FitzGerald J., Dennis A., Durcikova A. Business Data Communications and Networking

Подождите немного. Документ загружается.

6.4 WIRELESS ETHERNET 215

the shared circuit using a controlled-access technique, not the contention-based approach

of traditional Ethernet. With this approach, any computer wishing to transmit ﬁrst sends a

request to transmit (RTS) to the AP, which may or may not be heard by all computers.

The RTS requests permission to transmit and to reserve the circuit for the sole use of the

requesting computer for a speciﬁed time period. If no other computer is transmitting, the

AP responds with a clear to transmit (CTS), specifying the amount of time for which

the circuit is reserved for the requesting computer. All computers hear the CTS and

remain silent for the speciﬁed time period. The virtual carrier sense method is optional.

It can always be used, never used, or used just for frames exceeding a certain size, as

set by the WLAN manager.

Controlled-access methods provide poorer performance in low-trafﬁc networks

because computers must wait for permission before transmitting rather than just wait-

ing for an unused time period. However, controlled-access techniques work better in

high-trafﬁc WLANs because without controlled access there are many collisions. Think

of a large class discussion in which the instructor selects who will speak (controlled

access) versus one in which any student can shout out a comment at any time.

6.4.3 Wireless Ethernet Frame Layout

An 801.11 data frame is illustrated on Figure 6.7. We notice two major differences

when we compare the 802.11 frame to the 802.3 frame used in wired Ethernet (see

Chapter 4). First, the wireless Ethernet frame has four address ﬁelds rather than two

like the wired Ethernet. These four address ﬁelds are source address, transmitter address,

receiver address, and destination address. The source and destination address have the

same meaning as in wired Ethernet. However, because every NIC has to communicate

via an access point (it cannot directly communication with another NIC), there is a need

to add the address of the access point and also any other device that might be needed to

transmit the frame. To do this, the transmitter and received address ﬁelds are used.

Second, there is new ﬁeld called Sequence Control that indicates how a large frame

is fragmented—split into smaller pieces. Recall that in wired networks this is done by

the Transport Layer, not the Data Link Layer. Moving the segmentation to the Data Link

Layer for wireless makes the transmission transparent to the higher layers. The price,

however, is less efﬁciency because of the size of the frame and thus also higher error rate.

6.4.4 Types of Wireless Ethernet

Wi-Fi is one of the fastest changing areas in networking. There are four types of Wi-Fi

but one, the latest version (802.11n) is dominant today. The other three versions are

obsolete, but may still be in use in some companies.

Frame

Control

(2 bytes)

Duration

(2 bytes)

Address 1

(6 bytes)

Address 2

(6 bytes)

Address 3

(6 bytes)

Sequence

Control

(2 bytes)

Address 4

(6 bytes)

Data

(0-2312

bytes)

FCS

(6 bytes)

FIGURE 6.7 A Wireless Ethernet Frame

216 CHAPTER 6 WIRED AND WIRELESS LOCAL AREA NETWORKS

802.11a IEEE 802.11a is an obsolete, legacy technology, and no new products are

being developed. Under perfect conditions, it provides eight channels of 54 Mbps each

with a maximum range of 50 meters or 150 feet. Speeds of 20 Mbps at 50 foot ranges

are more common in the face of interference such as drywall or brick walls.

802.11b IEEE 802.11b is another obsolete, legacy technology. Under perfect condi-

tions, it provides three channels of 11 Mbps each with a maximum range of 150 meters

or 450 feet, although in practice both the speed and range are lower.

802.11g IEEE 802.11g is another obsolete, legacy technology, but many organiza-

tions still use it. Under perfect conditions, it provides three channels of 54 Mbps each

with a maximum range of 150 meters or 450 feet, although in practice both the speed

and range are lower.

802.11n IEEE 802.11n is the latest version of Wi-Fi that most ﬁrms use (or are in

the process of installing). Under perfect conditions, it provides three channels of about

200 Mbps each with a maximum range of 150 meters or 450 feet, although in practice

both the speed and range are lower. It is also possible to conﬁgure APs to use different

frequency ranges to provide fewer channels that run at higher speeds up to 600 Mbps each.

IEEE 802.11n is backward compatible with 802.11b and 802.11g, so that laptops

that use these older versions can use an 802.11n access point. However, this backward

compatibility comes with a price. These old laptops become confused when 802.11n

devices operate at high speeds near them, so when an 802.11n AP detects the presence

of an 802.11b or 802.11g device, it prohibits newer laptops that use 802.11n devices

from operating at high speeds. Thus one old laptop will slow down all the other laptops

around it.

6.4.5 Security

Security is important to all networks and types of technology, but it is especially important

for wireless networks. With a WLAN, anyone walking or driving within the range of an

AP (even outside the ofﬁces) can begin to use the network.

Finding WLANs is quite simple. You just walk or drive around different ofﬁce

buildings with your WLAN-equipped client computer and see if it picks up a signal.

There are also many special-purpose software tools available on the Internet that will

enable you to learn more about the WLANs you discover, with the intent of helping

you to break into them. This type of wireless reconnaissance is often called wardriving

(see www.wardriving.com). Warchalking refers to the practice of writing symbols in

chalk on sidewalks and walls to indicate the presence of an unsecured WLAN (see

www.warchalking.org).

WEP One wireless security technique is Wired Equivalent Privacy (WEP). With

WEP, the AP requires the user to have a key in order to communicate with it. All data

sent to and from the AP is encrypted so that it can only be understood by computers

or devices that have the key (encryption is discussed in more detail in Chapter 10).

6.4 WIRELESS ETHERNET 217

If a computer does not have the correct WEP key, it cannot understand any messages

transmitted by the access point and the access point will not accept any data that is not

encrypted with the correct key.

The key can be manually typed into the client computer and into the AP. Although

this is not a major problem in a small WLAN, it does become challenging for large

WLANs. Imagine the management time required when a WEP key needs to be changed

in an organization with dozens of APs and hundreds of client computers (or hundreds of

APs and thousands of computers).

With Extensible Authentication Protocol (EAP), the WEP keys are produced

dynamically, much like the way in which a DHCP server is used to dynamically produce

IP addresses. When an AP using EAP ﬁrst discovers a new client computer, it requires

the user to login before it will communicate with the client computer. The user ID

and password supplied by the user are transmitted to a login server, and if the server

determines that they are valid the server generates a WEP key that will be used by the

AP and client computer to communicate for this session. Once the client logs out or

leaves the WLAN, the WEP is discarded and the client must login again and receive a

new WEP key.

Wired Equivalent Privacy has a number of serious weaknesses, and most experts

agree that a determined hacker can break into a WLAN that uses only WEP security. A

good way to think about WEP is that it is like locking your doors when you leave: It

won’t keep out a professional criminal but it will protect against a casual thief.

WPA Wi-Fi Protected Access (WPA) is a newer, more secure type of security. WPA

works in ways similar to WEP and EAP: every frame is encrypted using a key, and the

key can be ﬁxed in the AP like WEP or can be assigned dynamically as users login like

EAP. The difference is that the WPA key is longer than the WEP key and thus is harder

to break. More importantly, the key is changed for every frame that is transmitted to the

client. Each time a frame is transmitted, the key is changed.

802.11i 802.11i (also called WPA2) is the newest, most secure type of WLAN secu-

rity. It uses EAP to obtain a master key—in other words, the user logs in to a login

server to obtain the master key. Armed with this master key, the user’s computer and

the AP negotiate a new key that will be used for this session until the users leaves the

WLAN. 802.11i uses the Advanced Encryption Standard (AES) discussed in Chapter 10

as its encryption method.

MAC Address Filtering With MAC address ﬁltering, the AP permits the owner to

provide a list of MAC addresses (i.e., layer-2 addresses). The AP only processes frames

sent by computers whose MAC address is in the address list; if a computer with a MAC

address not in the list sends a frame, the AP ignores it. Unfortunately, this provides no

security against a determined hacker. There is software available that will change the

MAC address on a wireless NIC, so a determined hacker could use a packet sniffer (e.g.,

Wireshark) to discover a valid MAC address and then use the software to change his

MAC address to one the AP would accept. MAC address ﬁltering is like WEP; it will

protect against a casual thief, but not a professional.

218 CHAPTER 6 WIRED AND WIRELESS LOCAL AREA NETWORKS

6.4 MOOCHING WI-FI

MANAGEMENT

FOCUS

If you connect into someone else’s Wi-Fi network

and start using their Internet connection are you:

a. Guilty of stealing from the owner

because you haven’t paid them

b. Guilty of stealing from the ISP because

you haven’t paid them

c. Committing an unethical but not illegal

act

d. Really frugal, and not unethical

e. All of the above

According to the St Petersburg, Florida, police

department, the answer is a. They arrested a man

named Benjamin Smith for ‘‘willfully, knowingly,

and without authorization’’ accessing the network

of a homeowner while sitting in a car parked on

the street.

According to Verizon and most ISPs, which

explicitly prohibit sharing, the answer is b. ‘‘It’s

obviously not good for Verizon to have its services

given away for free, just as a cable company won’t

want someone funneling their cable connection

next door,’’ said a Verizon spokeswoman.

According to Miss Manners, the answer is c.

It’s not nice to use other people’s stuff without

asking their permission.

According to Jennifer Granick, executive direc-

tor of the Center for Internet and Society at

Stanford Law School, the answer is d. ‘‘Such use

[i.e., sharing] might be allowed or even encour-

aged [by the owner].’’ Unless the owner states

you can’t enter their network, how do you know

you’re not invited?

As Lee Tien, a senior staff attorney at the Elec-

tronic Frontier Foundation says ‘‘Right now, we

don’t have a way of saying ‘Even though my wire-

less signal is open, I’m saying you can’t use it.’’’

Until we do, the answer is e. So, tread carefully.

Don’t leave your WLAN unsecured or you may

be legally inviting others to use it as well as your

Internet connection. Likewise, don’t intentionally

enter someone else’s WLAN and use their Inter-

net connection or you might end up like Benjamin

Smith—spending the night in jail.

SOURCE: John Cox, ‘‘Mooching Wi-Fi,’’ Network

World, August 8, 2005, pp. 1, 49.

6.5 THE BEST PRACTICE LAN DESIGN

The past few years have seen major changes in LAN technologies (e.g., gigabit Ethernet,

high-speed wireless Ethernet). As technologies have changed and costs have dropped, so

too has our understanding of the best practice design for LANs.

One of the key questions facing network designers is the relationship between

Wi-Fi and wired Ethernet. The data rates for Wi-Fi have increased substantially with the

introduction of 802.11n, so they are similar to the data rates offered by 100Base-T wired

Ethernet. The key difference is that 100Base-T wired Ethernet using switches provides

100 Mbps to each user, whereas Wi-Fi shares its available capacity among every user on

the same AP, so as more users connect to the APs, the network gets slower and slower.

Wi-Fi is considerably cheaper than wired Ethernet because the largest cost of LANs

is not the equipment, but in paying someone to install the cables. The cost to install a

cable in an existing building is typically between $150 and $400 per cable, depending

on whether the cable will have to be run through drywall, brick, ceilings, and so on.

Installing cable in a new building during construction is cheaper, typically $50 to $100

per cable.

6.5 THE BEST PRACTICE LAN DESIGN 219

6.5 WILL WI-FI REPLACE WIRED LANS?

MANAGEMENT

FOCUS

As KPMG, one of the largest consulting ﬁrms in

the world, began to build a new 2800-person head-

quarters near Amsterdam, KPMG’s IT group real-

ized that their traditional wired network approach

would have required 18,000 cable runs, 55 chassis

switches, and 260 LAN switches. The up-front cost

was expected to exceed $6 million and the recur-

ring operating costs would run into the millions

annually as well.

KPMG began to wonder if there was a better

way. Could they build an entirely wireless network

that would meet their needs?

After careful analysis, KPMG decided they were

not ready to go completely wireless. However,

they decided to shift a substantial portion of their

traditionally wired users to wireless. They cut

their wired network by half and installed over

500 802.11n access points throughout the new

facility to provide complete coverage for data and

voice. The new network design cut the initial cost

by $2 million and reduced annual operating costs

by $750,000 per year.

The new design also delivered substantial

green beneﬁts. Access Points use about 5% of the

electricity that 48-port switches require for power

and cooling. By eliminating half the switches, the

new design eliminated over 350 metric tons of

carbon dioxide emissions each year.

SOURCE: ‘‘KPMG Netherlands Counts on Aruba for

Network Rightsizing,’’ Enterprise Case Study, Aruba

Networks, 2009.

Most organizations today are still installing traditional wired Ethernet for desktop

users and using Wi-Fi as overlay networks. They build the usual switched Ethernet

networks as the primary LAN, but they also install Wi-Fi for laptops. Some organizations

have begun experimenting with Wi-Fi by moving groups of users off the wired networks

onto Wi-Fi as their primary network to see whether Wi-Fi is suitable as a primary network.

Today, we still believe the best practice is to use wired Ethernet for the primary

LAN, with Wi-Fi as an overlay network. However, this may change. Stay tuned.

6.5.1 Wired Ethernet

Many organizations today install switched 100Base-T over category 5e wiring for their

wired LANs. It is relatively low cost and fast. Another good alternative is 10/100/1000

over cat 5e wires.

In the early days of LANs, it was common practice to install network cable wher-

ever it was convenient. Little long-term planning was done. The exact placement of the

cables was often not documented, making future expansion more difﬁcult—you had to

ﬁnd the cable before you could add a new user.

With today’s explosion in LAN use, it is critical to plan for the effective installation

and use of LAN cabling. The cheapest point at which to install network cable is during the

construction of the building; adding cable to an existing building can cost signiﬁcantly

more. Indeed, the costs to install cable (i.e., paying those doing the installation and

additional construction) are usually substantially more than the cost of the hubs and

switches, making it expensive to reinstall the cable if the cable plan does not meet the

organization’s needs.

Most buildings under construction today have a separate LAN cable plan,asthey

have plans for telephone cables and electrical cables. Each ﬂoor has a telecommunications

220 CHAPTER 6 WIRED AND WIRELESS LOCAL AREA NETWORKS

wiring closet that contains one or more network hubs or switches. Cables are run from

each room on the ﬂoor to this wiring closet.

6.5.2 Wireless Ethernet

Selecting the best practice wireless technology is usually simple. You pick the newest

one, cost permitting. Today, 802.11n is the newest standard, but in time, there will be a

new one.

Designing the physical WLAN is more challenging than designing a wired LAN

because the potential for radio interference means that extra care must be taken in the

placement of access points. With the design of LANs there is considerable freedom in the

placement of switches, subject to the maximum limits to the length of network cables.

In WLANs, however, the placement of the access points needs to consider both the

placement of other access points as well as the sources of interference in the building.

The physical WLAN design begins with a site survey. The site survey determines

the feasibility of the desired coverage, the potential sources of interference, the current

locations of the wired network into which the WLAN will connect, and an estimate of

the number of APs required to provide coverage. WLANs work very well when there is

a clear line of sight between the AP and the wireless computer. The more walls there

are between the AP and the computer, the weaker the wireless signal becomes. The type

and thickness of the wall also has an impact; traditional drywall construction provides

less interference than does concrete block construction.

An asccess point with an omnidirectional antenna broadcasts in all directions. Its

coverage area is a circle with a certain radius. Wi-Fi has a long range, but real-world

tests of Wi-Fi in typical ofﬁce environments have shown that data rates slow down

dramatically when the distance from a laptop to the AP exceeds 50 feet. Therefore, many

wireless designers use a radius of 50 feet when planning traditional ofﬁce environments,

which ensures access high-quality coverage. It is also expensive, because many APs will

need to be purchased. Costs may be reduced by using a longer radius (e.g., 100 feet), so

that fewer APs are needed, but this may result in slower data rates.

One may design wireless LANs using this 50-foot radius circle, but because most

buildings are square, it is usually easier to design using squares. Figure 6.8 shows that

70 foot edge

50 foot

radius

70 foot

edge

FIGURE 6.8 Design parameters for Wi-Fi

access point range

6.5 THE BEST PRACTICE LAN DESIGN 221

450 ft

150 ft

300 ft

FIGURE 6.9 AWi-Fi

design (the numbers indicate the

channel numbers)

a 50-foot radius translates into a square that is approximately 70 feet on each edge. For

this reason, most designers plan wireless LANs using 50- to 75-foot squares, depending

on the construction of the building: smaller squares in areas where there are more walls

that can cause more interference and larger squares in areas with fewer walls.

Figure 6.9 shows a sample building that has two parts. The lower left corner is a

150 feet × 150 feet square, while the rest of the building is a 150 ft × 450 feet rectangle.

Let’s assume that the large rectangle part is an open ofﬁce environment, while the smaller

part uses drywall. If we put two rows of APs in the large rectangle part, we could probably

space them so that each AP covered a 75-foot square. This would take a total of 12 APs

for this area (see Figure 6.9). This same spacing probably won’t work for the small area

with drywall, so we would probably design using 50-foot squares, meaning we need 9

APs in this area (see Figure 6.9).

When designing a wireless LAN it is important to ensure that the APs don’t interfere

with each other. If all APs transmitted on the same frequency, the transmissions of one AP

would interfere with another AP where their signals overlapped—just like what happens

on your car radio when two stations are in the same frequency. Therefore, each AP is

set to transmit on a different channel, very much like the different channels on your TV.

Figure 6.8 shows how we could set the APs to the three commonly used channels (1, 6,

and 11) so that there is minimal overlap between APs using the same channel.

After the initial design is complete, a site survey is done using a temporary AP and

a computer or device that can actually measure the strength of the wireless signal. The

temporary AP is installed in the area as called for in the initial design, and the computer

or device is carried throughout the building measuring the strength of the signal. Actually

measuring the strength of the signal in the environment is far more accurate than relying

on estimated ranges.

Design becomes more difﬁcult in a multistory building because the signals from

the APs travel up and down as well as in all horizontal directions. The design must

include the usual horizontal mapping but also an added vertical mapping to ensure that

222 CHAPTER 6 WIRED AND WIRELESS LOCAL AREA NETWORKS

Channel 1

Channel 6

Channel 11

FIGURE 6.10 AWi-Fi

design in the three dimensions

(the numbers indicate the channel

numbers)

APs on different ﬂoors do not interfere with one another (Figure 6.10). Because ﬂoors

are usually thicker than walls, signals travel further horizontally than vertically, making

design a bit more difﬁcult. It becomes even more difﬁcult if your set of ﬂoors in a large

ofﬁce tower is surrounded by APs of other companies. You have to design your network

not to interfere with theirs.

6.5.3 Designing for SOHO Environments

Most of what we have discussed so far has focused on network design in large enterprises.

What about LAN design for SOHO environments? SOHO environments can be small

versions of enterprise designs, or can take a very different approach.

Figure 6.11a shows a SOHO LAN designed as a small enterprise design that pro-

vides both wired and wireless Ethernet (it’s in Alan’s house). Virtually all of the rooms

in the house are wired with 100Base-T Ethernet over Cat 5e cable, which terminates in a

24-port patch panel. You can see from the ﬁgure that only ﬁve of the rooms are actually

wired from the patch panel into the 16-port switch; one of those wires connects the AP

mounted in an upstairs hallway (not shown) that provides wireless access throughout the

house and onto the back deck and gazebo. There is a separate router and cable modem.

The AP, switch, and router are all Cisco or Linksys equipment and are the original 2001

equipment, which still work well. The cable modem is an off-brand provided by the ISP

and has broken and been replaced every three years.

Figure 6.11b shows a more modern—and probably more common—SOHO LAN

that provides only wireless access (it’s in Alexandra’s house). This has a cable modem

that connects into a wireless router; the wireless router is a wireless AP, a router, and a

switch for wired Ethernet all in one box. This network is simpler and cheaper because

it contains fewer devices and is used only for wireless access. Alexandra doesn’t have

a desktop computer at home, but she could easily connect one if she wanted by adding

a wireless NIC into a desktop; a 802.11n WLAN provides ample capacity for a small

SOHO network.

6.6 IMPROVING LAN PERFORMANCE 223

Patch Panel

Cable Modem

Switch

Router

(a) Alan’s home network

(b) Alexandra’s home network

Wireless

Router

Cable

Modem

FIGURE 6.11 SOHO LAN

designs

6.6 IMPROVING LAN PERFORMANCE

When LANs had only a few users, performance was usually very good. Today, however,

when most computers in an organization are on LANs, performance can be a problem.

Performance is usually expressed in terms of throughput (the total amount of user data

transmitted in a given time period). In this section, we discuss how to improve throughput.

We focus on dedicated-server networks because they are the most commonly used type

of LANs, but many of these concepts also apply to peer-to-peer networks.

To improve performance, you must locate the bottleneck, the part of the network

that is restricting the data ﬂow. Generally speaking, the bottleneck will lie in one of

224 CHAPTER 6 WIRED AND WIRELESS LOCAL AREA NETWORKS

6.3 ERROR CONTROL IN WIRED ETHERNET

TECHNICAL

FOCUS

Ethernet provides a strong error control method

using stop and wait ARQ with a CRC-32 error detec-

tion ﬁeld (see Chapter 4). However, the normal way

of installing wired Ethernet doesn’t use stop and wait

ARQ.

In the early days of Ethernet, LAN environments

were not very reliable, so error control was impor-

tant. However, today’s wired Ethernet LANs are very

reliable; errors seldom occur. Stop and wait ARQ

uses considerable network capacity because every

time a frame is transmitted, the sender must stop

and wait for the receiver to send an acknowledg-

ment. By eliminating the need to stop and wait and

the need to send acknowledgments, Ethernet can

signiﬁcantly improve network performance—almost

doubling the number of messages that can be

transmitted in the same time period. Ethernet does

still add the CRC and does still check it for errors, but

any frame with an error is simply discarded.

If Ethernet doesn’t provide error control, then

higher layers in the network model must. In gen-

eral, TCP is conﬁgured to provide error control by

using continuous ARQ (see Chapter 5) to ensure that

all frames that have been sent are actually received

at the ﬁnal destination. If a frame with an error is dis-

carded by Ethernet, TCP will recognize that a frame

has been lost and ask the sender to retransmit. This

moves responsibility for error control to the edges of

the network (i.e., the sender and receiver) rather than

making every computer along the way responsible

for ensuring reliable message delivery.

two places. The ﬁrst is the network server. In this case, the client computers have no

difﬁculty sending requests to the network server, but the server lacks sufﬁcient capacity

to process all the requests it receives in a timely manner. The second location is the

network circuit, connecting the LAN to the corporate BN. In this case, the server can

easily process all the client requests it receives, but the circuit lacks enough capacity to

transmit all the requests to the server.

The ﬁrst step in improving performance, therefore, is to identify whether the bot-

tleneck lies in the circuit or the server. To do so, you simply watch the utilization of the

server during periods of poor performance. If the server utilization is high (e.g., 60 to 100

percent), then the bottleneck is the server; it cannot process all the requests it receives

in a timely manner. If the server utilization is low during periods of poor performance

(e.g., 10 to 40 percent), then the problem lies with the network circuit; the circuit cannot

transmit requests to the server as quickly as necessary. Things become more difﬁcult if

utilization is in the midrange (e.g., 40 to 60 percent). This suggests that the bottleneck

may shift between the server and the circuit depending on the type of request, and it

suggests that both should be upgraded to provide the best performance.

Now we will focus attention on ways to improve the server and the circuit to

remove bottlenecks. These actions address only the supply side of the equation—that is,

increasing the capacity of the LAN as a whole. The other way to reduce performance

problems is to attack the demand side: reduce the amount of network use by the clients,

which we also discuss. Figure 6.12 provides a performance checklist.

6.6.1 Improving Server Performance

Improving server performance can be approached from two directions simultaneously:

software and hardware.