Working with Hubs and Switches

As switches become more commonplace and the technology becomes less expensive, you can expect the use of switches to all but eliminate hubs. However, the reliable nature of networking devices means you are likely to see hubs installed in networks for a long time to come. Therefore, it is important to know how to work with hubs and switches, sometimes in the same environment.

You must be aware of some of the aspects of hubs and switches when working with them. This is very important because you're likely to work with both hubs and switches in a production environment.

Hub and Switch Ports

Hubs and switches have two types of ports: medium dependent interface (MDI) and medium dependent interface crossed (MDI-X). The two types of ports differ in their wiring. As the X implies, an MDI-X port's wiring is crossed; this is because the transmit wire from the connected device must be wired to the receive line on the other. Rather than use a crossover cable (which is discussed in the next section, “Cables Connecting Hubs and Switches”), you can use the more simple straight-through cable (also discussed in the next section) to connect systems to the switch or hub.

On most modern hubs and switches, a special port called the uplink port allows you to connect two hubs and switches together to create larger networks. Because the aim of this type of network connection is to make each hub or switch think that it is simply part of a larger network, the connection for the port is not crossed; a straight-through network cable is used to connect the two hubs or switches together. Figure 3.6 shows the uplink port on an Ethernet switch.

In the absence of an uplink port, you can connect two hubs or switches together by using MDI-X ports, but you must use a crossover cable to do so.

Cables Connecting Hubs and Switches

Two types of cables are used to connect devices to hubs and switches: crossover cables and straight-through cables. The difference between the two types is that in a crossover cable, two of the wires are crossed; in a straight-through cable, all the wires run straight through.

Specifically, in a crossover cable, Wires 1 and 3 and Wires 2 and 6 are crossed: Wire 1 at one end becomes Wire 3 at the other end, Wire 2 at one end becomes Wire 6 at the other end, and vice versa in both cases. You can see the differences between the two cables in Figures 3.7 and 3.8. Figure 3.7 shows the pinouts for a straight-through cable, and Figure 3.8 shows the pinouts for a crossover cable.

IN THE FIELD: HOW MANY IS TOO MANY? Although Ethernet standards state that you can have as many as 1,024 nodes on a network, the practical maximum may be much lower. The number of nodes you can accommodate depends on a number of factors. Using switches instead of hubs makes a huge difference, particularly if you are using the full-duplex features of these devices. The amount of traffic generated by clients also has a significant effect, as does the type of traffic. On a more subtle level, you must consider the quality of the networking components and devices you use.

Hubs and switches are sometimes equipped with a network connection for another cable type, such as coaxial. Other higher-end devices simply have empty sockets into which you can plug connectivity modules of choice. This approach lets you create very fast networks. For example, three 24-port 10/100 Ethernet switches could be connected to each other by a Gigabit Ethernet fiber-optic connection. This would create a very fast network structure in which switch-to-system communication can occur at 200Mbps (in full-duplex mode) and switch-to-switch communication can occur at Gigabit Ethernet speeds. The result is a very fast local area network (LAN).

Hub and Switch Indicator Lights

Both hubs and switches use light-emitting diodes (LEDs) to indicate certain connection conditions. At the very least, a link light on the hub will indicate the existence of a live connection. On higher-end devices, additional lights might indicate activity, the speed of the connection, whether the connection is at half- or full-duplex, and sometimes errors or collisions. The LEDs provide an immediate visual indicator about the status of the device, so familiarizing yourself with their function is a worthwhile exercise. There is a further discussion of hub and switch LEDs in Chapter 14, “Troubleshooting Tools and Utilities.”

Rack Mount, Stackable, and Freestanding Devices

Some hubs and switches, as well as many other networking devices, are designed to be placed in a rack, whereas others are labeled as stackable or freestanding. Rack-mount devices are designed for placement into equipment racks, which are a common sight in computer rooms. The racks are approximately 19 inches wide; devices that are designed to be rack-mounted are slightly smaller than freestanding devices, so they can fit in the racks. Small metal brackets are screwed to the sides of the devices to allow them to be fitted into the racks.

If you don't have racks, you need to use stackable or freestanding devices. These devices can literally be placed on top of one another. Many network equipment manufacturers realize that not everyone has racks, and so they make their equipment usable in either a rack or a freestanding configuration.

Managed Hubs and Switches

Both hubs and switches come in managed and unmanaged versions. A managed device has an interface through which it can be configured to perform certain special functions. For example, it may allow for port mirroring, which can be useful for network monitoring, or allow ports to be specified to operate at a certain speed. Because of the extra functionality of a managed device, and because of the additional components required to achieve it, managed devices are considerably more expensive than unmanaged devices. When you're specifying switches or hubs, consider the need for manageability carefully. If a switch will be used to connect servers to the network, a managed device might make the most sense—the extra functionality might come in handy. On parts of the network that accommodate client computers, nonmanaged devices generally suffice.

At the time of this writing, switches are still quite a bit more expensive than hubs with equivalent capacity, but the gap is narrowing quickly. Some manufacturers have stopped producing hubs and instead are putting all their efforts into developing switches. This would seem to be a sound strategy. In all but the smallest networks or companies with the most restrictive budgets, hubs are rapidly being replaced by switches. In new implementations, hubs are very unlikely to be specified and installed.