Depending on the type of topology your network uses and the type of cabling you use (cabling is discussed in Chapter 4, "Building the Network Infrastructure"), your LAN might require some sort of connectivity device to connect the various network computers, printers, and other devices together. In cases where you need to extend your LAN (say, to the second floor of an office building) or add a large number of new users to the LAN, other connectivity devices might be required. Some of these connectivity devices merely serve to connect devices; others are used to boost the data signal traveling on the network medium, and still others actually participate in determining how data traffic should flow on the network.
Let's start our discussion of network connectivity devices with the hub, which is a device you would use on a small network, or even in a peer-to-peer networking situation, to connect computers. The other devices that we will look at, such as repeaters, switches, and routers, are often lumped under the blanket term internetworking devices . An internetwork is a network of LANs, meaning that some sort of connectivity technology is used to extend a LAN beyond its typical size or to connect different LANs together into one large network.
Hubs are commonly used LAN connectivity devices (although inexpensive switches are rapidly replacing hubs on networks). They serve as the central connection points for LANs (hubs are used on LANs that embrace the star topology discussed in Chapter 2, "Different Needs, Different Networks"). A basic hub contains no active electronics and therefore cannot be used to extend a LAN (that is, extend it past the cabling distance specifications discussed in the next chapter). It basically organizes your cables and relays data signals to all the computers on the LAN.
Hubs are used on networks that use twisted-pair cabling. Ports available on the hub provide the connection points for the devices on the network. Computers and other devices are attached to the hub by individual network cables. Hubs come in many sizes and shapes and supply different numbers of ports.
In cases where the LAN outgrows the size of the hub, a new hub can be attached (the hubs are "daisy chained" together using a short connection cable often referred to as a rat tail ) to the current hub, thus providing greater port density. Figure 3.5 shows a 24-port Ethernet hub.
Figure 3.5. Hubs provide a connecting point for LAN nodes.
Hubs come in all sizes and shapes and are available in a wide range of prices. Typically, the more ports on the hub, the more expensive the hub. Hubs that support faster varieties of Ethernet, such as Fast Ethernet (which we discuss in the next chapter), will also cost more.
As you will find out in the next chapter, the different types of network cabling all have a maximum distance that they can move a data signal. In cases where a LAN must be extended beyond the maximum run for a particular cabling type, repeaters are used. A repeater takes the signal that it receives from computers and other devices on the LAN and regenerates the signal so that the signal maintains its integrity along a longer media run than is normally possible.
Repeaters don't have any capabilities for directing network traffic or deciding the particular route certain data should take; they are simple devices that just sit on the network, boosting the data signals they receive. The problem with repeaters is that they amplify the entire signal that they receive, including any line noise. So, in worst-case scenarios, they pass on data traffic that is barely discernable from the background noise on the line.
A bridge is an internetworking device used to help conserve the bandwidth available on the network. When LANs really start to grow, network data traffic can begin to overwhelm the available bandwidth on network media.
One strategy for conserving network bandwidth is to chop the network up into smaller segments. These segments are connected to a bridge. Bridges are smarter than hubs and repeaters and actually use some software to help get the job done. A bridge is able to read the MAC address (also known as the hardware address remember it's burned onto the NIC in each computer on the network) of each data packet circulating on the network segments connected to the bridge. By learning which MAC addresses live on each of the network segments, the bridge can help keep data traffic that is local to a particular segment from spreading to the other network segments that are serviced by the bridge.
A switch is another internetworking device used to manage the bandwidth on a large network. Switches are rapidly becoming one of the most used internetworking devices for connecting even smaller networks because they allow you some control over the use of the bandwidth on the network. A switch, which is often referred to as a "bridge on steroids," controls the flow of data by using the MAC address that is placed on each data packet (which coincides with the MAC address of a particular computer's network card). Switches divide networks into what are called Virtual LANs or VLANs . The great thing about a VLAN, which is a logical grouping of computers on the network into a sort of communication group, is that the computers don't have to be in close proximity or even on the same floor. This allows you to group computers that serve similar types of users into a VLAN. For example, even if your engineers are spread all over your company's office building, their computers can still be made part of the same VLAN, which would share bandwidth.
Switches use a combination of software and hardware to switch packets between computers and other devices on the network. Switches have their own operating system. Figure 3.6 shows the status of a VLAN (VLAN1) on a Cisco 2900 switch. Understanding what is being shown in this figure requires an understanding of the switch's OS. Basically, this particular screen shows the hardware/MAC address of the switch and the IP address of the switch. Other statistics relate to the number of packets sent and received by the switch (which are all at 0 because the switch has just been placed on the network).
Figure 3.6. A proprietary switch operating system is used to configure and monitor a switch.
Because switches can offer a high density of connection ports, they can replace hubs on a network. This means that each computer on the network can be connected to its own port on the switch. When PCs are directly connected to a switch, the switch can supply each PC with a dedicated amount of bandwidth. For example, users on a 100Mbps Ethernet (fast Ethernet) network can realize bandwidth of 100Mbps. The computers don't compete for the bandwidth the way computers do on a network that is connected via a hub. This is why switches are rapidly replacing hubs. Inexpensive switches are even available for the small network and home network markets.
Some switch hardware can also take advantage of full-duplex access to the network media, which allows for the sending and receiving of data simultaneously on the network. This provides network access on an Ethernet network that would essentially be collision free (Ethernet networks experience data collisions pretty much as a rule; Ethernet is discussed in the next chapter). A computer on a Fast Ethernet network, which runs at 100Mbps, would actually realize a net total of 200Mbps throughput because sending and receiving can take place simultaneously on the full-duplex media.
Switches (because of the aforementioned reasons) are becoming very popular on both small and large networks. They have all but replaced bridges as the internetworking devices for conserving network bandwidth and expanding LANs into larger corporate internetworks. And they are also making the hub a thing of the past on smaller networks.
Routers are even smarter than bridges and switches (routers operate at the Network layera higher level in the OSI conceptual model than bridges and switches, which operate at the Data Link layer; we discuss all the layers of OSI in Chapter 5). A router uses a combination of hardware and software to actually "route" data from its source to its destination. (By software , I mean an operating system.) Routers actually have a very sophisticated OS that allows you to configure their various connection ports. You can set up a router to route data packets from a number of different network protocol stacks, including TCP/IP, IPX/SPX, and AppleTalk (protocols are discussed in Chapter 4).
Routers are used to segment LANs that have become very large and congested with data traffic. Routers are also used to connect remote LANs together using different WAN technologies.
Figure 3.7 shows a Cisco 2516 router. This router has a built-in hub and three different WAN connection points. Notice the BRI port marked in the figure. BRI stands for ISDN Basic Rate Interface , which allows this router to connect to a remote network using an ISDN connection (ISDN and other WAN technologies are discussed in Chapter 13, "Expanding a LAN with WAN Technology").
Figure 3.7. Routers are used to segment networks into logical subsets .
Routers divide large networks into logical segments called subnets . This division of the network is based on the addressing scheme the network uses, such as IP addresses. Data traffic related to a particular subnet is kept local. The router only forwards data that is meant for other subnets on the extended network. This routing of network data helps conserve network bandwidth.
Routers decide how to forward data packets to their destinations based on a routing table . Routers use protocols built in to their operating system to identify neighboring routers and their network addresses (such as IP addresses). This allows routers to build a routing table. Figure 3.8 shows the command-line interface used on a Cisco router. This figure also shows the IP routing table for a small network that consists of two connected Cisco routers. Each of the subnets shown at the bottom of the table (notice the list of IP addresses) represents a different router interface. 10.2.0.0 and 10.3.0.0 are on the router that supplied this screen. The subnets 10.1.0.0 and 220.127.116.11 were discovered by the router (using the RIP protocol) on a connected router.
Figure 3.8. Routers build and use a routing table to determine where data packets should be forwarded.