Upgrading and Repairing Networks

Building and Campus Topologies

Constructing a LAN within the confines of a building or campus immediately focuses you on the physical topology of your LAN environment. Let's take a look at how you can apply the topologies we've discussed to a physical implementation.

Connecting Network Segments Within a Building: The Backbone

You can think of a backbone as the spine of your network. The backbone integrates all the other LAN segments in one cohesive structure and facilitates the communication among these different segments. Take a look at Figure 2.12. You can see how a star topologybased backbone has been used to tie together three separate segments in this fictitious three-story building.

Backbones can be implemented with copper wiring, but fiber-optic cables are far more popular for a few reasons:

• The distance up through the risers of a building can be too long for copper wiring. Fiber-optic cables can usually handle a signal for far greater distances and therefore are a more popular medium for backbones. It is rare today to find a large network that uses coaxial cabling for a network backbone segment. There is still a large installed base of backbone segments using twisted-pair copper cables (usually running at 100Mbps), but new installations typically use fiber-optic cabling.

• Usually, electrical trunks servicing power for the building are inside the risers. Noisy electrical subsystems can wreak havoc with high-speed data communications, and copper wiring is sensitive to this electrical interference. Fiber optic, on the other hand, uses LEDs or lasers for its signaling and is not prone to electrical interference.

• Because fiber optic does not succumb to electrical interference, it can support far higher network speeds, making it the perfect medium for a backbone. With backbones, you might want to jump to higher-speed LAN technologies as they become available. Higher-speed LAN technologies are not always readily supported on old copper wiring standards and can force you to perform costly wiring upgrades.

Backbones can also span buildings. Once again, an important point to keep in mind is the distances with which you are working. If you design beyond the specifications of the network technology you are using, be prepared for a poor performing or inoperable network environment.

Design Considerations in a Campus LAN Environment

Integrating the LANs of several buildings creates a campus network. As a LAN grows to this scale, you face a few more design challenges: scalability, redundancy, and fault tolerance. This means you must give considerable thought to the many design considerations you have to keep in mind when designing a campus LAN environment (see Figure 2.13).

Figure 2.13. The topology for a campus network is more complex than a simple bus or star.

[View full size image]

In this figure you can see that this campus network has two buildings that are both three stories tall. Suppose that each "node" represents 100 computer workstations. In this scenario, you would have 300 computer workstations per floor, 900 per building, and 1,800 for the entire campus.

Scalability

Scalability is the capability of the design to meet all the network traffic requirements and to continue to accommodate them as the company grows. The scalability question enters into several areas in your network design. Let's consider just a few of them:

• Where will you place critical server resources that your computer workstations will be using?

• Will the network technology used in your risers be capable of supporting the 300 workstations you have placed on each floor?

• Will the backbone that ties your campus together be capable of supporting cross-building traffic among the 1,800 workstations?

• Will the hubs or switches have enough bandwidth capacity to process the traffic among the three floors and two buildings?

• Will the network protocols allow you to properly address every workstation in this environment?

• Will the LAN environment be capable of accommodating the volume of broadcasts?

• Will the overall distance between the two farthest nodes on the network be outside the bounds of the network technology you have chosen?

• Are there workgroups that have particularly high network demands? How will you accommodate them?

Redundancy

Redundancy is the capability of the network to fail-over to secondary paths when your primary paths are cut or experience some other type of hardware failure. Cost is often the determining factor in how much redundancy you need to deploy. The more redundancy you have in a network, the greater the cost of its implementation. On the other hand, if nodes or some part of your network infrastructure should fail, added redundancy can insolate your network from extended periods of downtime, something that can be even more costly. Referring to Figure 2.13, consider the design from a redundancy standpoint, giving some thought to the following questions:

• How many single points of failure do you have in the network design?

• Can the environment afford to accommodate downtime during normal hours of operation? Should you consider redundancy between buildings, within buildings, or both?

• Where can you place critical resources so that clients could still work even if you were to lose an entire floor of the building?

• Where will you place resources on the LAN that are critical to the entire campus?

• What if a backhoe were to cut through your cross-building backbone?

By using a multi-tiered network topology (see the next section), you can create a network that can handle network failures of various types.