scalable and flexible network, network technicians can easily modify the network to meet the changes in demand.

"Scalable" in network parlance means that the network can be easily modified to meet the changing demands (increasing or decreasing) in network traffic. Changes such as the addition of multimedia resources in the library may significantly increase the network traffic in a portion (or all) of a network. On the other hand, moving a multimedia lab from one room or building to another changes the network traffic in both areas.

"Flexible" means that the topology can be changed as the types of network traffic change. The network must be able to handle the movement of people and resources around the physical infrastructure.

Figure 1: Network Layers

Dynamics of Network Traffic

Overall, network traffic patterns in any network (LAN or WAN) change constantly. Traffic also fluctuates (in aggregate) during various times of the year (less during Christmas, spring break, or summer; more during 8-3 M-F, etc.). Newer multimedia technologies also demand more bandwidth; sometimes requiring a constant high-bandwidth stream, and sometimes requiring a constant low-bandwidth stream with intermittent "bursts" of traffic.

If you have a network currently in place, begin by monitoring the traffic levels. Then, armed with this information, you can begin to evaluate the "choke" points, as well as determine what additional applications you can add to the network without disruption to existing users.

As you develop an understanding of the network traffic dynamics, you can begin to develop Quality of Service (QoS) policies and then design and implement technologies that will help you monitor and enforce those policies. QoS is covered in extensive detail in a series of white papers available at http://www.more.net/services/bandwidth/index.html.

In addition, new and emerging technologies such as Voice over IP and H.323 videoconferencing are making more demands on the network to provide not just bandwidth, but guaranteed bandwidth along with acceptably low levels of latency and jitter. These QoS issues must be understood and dealt with early in the design process, or these and other high performance applications will not work well, and will likely wreak havoc on your network.