FitzGerald J., Dennis A., Durcikova A. Business Data Communications and Networking
Подождите немного. Документ загружается.
HANDS-ON ACTIVITY 7A 265
CASE STUDY
NEXT-DAY AIR SERVICE
See the Web site.
HANDS-ON ACTIVITY 7A
Network Mapping
A backbone network connects one or more LANs to each
other and usually to the Internet or to another backbone
that eventually leads to the Internet. Each of these back-
bones usually connects many computers.
Network mapping software enables you to generate
a map of the computers on all the LANs connected
to a backbone. There are many good network mapping
packages. Two of my favorites are LANState and LAN
Surveyor. LANState is simpler to use but works best for
small networks. LAN Surveyor is more complex, but can
map large networks.
Both work in the same way. They use the ping com-
mand (see Chapter 5) to send IMCP requests to all possi-
ble IP addresses in any range you specify. Any computer
that responds is added to the map.
Mapping a Small Network
The first step is to download and install LANState. A
demo version of the software is available free of charge
from 10-Strike Software (www.10-strike.com/lanstate).
You begin by creating a new network map (choose File
Create). Then use the Network Map Creation Wizard and
choose to Scan an IP-address range. You will be asked
to enter an address range. Choose some range, ideally
the address range of a small network. I choose to use my
home network range (192.168.1.1 through 192.168.1.254).
When the scan is complete you will see a list of computers.
Click Finish to see a map of these computers.
LANState does not do a good job of drawing a map,
but you can rearrange the computers by dragging and
dropping them. You can also add lines to make the map
look more like a network diagram. Figure 7.17 shows the
small network in my house. I have a router (192.168.1.1)
that connects a number of computers to the Internet. I also
have a wireless access point (192.168.1.100) and a music
server (192.168.1.52). When I did this map, three com-
puters were turned on and responded to LANState’s pings
(192.168.1.102, 192.168.1.103, 192.168.1.111). Comput-
ers and devices that are not turned on do not respond
to the pings and therefore are not mapped. Since I use
dynamic addressing, the addresses of my computers will
change every time I turn them on.
You can also left click on any device and choose Sys-
tem Information and General to learn more about that
device. Figure 7.17 also shows the information about
my son’s computer (192.168.1.103). It shows the MAC
address (i.e., the Ethernet address), the card manufacturer,
and Windows workgroup peer-to-peer network informa-
tion (i.e., application layer address) for this computer.
Mapping a Large Network
The first step is to download and install LAN Surveyor.
A demo version of the software is available free of
charge from Solarwinds (www.solarwinds.com/products/
lansurveyor). Installing the software and setting it up to
run is more complex, so be sure to follow the setup and
configuration instructions.
You begin by creating a new network map (choose
File New). You will be asked to enter an address range.
Choose some range, ideally the address range of a large
network. I choose to use part of the Indiana Univer-
sity network (129.79.1.1 through 129.79.1.254). There is
no rule preventing you from scanning anyone’s network,
but many companies (and individuals) feel that scanning
their networks is an invasion of privacy, so scan care-
fully. When the scan is complete, you will see a map of
computers.
My scan of this one small part of the Indiana Univer-
sity network found 124 computers. Figure 7.18 shows a
266 CHAPTER 7 BACKBONE NETWORKS
partial list of the computers and their IP addresses and
host names (i.e., application layer addresses).
Deliverables
1. Use the 10-Strike Software to draw a map of
you home network. Describe each component on
your map just like the example in the textbook
shows.
2. Use the System Information and provide addi-
tional information (e.g., MAC address, card man-
ufacturer) about at least two devices on your net-
work.
FIGURE 7.17 Network mapping with LANState
HANDS-ON ACTIVITY 7A 267
FIGURE 7.18 Network mapping with LAN Surveyor
CHAPTER8
WIDE AREA NETWORKS
The Three Faces of Networking
The Three Faces of Networking
Fundamental Concepts Network Technologies
Wide Area Network
Network Management
S
ecurit
y
N
e
t
w
o
r
k
D
e
s
i
g
n
N
e
t
w
o
r
k
M
a
n
a
g
e
m
e
n
t
Application layer
Network layer
Data Link layer
Physical layer
LAN
WLAN
Transport layer
Backbone
WAN
Internet
CHAPTER OUTLINE 269
MOST ORGANIZATIONS do not build their own metropoli-
tan or long-distance communication circuits, preferring instead to lease
them from common carriers or to use the Internet. Therefore, this chapter
focuses on the WAN architectures and telecommunications services offered
by common carriers for use in WANs, not the underlying technology that the
carriers use to provide them. We discuss the four principal types of WAN
services that are available: circuit-switched services, dedicated-circuit ser-
vices, packet-switched services, and virtual private network (VPN) services.
We conclude by discussing how to improve WAN performance and how to
select services to build WANs.
OBJECTIVES
▲
Understand circuit-switched services and architectures
Understand dedicated-circuit services and architectures
Understand packet-switched services and architectures
Understand Internet-based VPN services and architectures
Understand the best practice recommendations for WAN design
Be familiar with how to improve WAN performance
CHAPTER OUTLINE
▲
8.1 INTRODUCTION
8.2 CIRCUIT-SWITCHED NETWORKS
8.2.1 Basic Architecture
8.2.2 Plain Old Telephone Service
8.2.3 Integrated Services Digital Network
(ISDN)
8.3 DEDICATED-CIRCUIT NETWORKS
8.3.1 Basic Architecture
8.3.2 T Carrier Services
8.3.3 Synchronous Optical Network
8.4 PACKET-SWITCHED NETWORKS
8.4.1 Basic Architecture
8.4.2 Asynchronous Transfer Mode
8.4.3 Frame Relay
8.4.4 Ethernet Services
8.4.5 Multi-Protocol Label Switching
8.5 VIRTUAL PRIVATE NETWORKS
8.5.1 Basic Architecture
8.5.2 VPN Types
8.5.3 How VPNs Work
8.6 THE BEST PRACTICE WAN DESIGN
8.7 IMPROVING WAN PERFORMANCE
8.7.1 Improving Device Performance
8.7.2 Improving Circuit Capacity
8.7.3 Reducing Network Demand
8.8 IMPLICATIONS FOR MANAGEMENT
270 CHAPTER 8 WIDE AREA NETWORKS
8.1 INTRODUCTION
Wide area networks (WANs) typically run long distances connecting different offices in
different cities or countries. Some WANs run much shorter distances, connecting different
buildings in the same city. Most organizations do not own the land across which WANs
are built, so instead they rent or lease circuits from common carriers—private companies
such as AT&T, Bell Canada, Sprint, and BellSouth that provide communication services
to the public. As a customer, you do not lease physical cables per se; you simply lease
circuits that provide certain transmission characteristics. The carrier decides whether it
will use twisted-pair, coaxial, fiber optics, or other media for its circuits.
Common carriers are profit oriented, and their primary products are services for
voice and data transmissions, both over traditional wired circuits as well as cellular ser-
vices. Common carriers often supply a broad range of computer-based services, such as
the manufacturing and marketing of specialized communication hardware and software.
A common carrier that provides local telephone services (e.g., BellSouth) is commonly
called a local exchange carrier (LEC), whereas one that provides long-distance ser-
vices (e.g., AT&T) is commonly called an interexchange carrier (IXC). As the LECs
move into the long-distance market and IXCs move into the local telephone market, this
distinction may disappear.
In this chapter, we examine the WAN architectures and technologies from the
viewpoint of a network manager, rather than that of a common carrier. We focus less
on internal operations and how the specific technologies work, and more on how these
services are offered to network managers and how they can be used to build networks
because network managers are less concerned with how the services work and more
concerned with how they can use them effectively.
Likewise, we will focus on WAN services in North America because the majority
of our readers are in North America. Although there are many similarities in the way
data communications networks and services have evolved in different countries, there
also are many differences. Most countries have a federal government agency that
regulates data and voice communications. In the United States, the agency is the Federal
Communications Commission (FCC); in Canada, it is the Canadian Radio-Television
and Telecommunications Commission (CRTC). Each state or province also has
its own public utilities commission (PUC) to regulate communications within its
borders.
We will discuss three services that use common carrier networks (circuit-switched
services, dedicated-circuit services, and packet-switched services) and one that uses the
public Internet (virtual private network). The first three enable the customer to more
precisely design and manage the WAN and offer more reliable services, so these services
are most commonly chosen by large organizations that view the WAN as an important
part of their business operations. The public Internet is usually much cheaper than these
services, but less reliable, so it is usually only attractive to very small organizations that
are more price sensitive.
8.2 CIRCUIT-SWITCHED NETWORKS 271
8.2 CIRCUIT-SWITCHED NETWORKS
Circuit-switched networks are the oldest and simplest approach to WAN circuits. These
services operate over the public switched telephone network (PSTN); that is, the tele-
phone networks operated by the common carriers such as AT&T, BellSouth, and so on.
When you telephone someone, you are using the PSTN. The first service we discuss is
the dial-up service you use when you call an ISP with a dial-up modem—but first we
need to discuss the basic architecture shared by all circuit-switched services.
At this point, you’re probably wondering who uses dial-up any more? And you’d be
right. About 40% of the U.S. population in 2001 used dial-up services, but today that has
dropped to about 5%. However, many organizations still use dial-up services for specialty
services such as videoconferencing, although we expect these also will fade in time.
8.2.1 Basic Architecture
Circuit-switched services use a cloud architecture. The users lease connection points
(e.g., telephone lines) into the common carrier’s network, which is called the cloud
1
(Figure 8.1). A person (or computer) dials the telephone number of the destination com-
puter and establishes a temporary circuit between the two computers. The computers
LAN
LAN
LAN
LAN
LAN
Common
carrier
network
Modem
Modem
Modem
Modem
Modem
FIGURE 8.1 Dialed circuit
services. LAN = local area network
1
It is called a cloud because what happens inside the common carrier’s network is hidden from view. Network
managers really don’t care how the common carrier switches the circuit inside their network, just as long as
the network is fast, accurate, and reliable.
272 CHAPTER 8 WIDE AREA NETWORKS
exchange data, and when the task is complete, the circuit is disconnected (e.g., by hanging
up the phone).
This architecture is very flexible. Circuits can be established as needed between
any computers attached to the cloud at any point. However, data can be transmitted only
while a circuit is established, and only to the one location it connects to. If a computer
needs to send data to a number of other locations, a series of temporary circuits must be
established with and later disconnected from each location, one after another. In general,
only a limited number of circuits can be established from or to any one location at a
time (e.g., each location has only so many telephone lines).
Cloud-based designs are simpler for the organization because they move the burden
of network design and management inside the cloud from the organization to the common
carrier. Network managers do not need to worry about the amount of traffic sent between
each computer; they just need to specify the amount of traffic entering and leaving
each computer and buy the appropriate size and number of connections into the PSTN.
However, this comes at a price. Cloud-based designs can be more expensive because
users must pay for each connection into the network and pay on the basis of the amount
of time each circuit is used. Cloud-based designs are often used when network managers
are uncertain of network demand, particularly in a new or rapidly growing network.
There are two basic types of switched-circuit services in use today: POTS and
ISDN.
8.2.2 Plain Old Telephone Service
Plain old telephone service (POTS) is the name for the dial-up services you or your
parents used at one time. To use POTS, you need to lease a circuit into the network (i.e.,
a telephone line) and install special equipment (i.e., a modem) to enable your computer
to talk to the PSTN. To transfer data to and from another computer on the network,
you instruct your modem to dial the other computer’s telephone. Once the modem in
your computer connects to the modem at the other end, you can transfer data back and
forth. When you are done, you hang up and can then call another computer if you wish.
Today, POTS is most commonly used to connect to the Internet, but you can also use it
to communicate directly with a private non-Internet server.
Plain old telephone service may use different circuit paths between the two com-
puters each time a number is dialed. Some circuits have more noise and distortion than
others, so the quality and maximum data transmission rate can vary. Typical data rates
are between 33 kbps and 56 kbps.
8.2.3 Integrated Services Digital Network (ISDN)
The first generation of integrated services digital network (ISDN) combines voice, video,
and data over the same digital circuit. Because there is a newer version of ISDN, the original
version is occasionally called narrowband ISDN, but we will just use the term ISDN. ISDN
is widely available from a number of common carriers in North America.
To use ISDN, users first need to lease connection points in the PSTN, which
are telephone lines just like POTS. Next, they must have special equipment to connect
their computers (or networks) into the PSTN. Users need an ISDN network terminator
(NT-1 or NT-2) that functions much like a hub, and a NIC (called a terminal adapter
8.3 DEDICATED-CIRCUIT NETWORKS 273
[TA] or even an “ISDN modem”) in all computers attached to the NT-1/NT-2. In most
cases, the ISDN service appears identical to the regular dialed telephone service, with
the exception that usually (but not always) each device attached to the NT-1/NT-2 needs
a unique service profile identifier (SPID) to identify it. To connect to another computer
using ISDN, you dial that computer’s telephone number using the ISDN NIC in much
the same way as you would with a modem on a regular telephone line. There are two
types of ISDN, which we will discuss next.
Basic Rate Interface Basic rate interface (BRI) (sometimes called basic access
service or 2B+D) provides a communication circuit with two 64-Kbps digital transmis-
sion channels (called B channels) and one 16-Kbps control signaling channel (called
a D channel). The two B channels handle digitized voice, data, and image transmis-
sions, providing a total of 128 Kbps. The D channel is used for control messages such
as acknowledgments, call setup and termination, and other functions such as automatic
number identification. Some common carriers sell just one single 64-Kbps channel to
those customers needing less capacity than full BRI.
Primary Rate Interface Primary rate interface (PRI) (also called primary access
service or 23B+D) is typically offered to commercial customers. It consists of 23 64-Kbps
B channels plus 1 64-Kbps D channel. PRI has almost the same capacity as a T1 circuit
(1.544 Mbps). In Europe, PRI is defined as 30 B channels plus 1 D channel, making
interconnection between America and Europe difficult.
8.3 DEDICATED-CIRCUIT NETWORKS
There are two main problems with POTS and ISDN circuit-switched networks. First,
each time you want to transfer data, you need to make a separate connection. Second,
the data rates on these circuits are low, ranging from 56 Kbps for dialed POTS circuits to
128 Kbps or 1.5 Mbps for ISDN circuits. One alternative is to establish a dedicated-circuit
network, in which the user leases circuits from the common carrier for his or her exclusive
use 24 hours per day, 7 days per week. Dedicated-circuit networks are sometimes called
private line services.
8.3.1 Basic Architecture
With a dedicated-circuit network, you lease circuits from common carriers. All connec-
tions are point to point, from one building in one city to another building in the same or
a different city. The carrier installs the circuit connections at the two end points of the
circuit and makes the connection between them. The circuits still run through the com-
mon carrier’s cloud, but the network behaves as if you have your own physical circuits
running from one point to another (Figure 8.2).
Once again, the user leases the desired circuit from the common carrier (specifying
the physical end points of the circuit) and installs the equipment needed to connect
computers and devices (e.g., routers or switches) to the circuit. This equipment may
include multiplexers or a channel service unit (CSU) and/or a data service unit (DSU);
a CSU/DSU is the WAN equivalent of a NIC in a LAN.
274 CHAPTER 8 WIDE AREA NETWORKS
Common
carrier
network
LAN
Modem
Modem
Modem
Modem
Modem
MUX
MUX
LAN
CSU/DSU
CSU/DSU
LAN
LAN
LAN
FIGURE 8.2 Dedicated-circuit services. CSU = channel service unit; DSU = data service
unit; MUX = multiplexer
Unlike circuit-switched services that typically use a pay-per-use model, dedicated
circuits are billed at a flat fee per month, and the user has unlimited use of the circuit.
Once you sign a contract, making changes can be expensive because it means rewiring
the buildings and signing a new contract with the carrier. Therefore, dedicated circuits
require more care in network design than do switched circuits, both in terms of locations
and the amount of capacity you purchase.
There are three basic architectures used in dedicated-circuit networks: ring, star,
and mesh. In practice, most networks use a combination of architectures. For example, a
distributed star architecture has a series of star networks that are connected by a mesh
or ring architecture.
Ring Architecture A ring architecture connects all computers in a closed loop with
each computer linked to the next (Figure 8.3). The circuits are full-duplex or half-duplex
circuits, meaning that messages flow in both directions around the ring. Computers in
the ring may send data in one direction or the other, depending on which direction is the
shortest to the destination.
One disadvantage of the ring topology is that messages can take a long time to travel
from the sender to the receiver. Messages usually travel through several computers and
circuits before they reach their destination, so traffic delays can build up very quickly if