Network+ 2005 In Depth (P10) doc

net-Topologies and Access MethodsChapter 6 After reading this chapter and completing the exercises, you will be able to: ■ Describe the basic and hybrid LAN physical topologies, and thei

routing switch—See Layer 3 switch.

runt—An erroneously shortened packet.

single point of failure—A device or connection on a network that, were it to fail, could cause

the entire network to stop functioning

SOHO (small office-home office) router—A router designed for use on small office or home

office networks SOHO routers typically have no more than eight data ports and do not offeradvanced features such as traffic prioritization, network management, or hardware redun-dancy

stackable hub—A type of hub designed to be linked with other hubs in a single

telecommu-nications closet Stackable hubs linked together logically represent one large hub to the work

net-standalone hub—A type of hub that serves a workgroup of computers that are separate from

the rest of the network, also known as a workgroup hub

static routing—A technique in which a network administrator programs a router to use

spe-cific paths between nodes Because it does not account for occasional network congestion, failedconnections, or device moves, static routing is not optimal

store and forward mode—A method of switching in which a switch reads the entire data frame

into its memory and checks it for accuracy before transmitting it Although this method is moretime-consuming than the cut-through method, it allows store and forward switches to trans-mit data more accurately

switch—A connectivity device that logically subdivides a network into smaller, individual

col-lision domains A switch operates at the Data Link layer of the OSI Model and can interpretMAC address information to determine whether to filter (discard) or forward packets itreceives

system bus—See bus.

uplink port—A port on a connectivity device, such as a hub or switch, used to connect it to

another connectivity device

USB (universal serial bus) port—A standard external bus that can be used to connect

multi-ple types of peripherals, including modems, mice, and NICs, to a computer Two USB dards exist: USB 1.1 and USB 2.0 Most modern computers support the USB 2.0 standard

stan-virtual local area network—See VLAN.

VLAN (virtual local area network)—A network within a network that is logically defined by

grouping its devices’ switch ports in the same broadcast domain A VLAN can consist of anytype of network node in any geographic location and can incorporate nodes connected to dif-ferent switches

workgroup hub—See standalone hub.

Review Questions

1. _ are connectivity devices that enable a workstation,

server, printer, or other node to receive and transmit data over the network media

a. Network interface cards

b. Adapter cards

c. Routing protocols

d. Ports

2. A computer’s _ is the circuit, or signaling pathway, used

by the motherboard to transmit data to the computer’s components, including its

memory, processor, hard disk, and NIC

a. port

b. bus

c. switch

d. router

3. _ is a standard interface used to connect multiple types

of peripherals, including modems, mice, audio players, and NICs

a. OSPF

b. PCI

c. FireWire

d. USB

4. _ are physically designed to be linked with other hubs in

a single telecommunications closet

6. True or false? All peripheral devices are connected to a computer’s motherboard via anexpansion slot or peripheral bus.

7. True or false? A device’s base I/O port cannot be used by any other device

8. True or false? A repeater is limited in function but not in scope

9. True or false? A switch running in cut-through mode will read a frame’s header anddecide where to forward the data before it receives the entire packet

10. True or false? A router is a multiport connectivity device that directs data betweennodes on a network

11. A(n) _ is a small, removable piece of plastic that tains a metal receptacle

con-12. A(n) _ is a message to the computer that instructs it tostop what it is doing and pay attention to something else

13. The _ indicates, in hexadecimal notation, the area ofmemory that the NIC and CPU will use for exchanging, or buffering, data

14. A(n) _ is a connector that plugs into a port, such as aserial or parallel or an RJ-45 port, and crosses over the transmit line to the receive line

so that outgoing signals can be redirected into the computer for testing

15. A(n) _ is a logically or physically distinct Ethernet work segment on which all participating devices must detect and accommodate datacollisions

net-Topologies and Access Methods

Chapter 6

After reading this chapter and completing the exercises, you will be able to:

■ Describe the basic and hybrid LAN physical topologies, and their uses, advantages, and disadvantages

■ Describe the backbone structures that form the foundation for most LANs

■ Compare the different types of switching used in data transmission

■ Understand the transmission methods underlying Ethernet,Token Ring, FDDI, and ATM networks

■ Describe the characteristics of different wireless network technologies, including Bluetooth and the three IEEE 802.11 standards

unseen, when designing a network This chapter details some basic elements of network tecture: physical and logical topologies These elements are crucial to understanding network-ing design, troubleshooting, and management, all of which are discussed later in this book.

archi-In this chapter, you will also learn about the most commonly used network access methods:Ethernet, Token Ring, FDDI, ATM, and popular wireless access methods Once you masterthe physical and logical fundamentals of network architecture, you will have all the tools nec-essary to design a network as elegant as the Taj Mahal

Simple Physical Topologies

A physical topology is the physical layout, or pattern, of the nodes on a network It depicts a

network in broad scope; that is, it does not specify device types, connectivity methods, oraddressing schemes for the network Physical topologies are divided into three fundamentalgeometric shapes: bus, ring, and star These shapes can be mixed to create hybrid topologies.Before you design a network, you need to understand physical topologies, because they are inte-gral to the type of network (for example, Ethernet or Token Ring), cabling infrastructure, andtransmission media you use You must also understand a network’s physical topology to trou-bleshoot its problems or change its infrastructure A thorough knowledge of physical topolo-gies is necessary to obtain Network+ certification

Physical topologies and logical topologies (discussed later) are two different ing concepts You should be aware that when used alone, the word “topology” often

network-refers to a network’s physical topology.

On a bus topology network, devices share the responsibility for getting data from one point toanother Each node on a bus network passively listens for data directed to it When one nodewants to transmit data to another node, it broadcasts an alert to the entire network, informingall nodes that a transmission is being sent; the destination node then picks up the transmis-sion Nodes other than the sending and receiving nodes ignore the message.

For example, suppose that you want to send an instant message to your friend Diane, who worksacross the hall, asking whether she wants to have lunch with you You click the Send buttonafter typing your message, and the data stream that contains your message is sent to your NIC.Your NIC then sends a message across the shared wire that essentially says, “I have a messagefor Diane’s computer.” The message passes by every NIC between your computer and Diane’scomputer until Diane’s computer recognizes that the message is meant for it and responds byaccepting the data

At the ends of each bus network are 50-ohm resistors known as terminators Terminators stop

signals after they have reached the end of the wire Without these devices, signals on a busnetwork would travel endlessly between the two ends of the network—a phenomenon known

as signal bounce—and new signals could not get through To understand this concept,

imag-ine that you and a partner, standing at opposite sides of a canyon, are yelling to each other.When you call out, your words echo; when your partner replies, his words also echo Now imag-ine that the echoes never fade After a short while, you could not continue conversing becauseall of the previously generated sound waves would still be bouncing around, creating too muchnoise for you to hear anything else On a network, terminators prevent this problem by halt-ing the transmission of old signals In some cases, a hub provides termination for one end of asegment A bus network must also be grounded at one end to help remove static electricitythat could adversely affect the signal Figure 6-1 depicts a terminated bus network

Chapter 6 247

SIMPLE PHYSICAL TOPOLOGIES

FIGURE 6-1 A terminated bus topology network

NET+

1.1

Although networks based on a bus topology are relatively inexpensive to set up, they do not scalewell As you add more nodes, the network’s performance degrades Because of the single-chan-nel limitation, the more nodes on a bus network, the more slowly the network will transmit anddeliver data For example, suppose a bus network in your small office supports two workstationsand a server, and saving a file to the server takes two seconds During that time, your NIC firstchecks the communication channel to ensure it is free, then issues data directed to the server.When the data reaches the server, the server accepts it Suppose, however, that your businessexperiences tremendous growth, and you add five workstations during one weekend The fol-lowing Monday, when you attempt to save a file to the server, the save process might take fiveseconds, because the new workstations may also be using the communications channel, and yourworkstation may have to wait for a chance to transmit As this example illustrates, a bus topol-ogy is rarely practical for networks with more than a dozen workstations.

Bus networks are also difficult to troubleshoot, because it is a challenge to identify fault tions To understand why, think of the game called “telephone,” in which one person whispers

loca-a phrloca-ase into the eloca-ar of the next person, who whispers the phrloca-ase into the eloca-ar of loca-another son, and so on, until the final person in line repeats the phrase aloud The vast majority of thetime, the phrase recited by the last person bears little resemblance to the original phrase.When the game ends, it’s hard to determine precisely where in the chain the individual errorscropped up Similarly, errors may occur at any intermediate point on a bus network, but at thereceiving end it’s possible to tell only that an error occurred Finding the source of the errorcan prove very difficult

per-A final disadvantage to bus networks is that they are not very fault-tolerant, because a break

or a defect in the bus affects the entire network As a result, and because of the other vantages associated with this topology, you will rarely see a network run on a pure bus topol-ogy You may, however, encounter hybrid topologies that include a bus component

disad-Ring

In a ring topology, each node is connected to the two nearest nodes so that the entire network

forms a circle, as shown in Figure 6-2 Data is transmitted clockwise, in one direction rectionally), around the ring Each workstation accepts and responds to packets addressed to

(unidi-it, then forwards the other packets to the next workstation in the ring Each workstation acts

as a repeater for the transmission The fact that all workstations participate in delivery makes

the ring topology an active topology This is one way a ring topology differs from a bus

topol-ogy A ring topology also differs in that it has no “ends” and data stops at its destination Inmost ring networks, twisted-pair or fiber-optic cabling is used as the physical medium.The drawback of a simple ring topology is that a single malfunctioning workstation can dis-able the network For example, suppose that you and five colleagues share a pure ring topologyLAN in your small office You decide to send an instant message to Thad, who works threeoffices away, telling him you found his lost glasses Between your office and Thad’s office aretwo other offices, and two other workstations on the ring Your instant message must passthrough the two intervening workstations’ NICs before it reaches Thad’s computer If one ofthese workstations has a malfunctioning NIC, your message will never reach Thad

NET+

1.1

In addition, just as in a bus topology, the more workstations that must participate in datatransmission, the slower the response time Consequently, pure ring topologies are not very flex-ible or scalable Contemporary LANs rarely use pure ring topologies.

Star

In a star topology, every node on the network is connected through a central device, such as

a hub or switch Figure 6-3 depicts a typical star topology Star topologies are usually builtwith twisted-pair or fiber-optic cabling Any single cable on a star network connects only twodevices (for example, a workstation and a hub), so a cabling problem will affect two nodes atmost Devices such as workstations or printers transmit data to the hub, which then retrans-mits the signal to the network segment containing the destination node

Star topologies require more cabling than ring or bus networks They also require more figuration However, because each node is separately connected to a central connectivity device,they are more fault-tolerant A single malfunctioning workstation cannot disable an entire starnetwork A failure in the central connectivity device can take down a LAN segment, though.Because they include a centralized connection point, star topologies can easily be moved, iso-lated, or interconnected with other networks; they are therefore scalable For this reason, andbecause of their fault tolerance, the star topology has become the most popular fundamentallayout used in contemporary LANs Single star networks are commonly interconnected withother networks through hubs and switches to form more complex topologies Most Ethernetnetworks are based on the star topology

con-Chapter 6 249

SIMPLE PHYSICAL TOPOLOGIES

FIGURE 6-2 A typical ring topology network

NET+

1.1

Star networks can support a maximum of only 1024 addressable nodes on a logical network.For example, if you have a campus with 3000 users, hundreds of networked printers, and scores

of other devices, you must strategically create smaller logical networks Even if you had 1000

users and could put them on the same logical network, you wouldn’t, because doing so would

result in poor performance and difficult management Instead, you would use switches to divide clients and peripherals into many separate broadcast domains

sub-Hybrid Physical Topologies

Except in very small networks, you will rarely encounter a network that follows a pure bus, ring,

or star topology Simple topologies are too restrictive, particularly if the LAN must date a large number of devices More likely, you will work with a complex combination of

accommo-these topologies, known as a hybrid topology Several kinds of hybrid topologies are explained

in the following sections

Star-Wired Ring

The star-wired ring topology uses the physical layout of a star in conjunction with the ring

topology’s data transmission method In Figure 6-4, which depicts this architecture, the solidlines represent a physical connection and the dotted lines represent the flow of data Data issent around the star in a circular pattern This hybrid topology benefits from the fault toler-ance of the star topology (data transmission does not depend on each workstation to act as a

FIGURE 6-3 A typical star topology network

NET+

1.1

repeater) and the reliability of token passing (discussed later in this chapter) Token Ring works, as specified in IEEE 802.5, use this hybrid topology.

inter-Chapter 6 251

HYBRID PHYSICAL TOPOLOGIES

FIGURE 6-4 A star-wired ring topology network

FIGURE 6-5 A star-wired bus topology network

Although even the smallest LAN technically has a backbone, on an enterprise-wide network,

backbones are more complex and more difficult to plan In networking, the term enterprise

refers to an entire organization, including its local and remote offices, a mixture of computersystems, and a number of departments Enterprise-wide computing must therefore take intoaccount the breadth and diversity of a large organization’s computer needs The backbone isthe most significant building block of enterprise-wide networks It may take one of several dif-ferent shapes, as described in the following sections

Serial Backbone

A serial backbone is the simplest kind of backbone It consists of two or more

internetwork-ing devices connected to each other by a sinternetwork-ingle cable in a daisy-chain fashion In networkinternetwork-ing,

a daisy chain is simply a linked series of devices Hubs and switches are often connected in a

daisy chain to extend a network For example, suppose you manage a small star-wired bus ogy network in which a single hub serves a workgroup of eight users When new employeesare added to that department and you need more network connections, you could connect asecond hub to the first hub in a daisy-chain fashion The new hub would offer open ports fornew users Because the star-wired hybrids provide for modular additions, daisy chaining is alogical solution for growth Also, because hubs can easily be connected through cables attached

topol-to their ports, a LAN’s infrastructure can be expanded with little additional cost

Hubs are not the only devices that can be connected in a serial backbone Gateways, routers,switches, and bridges can also form part of the backbone Figure 6-6 illustrates a serial back-bone network, in which the backbone is indicated by a dashed line

The extent to which you can connect hubs in a serial backbone is limited For example, in a10BASE-T network, you may use a maximum of four hubs to connect five network segments

in a serial fashion Using more hubs than the standard suggests (in other words, exceeding themaximum network length) will adversely affect the functionality of a LAN On a 100BASE-

TX network, you may use a maximum of two hubs connecting three network segments And

on most 1-Gbps networks, you can use only one hub to extend the network If you extend aLAN beyond its recommended size, intermittent and unpredictable data transmission errorswill result Similarly, if you daisy-chain a topology with limited bandwidth, you risk overload-ing the channel and generating still more data errors

Distributed Backbone

A distributed backbone consists of a number of connectivity devices connected to a series of

central connectivity devices, such as hubs, switches, or routers, in a hierarchy, as shown in ure 6-7 In Figure 6-7, the dashed lines represent the backbone This kind of topology allowsfor simple expansion and limited capital outlay for growth, because more layers of devices can

Fig-be added to existing layers For example, suppose that you are the network administrator for asmall publisher’s office You might begin your network with a distributed backbone consisting

Chapter 6 253

BACKBONE NETWORKS

FIGURE 6-6 A serial backbone

FIGURE 6-7 A simple distributed backbone

of two switches that supply connectivity to your 20 users, 10 on each switch When your pany hires more staff, you can connect another switch to one of the existing switches, and usethe new switch to connect the new staff to the network.

com-A more complicated distributed backbone connects multiple Lcom-ANs or Lcom-AN segments usingrouters, as shown in Figure 6-8 In this example, the routers form the highest layer of thebackbone to connect the LANs or LAN segments

FIGURE 6-8 A distributed backbone connecting multiple LANs

A distributed backbone also provides network administrators with the ability to segregate groups and therefore manage them more easily It adapts well to an enterprise-wide networkconfined to a single building, in which certain hubs or switches can be assigned according tothe floor or department Note that distributed backbones may include hubs linked in a daisy-chain fashion This arrangement requires the same length considerations that serial back-bones demand Another possible problem in this design relates to the potential singlepoints of failure, such as the devices at the uppermost layers Despite these potential draw-backs, implementing a distributed backbone network can be relatively simple, quick, andinexpensive

work-Collapsed Backbone

The collapsed backbone topology uses a router or switch as the single central connection point

for multiple subnetworks, as shown in Figure 6-9 Contrast Figure 6-9 with Figure 6-8, inwhich multiple LANs are connected via a distributed backbone In a collapsed backbone, asingle router or switch is the highest layer of the backbone The router or switch that makes

up the collapsed backbone must contain multiprocessors to handle the heavy traffic goingthrough it This is risky because a failure in the central router or switch can bring down theentire network In addition, because routers cannot move traffic as quickly as hubs, using arouter may slow data transmission.

Nevertheless, a collapsed backbone topology offers substantial advantages Most significantly,this arrangement allows you to interconnect different types of subnetworks You can also cen-trally manage maintenance and troubleshooting chores

Parallel Backbone

A parallel backbone is the most robust type of network backbone This variation of the

col-lapsed backbone arrangement consists of more than one connection from the central router orswitch to each network segment In a network with more than one router or switch, the par-allel backbone calls for duplicate connections between those connectivity devices as well Fig-ure 6-10 depicts a simple parallel backbone topology As you can see, each hub is connected tothe router or switch by two cables, and the two routers are also connected by two cables Themost significant advantage of using a parallel backbone is that its redundant (duplicate) linksensure network connectivity to any area of the enterprise Parallel backbones are more expen-sive than other enterprise-wide topologies because they require much more cabling than theothers However, they make up for the additional cost by offering increased performance andbetter fault tolerance

Chapter 6 255

BACKBONE NETWORKS

FIGURE 6-9 A collapsed backbone

As a network administrator, you might choose to implement parallel connections to only some

of the most critical devices on your network For example, if the first and second hubs in ure 6-10 connected your Facilities and Payroll Departments to the rest of the network, and yourorganization could never afford to lose connectivity with those departments, you might use aparallel structure for those links If the third and fourth hubs in Figure 6-10 connected yourorganization’s Recreation and Training Departments to the network, you might decide that par-allel connections were unnecessary for these departments By selectively implementing theparallel structure, you can lower connectivity costs and leave available additional ports on theconnectivity devices

Fig-Bear in mind that an enterprise-wide LAN or WAN may include different combinations ofsimple physical topologies and backbone designs Now that you understand fundamentalphysical topologies and backbone networks, you are ready to understand the related concept oflogical topologies

Logical Topologies

The term logical topology refers to the way in which data is transmitted between nodes, rather

than the physical layout of the paths that data takes A network’s logical topology will not essarily match its physical topology

nec-The most common logical topologies are bus and ring In a bus logical topology, signals travelfrom one network device to all other devices on the network (or network segment) They may

FIGURE 6-10 A parallel backbone

NET+

1.1