Topologies, Backbones,

Switching, and Ethernet

ITNW 1325, Chapter V, Part I

Physical Topologies

Physical TopologiesOverview: Reflect geometry of physical connections only –

without devices, connectivity methods, or addressing Don’t reflect device types, connectivity methods, or

addressing schemes in use Three fundamental types are bus, ring, and star – can

be mixed to create hybrid topologies Important to understand in order to troubleshoot related

problems or change communications infrastructure Differ from logical topologies – reflect how digital data

propagates between nodes

Physical TopologiesOverview (continued): Physical and logical topologies used within the same

network may be very different The term topology commonly refers to a physical

topology when used alone – with logical used explicitly Too restrictive – rarely seen in their pure form in

medium-sized and large networks

Physical TopologiesBus: Implies nodes connected by a single cable without

employing connectivity devices Provides only one communications channel – baseband

transmission is supported only Enables only one node to transmit at a time – nodes

compete for the right to transmit Requires each node to passively listen for and accept

data directed to it – passive topology Nodes other than sending and receiving ones sense the

transmission but ignore the information sent

Physical TopologiesBus (continued): A broadcast transmission would be processed by all

connected nodes – parts of a single broadcast domain Requires resistors – terminators – at the cable ends to

prevent endless travel of the signal (signal bounce) Without terminators, old signals would keep bouncing

off the wire ends – prevent propagation of new signals Must be grounded at one end – helps to remove static

electricity that could adversely affect the signal Example – nodes connected with a coaxial cable and

sharing the available bandwidth (50-Ohm terminators)

Physical TopologiesBus (continued): Not scalable – performance degrades as more nodes are

added and compete for the right to transmit Hard to troubleshoot – errors are easily detected but

their exact source or location are difficult to locate Not fault tolerant – any single break or defect affects

the entire network disrupting transmissions Lack security – every connected node can read any data

transmission destined to it or to someone else The least expensive topology to set up – rarely used

today due to multiple carried drawbacks

Physical TopologiesBus (continued):

Physical TopologiesBus (continued):

BNC T-Connector BNC Terminator

Physical TopologiesRing: Implies that each node is connected to the two nearest

ones – with the entire topology forming a circle Each node accepts and responds to frames addressed to

it – while forwarding other packets to the next node Implies that each node to participates in delivery acting

as a repeater – active topology Employs twisted pair of fiber optic cable as medium

Physical TopologiesRing (continued): Not scalable – performance degrades as more nodes are

added and introduce additional transmission delays Not fault tolerant – a single malfunctioning node would

break the ring and disable the entire network Used by obsolete Token Ring networks

Physical TopologiesRing (continued):

Physical TopologiesStar: Implies nodes connected through a central connectivity

device – forwards frames to the recipient’s segment Requires more cabling – twisted pair of fiber optic –

and more configuration than bus or star topologies Requires proper configuration and constant availability

of the central device Enables connecting two devices only to each physical

segment – a cabling problem affects two nodes at most Enables many nodes to transmit at a time – depending

on the ability of the central device to handle the load

Physical TopologiesStar (continued): The most scalable topology – can be easily easily

moved, isolated, or interconnected with other networks The most fault tolerant – a malfunctioning node would

not affect any other node or a communication device The easiest to troubleshoot – having one node per

segment makes an error easier to locate Carries single point of failure – a problem with the

central connectivity device affects all connected nodes More expensive to set up and maintain – requires more

cabling and administration than other topologies

Physical TopologiesStar (continued): Limits the number of nodes per segment – may result in

reduced or eliminated competition for the medium Most widely used topology on modern networks

Physical TopologiesStar (continued):

Logical Topologies

Logical TopologiesOverview: Reflect how information propagates between nodes –

may differ from a physical topology used Important to understand when building networks,

troubleshooting them, or optimizing their performance Represented by two fundamental types – bus and ring

Logical TopologiesBus (“Local Broadcast”): Data travels from one network device to all other ones

on the segment – each connected node can access data Commonly supported by networks that use a bus, a star,

or a star-wired bus physical topology

Ring: Data follows a circular path between sender and

receiver – even in case physical connections form a star Supported by networks that use a ring or a star-wired

ring physical topology

Logical TopologiesBus (continued):

Logical TopologiesRing (continued):

Hybrid Physical Topologies

Hybrid Physical TopologiesOverview: Complex combinations of fundamental physical

topologies – more suitable for modern networks Minimize weaknesses and increase scalability of

networks – better fit large and growing networks Two primary kinds – star-wired ring and star-wired bus

Hybrid Physical TopologiesStar-Wired Bus: Implies groups of nodes that are star-connected to

connectivity devices that are connected via a bus Enables covering longer distances and interconnecting

or isolating different network segments Inherits fault-tolerance, scalability, and manageability

from a star topology Requires more cabling and more connectivity devices

than a star or a bus – more expensive than basic ones A basis for modern midsize and large Ethernet

networks

Hybrid Physical TopologiesStar-Wired Bus (continued):

Hybrid Physical TopologiesStar-Wired Ring: Implies groups of nodes that are star-connected to

connectivity devices – and the ring logical topology Data flows in a circular pattern over the star-like wiring Inherits fault-tolerance, scalability, and manageability

from a star topology A basis for obsolete Token Ring networks

Hybrid Physical TopologiesStar-Wired Ring (continued):

Backbone Networks

Backbone NetworksOverview: Cabling that interconnects various parts of enterprise –

local and remote offices, departments, and computers Commonly carry substantially more traffic than cables

connecting to workstations – possess increased capacity Designed for continuous high throughput to avoid

congestion – complex and require careful planning Four fundamental types – serial, distributed, collapsed,

and parallel

Backbone NetworksSerial: Implies two or more internetworking devices connected

to each other in a daisy-chain fashion (linked series) Used for extending networks and adding device ports to

connect more user workstations Requires to observe the maximum number of connected

devices and segments – depends on the network type Not scalable – delays in information delivery increase

as more devices are added to the backbone Not fault tolerant – any single break or defect affects

the entire backbone disrupting transmissions

Backbone NetworksSerial (continued): The simplest logically, the least expensive, and the

easiest to implement backbone type

Backbone NetworksDistributed: Consists of a number of connectivity devices connected

to multiple central devices in a hierarchy More devices can be added to existing layers – allows

for simple expansion at lower costs of adding networks Can employ advanced devices for connecting LAN

segments – raise effectiveness of data transmissions Maps onto the structure of a building – with some

devices serving floors and/or departments and other ones connecting these segments together

Enables segregation and easy management of networks

Backbone NetworksDistributed (continued): May include a daisy-chain linked bus – inherits its

limitations requiring to place it thoughtfully Device at the upper layers represent potential single

points of failure – can damage the entire network Brings relatively simple, quick, and inexpensive

implementation – popular on today’s LANs and MANs

Backbone NetworksDistributed (continued):

Backbone NetworksCollapsed: Implies having the single central connection point for

multiple networks – connects multiple LANs together Makes the central device the highest level of the

backbone – must be able to handle heavy traffic loads Scalable – makes addition of new segments easy, with

potential necessity to upgrade the central device only The central network device represents single point of

failure for the entire network – must be available Fault tolerant – a failed segment does not affect others

Backbone NetworksCollapsed (continued): Centralizes maintenance and troubleshooting and

enables interconnecting networks of different types

Backbone NetworksCollapsed (continued):

Backbone NetworksParallel: Resembles other backbone types – implies duplicate

connections between connectivity devices Doubles the amount of cable needed and physical ports

used on network devices – can be quite expensive Provides network load balancing, redundancy, and

increased performance Most robust backbone type – commonly implemented

within critical segments of the network