CompTIA Network+ N10-006 Cert Guide: Network Components

This chapter is from the book 

After completion of this chapter, you will be able to answer the following questions:

• What are the characteristics of various media types?
• What is the role of a given network infrastructure component?
• What features are provided by specialized network devices?
• How are virtualization technologies impacting traditional corporate data center designs?
• What are some of the primary protocols and hardware components found in a Voice over IP (VoIP) network?

Many modern networks contain a daunting number of devices, and it is your job to understand the function of each device and how they work with one another. To create a network, these devices obviously need some sort of interconnection. That interconnection uses one of a variety of media types. Therefore, this chapter begins by delving into the characteristics of media types, such as coaxial cable, twisted-pair cable, fiber-optic cable, and wireless technologies.

Next, infrastructure components (for example, hubs, bridges, switches, multilayer switches, and routers) are identified, along with their purpose. Special attention is given to switches, because they make up a significant part of a local-area network’s (LAN) infrastructure.

Finally, this chapter introduces you to a collection of specialized network devices. These include a virtual private network (VPN) concentrator, a firewall, a Domain Name System (DNS) server, a Dynamic Host Configuration Protocol (DHCP) server, a proxy server, a caching engine, and a content switch.

Foundation Topics

Media

By definition, a network is an interconnection of devices. Those interconnections occur over some type of media. The media might be physical, such as a copper or fiber-optic cable. Alternatively, the media might be the air, through which radio waves propagate (as is the case with wireless networking technologies).

This section contrasts various media types, including physical and wireless media. Although wireless technologies are introduced, be aware that wireless technologies are examined more thoroughly in Chapter 8, “Wireless LANs.”

Coaxial Cable

Coaxial cable (commonly referred to as coax) is composed of two conductors. As illustrated in Figure 3-1, one of the conductors is an inner insulated conductor. This inner conductor is surrounded by another conductor. This second conductor is sometimes made of a metallic foil or woven wire.

Because the inner conductor is shielded by the metallic outer conductor, coaxial cable is resistant to electromagnetic interference (EMI). For example, EMI occurs when an external signal is received on a wire and might result in a corrupted data transmission. As another example, EMI occurs when a wire acts as an antenna and radiates electromagnetic waves, which might interfere with data transmission on another cable. Coaxial cables have an associated characteristic impedance that needs to be balanced with the device (or terminator) with which the cable connects.

Three of the most common types of coaxial cables include the following:

• RG-59: Typically used for short-distance applications, such as carrying composite video between two nearby devices. This cable type has loss characteristics such that it is not appropriate for long-distance applications. RG-59 cable has a characteristic impedance of 75 Ohms.
• RG-6: Commonly used by local cable companies to connect individual homes to the cable company’s distribution network. Like RG-59 cable, RG-6 cable has a characteristic impedance of 75 Ohms.
• RG-58: Has loss characteristics and distance limitations similar to those of RG-59. However, the characteristic impedance of RG-58 is 50 ohms, and this type of coax was popular with early 10BASE2 Ethernet networks (which are discussed in Chapter 4, “Ethernet Technology”).

Although RG-58 coaxial cable was commonplace in early computer networks (in 10BASE2 networks), coaxial cable’s role in modern computer networks is as the media used by cable modems. Cable modems are commonly installed in residences to provide high-speed Internet access over the same connection used to receive multiple television stations.

Common connectors used on coaxial cables are as follows:

• BNC: A Bayonet Neill-Concelman (BNC) (also referred to as British Naval -Connector in some literature) connector can be used for a variety of applications, including being used as a connector in a 10BASE2 Ethernet network. A BNC coupler could be used to connect two coaxial cables together back to back.
• F-connector: An F-connector is often used for cable TV (including cable modem) connections.

Figure 3-2 shows what both of these connectors look like.

Twisted-Pair Cable

Today’s most popular LAN media type is twisted-pair cable, where individually insulated copper strands are intertwined into a twisted-pair cable. Two categories of twisted-pair cable include shielded twisted pair (STP) and unshielded twisted pair (UTP). A UTP coupler could be used to connect two UTP cables, back to back. Also, for adherence to fire codes, you might need to select plenum cable versus nonplenum cable.

To define industry-standard pinouts and color coding for twisted-pair cabling, the TIA/EIA-568 standard was developed. The first iteration of the TIA/EIA-568 standard has come to be known as the TIA/EIA-568-A standard, which was released in 1991.

In 2001, an updated standard was released, which became known as TIA/EIA-568-B. Interestingly, the pinout of these two standards is the same. However, the color coding of the wiring is different.

Shielded Twisted Pair

If wires in a cable are not twisted or shielded, that cable can act as an antenna, which might receive or transmit EMI. To help prevent this type of behavior, the wires (which are individually insulated) can be twisted together in pairs.

If the distance between the twists is less than a quarter of the wavelength of an electromagnetic waveform, the twisted pair of wires will not radiate that wavelength or receive EMI from that wavelength (in theory, if the wires were perfect conductors). However, as frequencies increase, wavelengths decrease.

One option of supporting higher frequencies is to surround a twisted pair in a metallic shielding, similar to the outer conductor in a coaxial cable. This type of cable is referred to as a shielded twisted-pair (STP) cable.

Figure 3-3 shows an example of STP cable. These outer conductors shield the copper strands from EMI; however, the addition of the metallic shielding adds to the expense of STP.

Unshielded Twisted Pair

Another way to block EMI from the copper strands making up a twisted-pair cable is to twist the strands more tightly (that is, more twists per centimeters [cm]). By wrapping these strands around each other, the wires insulate each other from EMI.

Figure 3-4 illustrates an example of UTP cable. Because UTP is less expensive than STP, it has grown in popularity since the mid 1990s to become the media of choice for most LANs.

UTP cable types vary in their data carrying capacity. Common categories of UTP cabling include the following:

• Category 3: Category 3 (Cat 3) cable was used in older Ethernet 10BASE-T networks, which carried data at a rate of 10 Mbps (where Mbps stands for megabits per second, meaning millions of bits per second). However, Cat 3 cable can carry data at a maximum rate of 16 Mbps, as seen in some older Token Ring networks.
• Category 5: Category 5 (Cat 5) cable is commonly used in Ethernet 100BASE-TX networks, which carry data at a rate of 100 Mbps. However, Cat 5 cable can carry ATM traffic at a rate of 155 Mbps. Most Cat 5 cables consist of four pairs of 24-gauge wires. Each pair is twisted, with a different number of twists per meter. However, on average, one pair of wires has a twist every 5 cm.
• Category 5e: Category 5e (Cat 5e) cable is an updated version of Cat 5 and is commonly used for 1000BASE-T networks, which carry data at a rate of 1 Gbps. Cat 5e cable offers reduced crosstalk, as compared to Cat 5 cable.
• Category 6: Like Cat 5e cable, Category 6 (Cat 6) cable is commonly used for 1000BASE-T Ethernet networks. Some Cat 6 cable is made of thicker conductors (for example, 22-gauge or 23-gauge wire), although some Cat 6 cable is made from the same 24-gauge wire used by Cat 5 and Cat 5e. Cat 6 cable has thicker insulation and offers reduced crosstalk, as compared with Cat 5e.
• Category 6a: Category 6a (Cat 6a), or augmented Cat 6, supports twice as many frequencies as Cat 6 and can be used for 10GBASE-T networks, which can transmit data at a rate of 10 billion bits per second (10 Gbps).

Although other wiring categories exist, those presented in the previous list are the categories most commonly seen in modern networks.

Most UTP cabling used in today’s networks is considered to be straight-through, meaning that the RJ-45 jacks at each end of a cable have matching pinouts. For example, pin 1 in an RJ-45 jack at one end of a cable uses the same copper conductor as pin 1 in the RJ-45 jack at the other end of a cable.

However, some network devices cannot be interconnected with a straight-through cable. For example, consider two PCs interconnected with a straight-through cable. Because the network interface cards (NICs) in these PCs use the same pair of wires for transmission and reception, when one PC sends data to the other PC, the receiving PC would receive the data on its transmission wires, rather than its reception wires. For such a scenario, you can use a crossover cable, which swaps the transmit and receive wire pairs between the two ends of a cable.

Common connectors used on twisted-pair cables are as follows:

• RJ-45: A type 45 registered jack (RJ-45) is an eight-pin connector found in most Ethernet networks. However, most Ethernet implementations only use four of the eight pins.
• RJ-11: A type 11 registered jack (RJ-11) has the capacity to be a six-pin connector. However, most RJ-11 connectors have only two or four conductors. An RJ-11 connector is found in most home telephone networks. However, most home phones only use two of the six pins.
• DB-9 (RS-232): A 9-pin D-subminiature (DB-9) connector is commonly used as a connector for asynchronous serial communications. One of the more popular uses of a DB-9 connector is to connect the serial port on a computer with an external modem.

Figure 3-5 shows what these connectors look like.

Plenum Versus Nonplenum Cable

If a twisted-pair cable is to be installed under raised flooring or in an open-air return, fire codes must be considered. For example, imagine that there was a fire in a building. If the outer insulation of a twisted-pair cable caught on fire or started to melt, it could release toxic fumes. If those toxic fumes were released in a location such as an open-air return, those fumes could be spread throughout a building, posing a huge health risk.

To mitigate the concern of pumping poisonous gas throughout a building’s heating, ventilation, and air conditioning (HVAC) system, plenum cabling can be used. The outer insulator of a plenum twisted-pair cable is not only fire retardant; some plenum cabling uses a fluorinated ethylene polymer (FEP) or a low-smoke polyvinyl chloride (PVC) to minimize dangerous fumes.

Fiber-Optic Cable

An alternative to copper cabling is fiber-optic cabling, which sends light (instead of electricity) through an optical fiber (typically made of glass). Using light instead of electricity makes fiber optics immune to EMI. Also, depending on the Layer 1 technology being used, fiber-optic cables typically have greater range (that is, a greater maximum distance between networked devices) and greater data-carrying capacity.

Lasers are often used to inject light pulses into a fiber-optic cable. However, lower-cost light emitting diodes (LED) are also on the market. Fiber-optic cables are generally classified according to their diameter and fall into one of two categories: multimode fiber (MMF) and single-mode fiber (SMF).

The wavelengths of light also vary between MMF and SMF cables. Usually, wavelengths of light in a MMF cable are in the range of 850–1300 nm, where nm stands for nanometers. A nanometer is one billionth of a meter. Conversely, the wavelengths of light in a SMF cable use usually in the range of 1310–1550 nm. A fiber coupler could be used to connect two fiber cables, back to back.

Multimode Fiber

When a light source, such as a laser, sends light pulses into a fiber-optic cable, what keeps the light from simply passing through the glass and being dispersed into the surrounding air? The trick is that fiber-optic cables use two different types of glass. There is an inner strand of glass (that is, a core) surrounded by an outer cladding of glass, similar to the construction of the previously mentioned coaxial cable.

The light injected by a laser (or LED) enters the core, and the light is prevented from leaving that inner strand and going into the outer cladding of glass. Specifically, the indices of refraction of these two different types of glass are so different that if the light attempts to leave the inner strand, it hits the outer cladding and bends back on itself.

To better understand this concept, consider a straw in a glass of water, as shown in Figure 3-6. Because air and water have different indices of refraction (that is, light travels at a slightly different speed in air and water), the light that bounces off of the straw and travels to our eyes is bent by the water’s index of refraction. When a fiber-optic cable is manufactured, dopants are injected into the two types of glasses, making up the core and cladding to give them significantly different indices of refraction, thus causing any light attempting to escape to be bent back into the core.

The path that light travels through a fiber-optic cable is called a mode of propagation. The diameter of the core in a multimode fiber is large enough to permit light to enter the core at different angles, as depicted in Figure 3-7. If light enters at a steep angle, it bounces back and forth much more frequently on its way to the far end of the cable as opposed to light that enters the cable perpendicularly. If pulses of light representing different bits travel down the cable using different modes of propagation, it is possible that the bits (that is, the pulses of light representing the bits) will arrive out of order at the far end (where the pulses of light, or absence of light, are interpreted as binary data by photoelectronic sensors).

For example, perhaps the pulse of light representing the first bit intersected the core at a steep angle and bounced back and forth many times on its way to the far end of the cable, while the light pulse representing the second bit intersected the core perpendicularly and did not bounce back and forth very much. With all of its bouncing, the first bit has to travel further than the second bit, which might cause the bits to arrive out of order. Such a condition is known as multimode delay distortion. To mitigate the issue of multimode delay distortion, MMF typically has shorter distance limitations, as opposed to SMF.

Single-Mode Fiber

SMF eliminates the issue of multimode delay distortion by having a core with a diameter so small that it only permits one mode (that is, one path) of propagation, as shown in Figure 3-8. With the issue of multimode delay distortion mitigated, SMF typically has longer distance limitations than MMF.

A potential downside to SMF, however, is cost. Because SMF has to be manufactured to very exacting tolerances, you usually pay more for a given length of fiber-optic cabling. However, for some implementations, where greater distances are required, the cost is an acceptable trade-off to reach greater distances.

Some common connectors used on fiber-optic cables are as follows:

• ST: A straight tip (ST) connector is sometimes referred to as a bayonet connector, because of the long tip extending from the connector. ST connectors are most commonly used with MMF. An ST connector connects to a terminating device by pushing the connector into the terminating equipment and then twisting the connector housing to lock it in place.
• SC: Different literature defines an SC connector as subscriber connector, standard connector, or square connector. The SC connector is connected by pushing the connector into the terminating device, and it can be removed by pulling the connector from the terminating device.
• LC: A Lucent connector (LC) connects to a terminating device by pushing the connector into the terminating device, and it can be removed by pressing the tab on the connector and pulling it out of the terminating device.
• MTRJ: The most unique characteristic of a media termination recommended jack (MTRJ) connector is that two fiber strands (a transmit strand and a receive strand) are included in a single connector. An MTRJ connector is connected by pushing the connector into the terminating device, and it can be removed by pulling the connector from the terminating device.

Figure 3-9 shows what these connectors look like.

Fiber Connector Polishing Styles

Fiber optic cables have different types of mechanical connections. The type of connection impacts the quality of the fiber optic transmission. Listed from basic to better, the options include Physical Contact (PC), Ultra Physical Contact (UPC), and Angled Physical Contact (APC), which refer to the polishing styles of fiber optic connectors. The different polish of the fiber optic connectors results in different performance of the connector. The less back reflection, the better the transmission. The PC back reflection is –40 dB, the UPC back reflection is around –55 dB, and the APC back reflection is about –70 dB.

Media Converters

There may be times when the media needs to be converted. To do this, a media converter could be used. Examples may include single-mode fiber to Ethernet, multimode fiber to Ethernet, fiber to coaxial, or single-mode to multimode fiber.

Cable Distribution

After deciding on what type of media you are going to use in your network (for example, UTP, STP, MMF, or SMF), you should install that media as part of an organized cable distribution system. Typically, cable distribution systems are hierarchical in nature.

Consider the example profiled in Figure 3-10. In this example, cable from end-user offices runs back to common locations within the building. These locations are sometimes referred to as wiring closets. Cables in these locations might terminate in a patch panel. This patch panel might consist of some sort of cross-connect block wired into a series of ports (for example, RJ-45 ports), which can be used to quickly interconnect cables coming from end-user offices with a network device, such as an Ethernet switch. A building might have multiple patch panels (for example, on different floors of a building). These common locations, where cables from nearby offices terminate, are often called intermediate distribution frames (IDFs).

The two most popular types of cross-connect blocks found in an IDF are as follows:

• 66 block: A 66 block, as shown in Figure 3-11, was traditionally used in corporate environments for cross-connecting phone system cabling. As 10-Mbps LANs grew in popularity, in the late 1980s and early 1990s, these termination blocks were used to cross-connect Cat 3 UTP cabling. The electrical characteristics (specifically, crosstalk) of a 66 block, however, do not support higher-speed LAN technologies, such as 100-Mbps Ethernet networks.

  Figure 3-11 66 Block

• 110 block: Because 66 blocks are subject to too much crosstalk (that is, interference between different pairs of wires) for higher-speed LAN connections, 110 blocks, an example of which is provided in Figure 3-12, can terminate a cable (for example, a Cat 5 cable) being used for those higher-speed LANs.

  Figure 3-12 110 Block

This centralized distribution frame, which connects out to multiple IDFs, is called the main distribution frame (MDF).

With such a wide variety of copper and fiber cabling used by different network devices, you might need one or more media converters. Examples of media converters include the following:

• Fiber (MMF or SMF) to Ethernet
• Fiber to coaxial
• SMF to MMF

Wireless Technologies

Not all media is physical, as is the case of wireless network technologies. This book dedicates Chapter 8 to these technologies. However, for now, you just need to understand the basics.

Consider the sample wireless topology presented in Figure 3-13. Notice that wireless clients gain access to a wired network by communicating via radio waves with a wireless access point (AP). The AP is then hardwired to a LAN.

As discussed in Chapter 8, wireless LANs include multiple standards that support various transmission speeds and security features. However, you need to understand, at this point, that all wireless devices connecting to the same AP are considered to be on the same shared network segment, which means that only one device can send data to and receive data from an AP at any one time.

Network Infrastructure Devices

The devices used in a network infrastructure can vary based on the Layer 1 technology used. For example, a Token Ring network (which is rare today) might use a multistation access unit (MAU), while an Ethernet network might use a switch.

Because Ethernet-based networks are dominant in today’s LANs, however, the infrastructure devices presented here lend themselves to networks using Ethernet as the Layer 1 transport. Some devices (such as a router, for example) function basically the same regardless of the Layer 1 transport being used.

Hubs

As mentioned in Chapter 2, “The OSI Reference Model,” a hub (specifically, an Ethernet hub in this discussion) lives at Layer 1 of the OSI model. As a result, a hub does not make forwarding decisions. Instead, a hub receives bits in on one port and then retransmits those bits out all other ports (that is, all ports on the hub other than the port on which the bits were received). This basic function of a hub has caused it to gain the nickname of a bit spitter.

Hubs most often use UTP cabling to connect to other network devices; however, some early versions of Ethernet hubs (prior to the popularization of Ethernet switches) supported fiber-optic connections.

The three basic types of Ethernet hubs are as follows:

• Passive hub: Does not amplify (that is, electrically regenerate) received bits.
• Active hub: Regenerates incoming bits as they are sent out all the ports on a hub, other than the port on which the bits were received.
• Smart hub: The term smart hub usually implies an active hub with enhanced features, such as Simple Network Management Protocol (SNMP) support.

A significant downside to hubs, and the main reason they have largely been replaced with switches, is that all ports on a hub belong to the same collision domain. As discussed in Chapter 4, a collision domain represents an area on a LAN on which there can be only one transmission at a time. Because multiple devices can reside in the same collision domain, as is the case with multiple PCs connected to a hub, if two devices transmit at the same time, those transmissions collide and have to be retransmitted.

Because of the collision-domain issue, and the inefficient use of bandwidth (that is, bits being sent out all ports rather than only the port needing the bits), hubs are rarely seen in modern LANs. However, hubs are an important piece of the tapestry that makes up the history of Ethernet networks and represent characteristics found in different areas of modern Ethernet networks. For example, a wireless AP is much like a hub, in that all the wireless devices associated with the AP belong to the same collision domain.

Consider Figure 3-14. Notice that the PCs depicted are interconnected using an Ethernet hub, but they are all in the same collision domain. As a result, only one of the connected PCs can transmit at any one time. This characteristic of hubs can limit scalability of hub-based LANs.

Also notice that all devices on a hub belong to the same broadcast domain, which means that a broadcast sent into the hub will be propagated out all of the ports on the hub (other than the port on which the broadcast was received).

Bridges

A bridge joins two or more LAN segments, typically two Ethernet LAN segments. Each LAN segment is in separate collision domains, as shown in Figure 3-15. As a result, an Ethernet bridge can be used to scale Ethernet networks to a larger number of attached devices.

Unlike a hub, which blindly forwards received bits, a bridge (specifically, an Ethernet bridge in this discussion) makes intelligent forwarding decisions based on the destination MAC address present in a frame. Specifically, a bridge analyzes source MAC address information on frames entering the bridge and populates an internal MAC address table based on the learned information. Then, when a frame enters the bridge destined for a MAC address known by the bridge’s MAC address table to reside off of a specific port, the bridge can intelligently forward the frame out the appropriate port. Because this operation is logically the same as switch operation, a more detailed description is presented in the upcoming discussion on switches. Because a bridge makes forwarding decisions based on Layer 2 information (that is, MAC addresses), a bridge is considered to be a Layer 2 device.

Although a bridge segments a LAN into multiple collision domains (that is, one collision domain per bridge port), all ports on a bridge belong to the same broadcast domain. To understand this concept, think about the destination MAC address found in a broadcast frame. At Layer 2, the destination MAC address of a broadcast frame is FFFF.FFFF.FFFF in hexadecimal notation. Also, recall that a bridge filters frames (that is, sends frames only out necessary ports) if the bridge has previously learned the destination MAC address in its MAC address table. Because no device on a network will have a MAC address of FFFF.FFFF.FFFF, a bridge will never enter that MAC address in its MAC address table. As a result, broadcast frames are flooded out all bridge ports other than the port that received the frame.

Popular in the mid to late 1980s and early 1990s, bridges have largely been replaced with switches, for reasons including price, performance, and features. From a performance perspective, a bridge makes its forwarding decisions in software, whereas a switch makes its forwarding decisions in hardware, using application-specific integrated circuits (ASICs). Also, not only do these ASICs help reduce the cost of switches, they enable switches to offer a wider array of features. For example, Chapter 4 discusses a variety of switch features, including VLANs, trunks, port mirroring, Power over Ethernet (PoE), and 802.1X authentication.

Switches

Like a bridge, a switch (specifically, a Layer 2 Ethernet switch in this discussion) can dynamically learn the MAC addresses attached to various ports by looking at the source MAC address on frames coming into a port. For example, if switch port Gigabit Ethernet 1/1 received a frame with a source MAC address of DDDD.DDDD.DDDD, the switch could conclude that MAC address DDDD.DDDD.DDDD resided off of port Gigabit Ethernet 1/1. In the future, if the switch received a frame destined for a MAC address of DDDD.DDDD.DDDD, the switch would only send that frame out of port Gigabit Ethernet 1/1.

Initially, however, a switch is unaware of what MAC addresses reside off of which ports (unless MAC addresses have been statically configured). Therefore, when a switch receives a frame destined for a MAC address not yet present in the switch’s MAC address table, the switch floods that frame out of all the switch ports except the port on which the frame was received. Similarly, broadcast frames (that is, frames with a destination MAC address of FFFF.FFFF.FFFF) are always flooded out all switch ports except the port on which the frame was received. As mentioned in the discussion on bridges, the reason broadcast frames are always flooded is that no endpoint will have a MAC address of FFFF.FFFF.FFFF, meaning that the FFFF.FFFF.FFFF MAC address will never be learned in a switch’s MAC address table.

To illustrate how a switch’s MAC address table becomes populated, consider an endpoint named PC1 that wants to form a Telnet connection with a server. Also, assume that PC1 and its server both reside on the same subnet. (That is, no routing is required to get traffic between PC1 and its server.) Before PC1 can send a Telnet session to its server, PC1 needs to know the IP address (that is, the Layer 3 address) and the MAC address (Layer 2 address) of the server. The IP address of the server is typically known or is resolved via a Domain Name System (DNS) lookup. In this example, assume the server’s IP address is known. To properly form a Telnet segment, however, PC1 needs to know the server’s Layer 2 MAC address. If PC1 does not already have the server’s MAC address in its ARP cache, PC1 can send an Address Resolution Protocol (ARP) request in an attempt to learn the server’s MAC address, as shown in Figure 3-16.

When switch SW1 sees PC1’s ARP request enter port Gigabit 0/1, PC1’s MAC address of AAAA.AAAA.AAAA is added to switch SW1’s MAC address table. Also, because the ARP request is a broadcast, its destination MAC address is FFFF.FFFF.FFFF. Because the MAC address of FFFF.FFFF.FFFF is not known to switch SW1’s MAC address table, switch SW1 floods a copy of the incoming frame out all switch ports, other than the port on which the frame was received, as shown in Figure 3-17.

When switch SW2 receives the ARP request over its Gig 0/1 trunk port, the source MAC address of AAAA.AAAA.AAAA is added to switch SW2’s MAC address table, as illustrated in Figure 3-18. Also, similar to the behavior of switch SW1, switch SW2 floods the broadcast.

The server receives the ARP request and responds with an ARP reply, as shown in Figure 3-19. Unlike the ARP request, however, the ARP reply frame is not a broadcast frame. The ARP reply, in this example, has a destination MAC address of AAAA.AAAA.AAAA.

Upon receiving the ARP reply from the server, switch SW2 adds the server’s MAC address of BBBB.BBBB.BBBB to its MAC address table, as shown in Figure 3-20. Also, the ARP reply is only sent out port Gig 0/1 because switch SW1 knows that the destination MAC address of AAAA.AAAA.AAAA is available off of port Gig 0/1.

When receiving the ARP reply in its Gig 0/2 port, switch SW1 adds the server’s MAC address of BBBB.BBBB.BBBB to its MAC address table. Also, like switch SW2, switch SW1 now has an entry in its MAC address table for the frame’s destination MAC address of AAAA.AAAA.AAAA. Therefore, switch SW1 forwards the ARP reply out port Gig 0/1 to the endpoint of PC1, as illustrated in Figure 3-21.

After receiving the server’s ARP reply, PC1 now knows the MAC address of the server. Therefore, PC1 can now properly construct a Telnet segment destined for the server, as depicted in Figure 3-22.

Switch SW1 has the server’s MAC address of BBBB.BBBB.BBBB in its MAC address table. Therefore, when switch SW1 receives the Telnet segment from PC1, that segment is forwarded out of switch SW1’s Gig 0/2 port, as shown in Figure 3-23.

Similar to the behavior of switch SW1, switch SW2 forwards the Telnet segment out of its Gig 0/2 port. This forwarding, shown in Figure 3-24, is possible, because switch SW2 has an entry for the segment’s destination MAC address of BBBB.BBBB.BBBB in its MAC address table.

Finally, the server responds to PC1, and a bidirectional Telnet session is established between PC1 and the server, as illustrated in Figure 3-25. Because PC1 learned the server’s MAC address as a result of its earlier ARP request and stored that result in its local ARP cache, the transmission of subsequent Telnet segments does not require additional ARP requests. However, if unused for a period of time, entries in a PC’s ARP cache can time out. Therefore, the PC would have to broadcast another ARP frame if it needed to send traffic to the same destination IP address. The sending of the additional ARP adds a small amount of delay when reestablishing a session with that destination IP address.

As shown in Figure 3-26, like a bridge, each port on a switch represents a separate collision domain. Also, all ports on a switch belong to the same broadcast domain, with one exception.

The exception is when the ports on a switch have been divided up into separate virtual LANs (VLANs). As discussed in Chapter 5, “IPv4 and IPv6 Addresses,” each VLAN represents a separate broadcast domain, and for traffic to travel from one VLAN to another, that traffic must be routed by a Layer 3 device.

Multilayer Switches

Although a Layer 2 switch, as previously described, makes forwarding decisions based on MAC address information, a multilayer switch can make forwarding decisions based on upper-layer information. For example, a multilayer switch could function as a router and make forwarding decisions based on destination IP address information.

Some literature refers to a multilayer switch as a Layer 3 switch because of the switch’s capability to make forwarding decisions like a router. The term multilayer switch is more accurate, however, because many multilayer switches have policy-based routing features that allow upper-layer information (for example, application port numbers) to be used in making forwarding decisions.

Figure 3-27 makes the point that a multilayer switch can be used to interconnect not just network segments, but entire networks. Specifically, Chapter 6, “Routing IP Packets,” explains how logical Layer 3 IP addresses are used to assign network devices to different logical networks. For traffic to travel between two networked devices that belong to different networks, that traffic must be routed. (That is, a device, such as a multilayer switch, has to make a forwarding decision based on Layer 3 information.)

Like a Layer 2 switch, each port on a multilayer switch represents a separate collision domain; however, a characteristic of a multilayer switch (and a router) is that it can become a boundary of a broadcast domain. Although all ports on a Layer 2 switch belong to the same broadcast domain, if configured as such, all ports on a multilayer switch can belong to different broadcast domains.

Routers

A router is a Layer 3 device, meaning that it makes forwarding decisions based on logical network address (for example, IP address) information. Although a router is considered to be a Layer 3 device, like a multilayer switch, a router has the capability to consider high-layer traffic parameters (for example, quality of service [QoS] settings) in making its forwarding decisions.

As shown in Figure 3-28, each port on a router is a separate collision domain and a separate broadcast domain. At this point in the discussion, routers are beginning to sound much like multilayer switches. So, why would a network designer select a router rather than a multilayer switch in his design?

One reason a router is preferable to a multilayer switch, in some cases, is that routers are usually more feature rich and support a broader range of interface types. For example, if you need to connect a Layer 3 device out to your Internet service provider (ISP) using a serial port, you will be more likely to find a serial port expansion module for your router, rather than your multilayer switch.

Infrastructure Device Summary

Table 3-1 summarizes the characteristics of the network infrastructure devices discussed in this section.

Table 3-1 Network Infrastructure Device Characteristics

Device	Number of Collision Domains Possible	Number of Broadcast Domains Possible	OSI Layer of Operation
Hub	1	1	1
Bridge	1 per port	1	2
Switch	1 per port	1 per port	2
Multilayer switch	1 per port	1 per port	3+
Router	1 per port	1 per port	3+

Specialized Network Devices

Although network infrastructure devices make up the backbone of a network, for added end-user functionality, many networks integrate various specialized network devices, such as VPN concentrators, firewalls, DNS servers, DHCP servers, proxy servers, caching engines, and content switches.

VPN Concentrators

Companies with locations spread across multiple sites often require secure communications between those sites. One option is to purchase multiple WAN connections interconnecting those sites. Sometimes, however, a more cost-effective option is to create secure connections through an untrusted network, such as the Internet. Such a secure tunnel is called a virtual private network (VPN). Depending on the VPN technology being used, the devices that terminate the ends of a VPN tunnel might be required to perform heavy processing. For example, consider a company headquarters location with VPN connections to each of 100 remote sites. The device at the headquarters terminating these VPN tunnels might have to perform encryption and authentication for each tunnel, resulting in a heavy processor burden on that device.

Although several router models can terminate a VPN circuit, a dedicated device, called a VPN concentrator, can be used instead. A VPN concentrator performs the processor-intensive process required to terminate multiple VPN tunnels. Figure 3-29 shows a sample VPN topology, with a VPN concentrator at each corporate location.

The term encryption refers to the capability of a device to scramble data from a sender in such a way that the data can be unscrambled by the receiver, but not by any other party who might intercept the data. With a VPN concentrator’s capability to encrypt data, it is considered to belong to a class of devices called encryption devices, which are devices (for example, routers, firewalls, and VPN concentrators) capable of participating in an encrypted session.

Firewalls

A firewall is primarily a network security appliance, and it is discussed in Chapter 12, “Network Security.” As depicted in Figure 3-30, a firewall stands guard at the door of your network, protecting it from malicious Internet traffic.

For example, a stateful firewall allows traffic to originate from an inside network (that is, a trusted network) and go out to the Internet (an untrusted network). Likewise, return traffic coming back from the Internet to the inside network is allowed by the firewall. However, if traffic were originated from a device on the Internet (that is, not returning traffic), the firewall blocks that traffic.

DNS Servers

A Domain Name System (DNS) server performs the task of taking a domain name (for example, www.ciscopress.com) and resolving that name into a corresponding IP address (for example, 10.1.2.3). Because routers (or multilayer switches) make their forwarding decisions based on Layer 3 information (for example, IP addresses), an IP packet needs to contain IP address information, not DNS names. However, as humans, we more readily recall meaningful names rather than 32-bit numbers.

As shown in Figure 3-31, an end user who wants to navigate to the www.ciscopress.com website enters that fully qualified domain name (FQDN) into her web browser; however, the browser cannot immediately send a packet destined for www.ciscopress.com. First, the end user’s computer needs to take the FQDN of www.ciscopress.com and resolve it into a corresponding IP address, which can be inserted as the destination IP address in an IP packet. This resolution is made possible by a DNS server, which maintains a database of local FQDNs and their corresponding IP addresses, in addition to pointers to other servers that can resolve IP addresses for other domains.

An FQDN is a series of strings delimited by a period (as in the previous example, www.ciscopress.com). The rightmost string represents the root domain. Examples of root domains include .com, .mil, .gov, and .edu. Although there are many other root domains, these are among some of the more common domains seen in the United States.

Lower-level domains can point upward to higher-level DNS servers, to resolve nonlocal FQDNs, as illustrated in Figure 3-32.

A DNS server’s database contains not only FQDNs and corresponding IP addresses, but also DNS record types. For example, a Mail Exchange (MX) record would be the record type for an e-mail server. As a few examples, Table 3-2 lists a collection of common DNS record types.

Table 3-2 Common DNS Record Types

Record Type	Description
A	An address record (that is, A record) maps a hostname to an IPv4 address.
AAAA	An IPv6 address record (that is, AAAA record) maps a hostname to an IPv6 address.
CNAME	A canonical name record (that is, CNAME record) is an alias of an existing record, thus allowing multiple DNS records to map to the same IP address.
MX	A mail exchange record (that is, MX record) maps a domain name to an e-mail (or message transfer agent) server for that domain.
PTR	A pointer record (that is, PTR record) points to a canonical name. A PTR record is commonly used when performing a reverse DNS lookup, which is a process used to determine what domain name is associated with a known IP address.
SOA	A start of authority record (that is, SOA record) provides authoritative information about a DNS zone, such as e-mail contact information for the zone’s administrator, the zone’s primary name server, and various refresh timers.

A potential challenge when setting up DNS records is when you want to point to the IP address of a device, which might change its IP address. For example, if you have a cable modem or digital subscriber line (DSL) modem in your home, that device might obtain its IP address from your service provider via DHCP (as discussed in the next section, “DHCP Servers”). As a result, if you add the IP address of your cable modem or DSL modem to a DNS record (to allow users on the Internet to access one or more devices inside your network), that record could be incorrect if your device obtains a new IP address from your service provider.

To overcome such a challenge, you can turn to dynamic DNS (DDNS). A DDNS provider supplies software that you run on one of your PCs, which monitors the IP address of the device referenced in the DNS record (that is, your cable modem or DSL modem in this example). If the software detects a change in the monitored IP address, that change is reported to your service provider, which is also providing DNS service.

Yet another DNS variant is Extension Mechanisms for DNS (EDNS). The original specification for DNS had size limitations that prevented the addition of certain features, such as security. EDNS supports these additional features, while maintaining backward compatibility with the original DNS implementation. Rather than using new flags in the DNS header, which would negate backwards compatibility, EDNS sends optional pseudo-resource-records between devices supporting EDNS. These records support 16 new DNS flags. If a legacy DNS server were to receive one of these optional records, the record would simply be ignored. Therefore, backward compatibility is maintained, while new features can be added for newer DNS servers.

When you enter a web address into your browser in the form of http://FQDN (for example, http://www.1ExamAMonth.com), notice that you not only indicate the FQDN of your web address, you also specify that you want to access this location using the HTTP protocol. Such a string, which indicates both an address (for example, www.1ExamAMonth.com) and a method for accessing that address (for example, http://), is called a uniform resource locator (URL).

DHCP Servers

Most modern networks have IP addresses assigned to network devices, and those logical Layer 3 addresses are used to route traffic between different networks. However, how does a network device receive its initial IP address assignment?

One option is to manually configure an IP address on a device; however, such a process is time consuming and error prone. A far more efficient method of IP address assignment is to dynamically assign IP addresses to network devices. The most common approach for this auto assignment of IP addresses is Dynamic Host Configuration Protocol (DHCP). Not only does DHCP assign an IP address to a network device, it can assign a wide variety of other IP parameters, such as a subnet mask, a default gateway, and the IP address of a DNS server.

If you have a cable modem or DSL connection in your home, your cable modem or DSL router might obtain its IP address from your service provider via DHCP. In many corporate networks, when a PC boots, that PC receives its IP address configuration information from a corporate DHCP server.

Figure 3-33 illustrates the exchange of messages that occur as a DHCP client obtains IP address information from a DHCP server. The following list describes each step in further detail.

1. When a DHCP client initially boots, it has no IP address, default gateway, or other such configuration information. Therefore, the way a DHCP client initially communicates is by sending a broadcast message (that is, a DHCPDISCOVER message to a destination address of 255.255.255.255) in an attempt to discover a DHCP server.
2. When a DHCP server receives a DHCPDISCOVER message, it can respond with a unicast DHCPOFFER message. Because the DHCPDISCOVER message is sent as a broadcast, more than one DHCP server might respond to this discover request. However, the client typically selects the server that sent the first DHCPOFFER response received by the client.
3. The DHCP client communicates with this selected server by sending a unicast DHCPREQUEST message asking the DHCP server to provide IP configuration parameters.
4. The DHCP server responds to the client with a unicast DHCPACK message. This DHCPACK message contains a collection of IP configuration parameters.

Notice that in Step 1, the DHCPDISCOVER message was sent as a broadcast. By default, a broadcast cannot cross a router boundary. Therefore, if a client resides on a different network than the DHCP server, the client’s next-hop router should be configured as a DHCP relay agent, which allows a router to relay DHCP requests to either a unicast IP address or a directed broadcast address for a network.

A DHCP server can be configured to assign IP addresses to devices belonging to different subnets. Specifically, the DHCP server can determine the source subnet of the DHCP request and select an appropriate address pool from which to assign an address. One of these address pools (which typically corresponds to a single subnet) is called a scope.

When a network device is assigned an IP address from an appropriate DHCP scope, that assignment is not permanent. Rather, it is a temporary assignment referred to as a lease. Although most client devices on a network work well with this dynamic addressing, some devices (for example, servers) might need to be assigned a specific IP address. Fortunately, you can configure a DHCP reservation, where a specific MAC address is mapped to a specific IP address, which will not be assigned to any other network device. This static addressing approach is referred to as a DHCP reservation.

A method for remembering the four main steps of DHCP is D.O.R.A., with each letter representing the steps discover, offer, request, and acknowledge.

Proxy Servers

Some clients are configured to forward their packets, which are seemingly destined for the Internet, to a proxy server. This proxy server receives the client’s request, and on behalf of that client (that is, as that client’s proxy), the proxy server sends the request out to the Internet. When a reply is received from the Internet, the proxy server forwards the response on to the client. Figure 3-34 illustrates the operation of a proxy server.

What possible benefit could come from such an arrangement? Security is one benefit. Specifically, because all requests going out to the Internet are sourced from the proxy server, the IP addresses of network devices inside the trusted network are hidden from the Internet.

Yet another benefit could come in the form of bandwidth savings, because many proxy servers perform content caching. For example, without a proxy server, if multiple clients all visited the same website, the same graphics from the home page of the website would be downloaded multiple times (one time for each client visiting the website). However, with a proxy server performing content caching, when the first client navigates to a website on the Internet, and the Internet-based web server returns its content, the proxy server not only forwards this content to the client requesting the web page but stores a copy of the content on its hard drive. Then, when a subsequent client points its web browser to the same website, after the proxy server determines that the page has not changed, the proxy server can locally serve up the content to the client, without having to once again consume Internet bandwidth to download all the graphic elements from the Internet-based website.

As a final example of a proxy server benefit, some proxy servers can perform content filtering. Content filtering restricts clients from accessing certain URLs. For example, many companies use content filtering to prevent their employees from accessing popular social networking sites, in an attempt to prevent a loss of productivity. A reverse proxy receives requests on behalf of a server or servers and replies back to the clients on behalf of those servers. This can also be used with load-balancing and caching to better utilize a group of servers.

Content Engines

As previously described, many proxy servers are capable of performing content caching; however, some networks used dedicated appliances to perform this content caching. These appliances are commonly referred to as caching engines or content engines.

Figure 3-35 demonstrates how a corporate branch office can locally cache information from a server located at the corporate headquarters location. Multiple requests from branch office clients for the content can then be serviced from the content engine at the branch office, thus eliminating the repetitive transfer of the same data. Depending on traffic patterns, such an arrangement might provide significant WAN bandwidth savings.

Content Switches

Consider the server farm presented in Figure 3-36. The servers making up this server farm house the same data. For companies with a large Internet presence (for example, a search engine company, an online book store, or a social networking site), a single server could be overwhelmed with the glut of requests flooding in from the Internet. To alleviate the burden placed on a single server, a content switch (also known as a load balancer) distributes incoming requests across the various servers in the server farm, each of which maintains an identical copy of data and applications.

A major benefit of content switching is that it allows a server farm to scale. Specifically, as demand increases, new servers can be added to the group of servers across which requests are load balanced. Also, if maintenance (for example, applying an operating system patch) needs to be performed on a server, a server can simply be taken out of the load-balancing rotation, with the remaining servers picking up the slack. Then, after the maintenance is complete, the server can once again be added back to the defined server group.

Virtual Network Devices

A major data center paradigm shift is underway. This shift is away from a company having its own data center (with its raised flooring and large air conditioning system) containing multiple physical servers, each of which offered a specific service (for example, e-mail, DNS services, or Microsoft Active Directory).

Virtual Servers

The computing power available in a single high-end server is often sufficient to handle the tasks of multiple independent servers. With the advent of virtualization, multiple servers (which might be running different operating systems) can run in virtual server instances on one physical device. For example, a single high-end server might be running an instance of a Microsoft Windows Server providing Microsoft Active Directory (AD) services to an enterprise, while simultaneously running an instance of a Linux server acting as a corporate web server, and at the same time acting as a Sun Solaris UNIX server providing corporate DNS services. Figure 3-37 illustrates this concept of a virtual server. Although the virtual server in the figure uses a single network interface card (NIC) to connect out to an Ethernet switch, many virtual server platforms support multiple NICs. Having multiple NICs offers increased throughput and load balancing.

Virtual Routers and Firewalls

Most of the vendors who create physical routers and firewalls also have an offering that includes virtualized routers and firewalls. The benefit of using a virtualized firewall or router is that the same features of routing and security can be available in the virtual environment as they are in the physical environment. As part of interfacing with virtual networks, virtual network adapters can be used. For connectivity between the virtual world and the physical one, there would be physical interfaces involved that connect to the logical virtual interfaces.

Virtual Switches

One potential trade-off you make with the previously described virtual server scenario is that all servers belong to the same IP subnet, which could have QoS and security implications. If these server instances ran on separate physical devices, they could be attached to different ports on an Ethernet switch. These switch ports could belong to different VLANs, which could place each server in a different broadcast domain.

Fortunately, some virtual servers allow you to still have Layer 2 control (for example, VLAN separation and filtering). This Layer 2 control is made possible by the virtual server not only virtualizing instances of servers, but also virtualizing a Layer 2 switch. Figure 3-38 depicts a virtual switch. Notice that the servers logically reside on separate VLANs, and frames from those servers are appropriately tagged when traveling over a trunk to the attached Ethernet switch.

Virtual Desktops

Another emerging virtualization technology is virtual desktops. With today’s users being more mobile than ever, they need access to information traditionally stored on their office computers’ hard drives from a variety of other locations. For example, a user might be at an airport using her smartphone, and she needs access to a document she created on her office computer. With virtual desktops, a user’s data is stored in a data center rather than on an office computer’s hard drive. By providing authentication credentials, a secure connection can be established between the centralized repository of user data and that user’s device, as shown in Figure 3-39, thus allowing the user to remotely access her document.

Other Virtualization Solutions

Although the previously discussed virtualization technologies (that is, virtual servers, virtual switches, and virtual desktops) were described as residing at a corporate location (that is, on-site), some service providers offer off-site options. Specifically, if a service provider’s customer did not want to house and maintain his own data center, these virtualization technologies could be located at a service provider’s data center, and the customer could be billed based on usage patterns. Such a service provider offering is called network as a service (NaaS), implying that network features can be provided by a service provider, just as a telephony service provider offers access to the Public Switched Telephone Network (PSTN), and an ISP offers access to the public Internet.

Cloud Computing

Virtualized services and solutions are often offered by service providers as cloud computing. A company purchasing cloud computing services has the option of public, private, or hybrid cloud services. Private cloud services include systems that only have interactions and communications with other devices inside that same private cloud or system. Public cloud services interact with devices on public networks such as the Internet and potentially other public clouds. An environment in which there are private cloud services but some of those services interact with public cloud is referred to as hybrid cloud services. Some of the types of services that can be available as part of cloud computing include infrastructure as a service, IaaS, where the company rents virtualized servers (which are hosted by a service provider) and then runs specific applications on those servers. Another type of cloud service is software as a service, SaaS, where the details of the servers are hidden from the customer and the customer’s experience is similar to using a web-based application. Another cloud service is called platform as a service, PaaS, which can provide a development platform for companies that are developing applications and want to focus on creating the software and not have to worry about the servers and infrastructure that are being used for that development. Another type of cloud is the community cloud, which is a term referring to cloud services used by individuals, companies or entities with similar interests. In cloud computing, it is likely that virtualized switches, routers, servers, and firewalls will be used as part of cloud-based services.

Similar to outsourcing the features of a data network with NaaS, a corporate telephony solution might also be outsourced. Many companies own and maintain their own private branch exchange (PBX), which is a privately owned telephone system. One option for companies that want to outsource their telephony service is to use a service provider’s virtual PBX. A virtual PBX is usually a Voice over IP (VoIP) solution, where voice is encapsulated inside data packets for transmission across a data network. Typically, a service provider provides all necessary IP telephony gateways to convert between a customer’s existing telephony system and the service provider’s virtual PBX.

Software-Defined Networking

Software-defined networking is changing the landscape of our traditional networks. A well-implemented software defined network will allow the administrator to implement features and functions and configurations without the need to do the individual command-line configuration on the network devices. The front end that the administrator interfaces with can alert the administrator to what the network is currently doing, and then through that same graphical user interface the administrator can indicate what he wants done, and then behind the scenes the detailed configurations across multiple network devices can be implemented by the software-defined network.

Voice over IP Protocols and Components

As previously mentioned, a Voice over IP (VoIP) network digitizes the spoken voice into packets and transmits those packets across a data network. This allows voice, data, and even video to share the same medium. In a network with unified communications (UC) such as voice, video, and data, specialized UC servers, controllers, devices, and gateways are also likely to be used. In a cloud computing environment, they may be virtualized as well. Figure 3-40 shows a sample VoIP network topology. Not only can a VoIP network provide significant cost savings over a traditional PBX solution, many VoIP networks offer enhanced services (for example, integration with video conferencing applications and calendaring software to determine availability) not found in traditional corporate telephony environments.

Table 3-3 defines the VoIP devices and protocols shown in Figure 3-40.

Table 3-3 VoIP Network Elements

Protocol/Device	Description
IP phone	An IP phone is a telephone with an integrated Ethernet connection. Although users speak into a traditional analog handset (or headset) on the IP phone, the IP phone digitizes the spoken voice, packetizes it, and sends it out over a data network (via the IP phone’s Ethernet port).
Call agent	A call agent is a repository for a VoIP network’s dial plan. For example, when a user dials a number from an IP phone, the call agent analyzes the dialed digits and determines how to route the call toward the destination.
Gateway	A gateway in a VoIP network acts as a translator between two different telephony signaling environments. In the figure, both gateways interconnect a VoIP network with the PSTN. Also, the gateway on the right interconnects a traditional PBX with a VoIP network.
PBX	A Private Branch Exchange (PBX) is a privately owned telephone switch traditionally used in corporate telephony systems. Although a PBX is not typically considered a VoIP device, it can connect into a VoIP network through a gateway, as shown in the figure.
Analog phone	An analog phone is a traditional telephone, like you might have in your home. Even though an analog phone is not typically considered a VoIP device, it can connect into a VoIP network via a VoIP or, as shown in the figure, via a PBX, which is connected to a VoIP network.
SIP	Session Initiation Protocol (SIP) is a VoIP signaling protocol used to set up, maintain, and tear down VoIP phone calls. Notice in the figure that SIP is spoken between the IP phone and the call agent to establish a call. The call agent then uses SIP to signal a local gateway to route the call, and that gateway uses SIP (across an IP WAN) to signal the remote gateway (on the right) about the incoming call.
RTP	Real-time Transport Protocol (RTP) is a protocol that carries voice (and interactive video). Notice in the figure that the bidirectional RTP stream does not flow through the call agent.

Real-World Case Study

Acme Inc. has decided that to keep pace with the growing customer demand, it will be using software as a service from a cloud provider for its primary business application. This will allow the company to focus on its business and use the application instead of managing and maintaining that application.

There will be some desktop computers in the office for the users, and those computers will be networked using UTP cabling that goes to a switch. The switches on each floor of the building will be secured in a locked intermediate distribution frame (IDF) in a wiring closet on each floor. For the interconnections between the switches on each of the floors, multi-mode fiber-optic cabling will be used. When purchasing their hardware and their fiber-optic cabling, Acme will want to make sure that the fiber-optic connector type matches the correct fiber interface type on the switches. In the basement of the building is an area for Acme Inc. to use as its own dedicated MDF. From the MDF, there will be cabling that goes to the demarcation point for the service provider for the WAN and Internet connectivity provided by the service provider. This connectivity will be used to access the cloud services (SaaS specifically) from the service provider and for WAN and Internet access.

Inside the building, a few of the users have mobile devices. To facilitate network access for these mobile users, wireless APs, which are physically connected through UTP cabling to the switches on each floor, will be used. Hubs will not be used because they are not very secure or effective and because all network traffic is sent to every other port on a hub, whereas a switch only forwards unicast frames to the other ports that need to see that traffic. To consolidate hardware in the MDF, multilayer switches will be used to provide Layer 2 forwarding of frames based on MAC addresses, and because they are multilayer switches, they can also provide Layer 3 forwarding of packets based on IP addresses. On the LAN, Acme intends to use a set of redundant servers near the MDF to provide services such as DHCP, DNS, and time synchronization to each of its offices on each floor. The servers can coordinate the DNS and time with other servers on the public Internet. The local servers can also be used for network authentication to control user access to the network regardless of the source including wireless, wired, or VPN. Instead of purchasing multiple physical servers, the company is going to virtualize the servers onto specialized hardware that is fault tolerant. With this solution, the company can easily add additional logical servers without purchasing a physical system for every new server. This could include unified communications servers that may be involved with voice, video, and other types of streaming data.

A VPN device will also be installed in the MDF to allow users who are connected to the Internet from their home or other locations to build a secure VPN remote access connection over the Internet to the corporate headquarters. Instead of buying a dedicated VPN device such as a concentrator, Acme is going to use a firewall that has this VPN capability integrated as part of its services.

Summary

The main topics covered in this chapter are the following:

• This chapter contrasted various media types, including coaxial cable, shielded twisted pair, unshielded twisted pair, fiber-optic cable, and wireless technologies.
• The roles of various network infrastructure components were contrasted. These components include hubs, bridges, switches, multilayer switches, and routers.
• This chapter provided examples of specialized network devices and explained how they could add network enhancements. These devices include VPN concentrators, firewalls, DNS servers, DHCP servers, proxy servers, content engines, and content switches.
• Virtual networking components were described. These components include virtual server, virtual switch, virtual desktop, and virtual PBX technologies.
• This chapter introduced VoIP and described some of the protocols and hardware components that make up a VoIP network.