Download Applying the OSI model

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project

Document related concepts

Computer security compromised by hardware failure wikipedia , lookup

Computer and network surveillance wikipedia , lookup

Transcript
COS226 notes
Networking standards organisations
ANSI (American National Standards Institute)



Represents the USA in setting international standards
Determines standards for the electronics industry and other fields
Requests manufacturers to comply with its standards, rather than dictating
EIA (Electronic Industries Alliance)



Composed of representatives from electronics manufacturing firms across USA
Sets standards for its members and helps write ANSI standards
Divided into subgroups, one of which is the TIA (Telecommunications Industry Association)
IEEE (Institute of Electrical and Electronic Engineers)




An international society composed of engineering professionals
Goals: Promote development & education in electrical engineering & computer science
Establish their own standards (for electronics etc) and contribute to the work of ANSI
Set standards for protocols used at the Physical layer (e.g. IEEE 802 standards)
ISO (International Organisation for Standardisation)




A collection of standards organisations representing 130 countries (HQ is in Geneva)
Developed the OSI model
Goal: Establish international technological standards to facilitate global information exchange
Authority extends to the fields of textiles, packaging, energy production, shipbuilding…
ITU (International Telecommunication Union)


A specialised United Nations agency that regulates international telecommunications, including radio &
TV frequencies, satellite & telephony specifications, networking infrastructure…
Provides developing countries with technical expertise and equipment
The OSI model
1 Physical layer
 Function: Manages hardware connections
 Contains the physical networking medium
 Protocols generate & detect voltage so as to transmit & receive signals carrying data
 Sets the data transmission rate & monitors data error rates
 Doesn’t provide error correction services
Devices / media: Cabling, connectors, repeaters, hubs, amplifiers, transceivers, and NICs
2 Data Link layer

Primary function: Divides data received from the Network layer into frames that can be transmitted by
the Physical layer
 Inserts a frame check sequence at the end of each frame
 Function: Codes, addresses, and transmits information
 Retransmits information if there’s no acknowledgement from the receiver
 Doesn’t try to figure out what went wrong in the transmission
 Controls the flow of information if multiple requests, that include many frames, are received
 Functions independently of the type of Physical layer used by the network
 Ethernet is an example of a Data Link layer technology
The IEEE separated the Data Link layer into two sub-layers:
1. LLC (Logical Link Control)
 Provides a common interface and supplies reliability and flow control services
2. MAC (Media Access Control)
 Appends the physical address of the destination computer onto the data frame
Connectivity devices: Bridges, switches, and NICs
1
3 Network layer

Primary function: Translates logical network addresses into physical ones, and decides how to route
data from sender to receiver
 Determines the best path between two points on different networks, by factoring in delivery priorities,
network congestion, quality of service, and cost of alternative routes
Connectivity devices: Routers
4 Transport layer






Primary function: Ensures that data are transferred reliably, in the correct sequence, and accurately
The most important layer in the OSI model
Sends an acknowledgment (ACK) to tell the sender that data were received correctly
Requests retransmission if the data contained errors
Retransmits data if there’s no acknowledgement within a given time period
Transport protocols handle flow control (= gauging the appropriate transmission rate based on how fast
the recipient can accept data)
 Long packets are broken into the maximum size that the type of network can handle
 Segmentation = decreasing the size of data units for a network that can handle only small units
 Reassembly = reconstructing the segmented data units
 Sequencing = assigning sequence numbers to each piece, so data can be reassembled correctly
Services: TCP, SPX
5 Session layer





Primary function: Establishes and maintains communication between two nodes on the network
Synchronises the dialogue between two nodes
Determines whether communications have been cut off, figuring out where to restart transmission
Decides which node will communicate first and how long a node can communicate
Monitors the identification of session participants, giving access to only authorised nodes
6 Presentation layer





Function: Handles text formatting and displays code conversion
Serves as a translator between the application and the network
Formats data in a schema that the network can understand
Manages data encryption & decryption
Protocols code & decode graphics and file format information
7 Application layer





Function: Transfers information from program to program
Provides interfaces to the software that enable programs to use network services
Some services provided: file transfer, file management, and message handling for email
APIs (Application Program Interface) = routines that allow programs to interact with the OS
MSMQ (Microsoft Message Queuing) = an API that stores messages sent between nodes in queues,
and then forwards them to their destinations
Applying the OSI model
Communication between two systems
How a message file is retrieved from the server:
 The APPLICATION layer formulates a request for data from a remote node, and transfers the request
to the Presentation layer
 The PRESENTATION layer determines whether the request should be formatted / encrypted, and adds
any necessary translation / codes required, before passing the request to the Session layer
 The SESSION layer picks up the formatted request and assigns a data token to it (to indicate that you
have the right to transmit data) before passing the data to the Transport layer
 The TRANSPORT layer breaks down the data & control info into chunks. If the data is too large for one
frame, it is subdivided into smaller blocks. One block at a time is passed to the Network layer
 The NETWORK layer adds addressing information to the data so that subsequent layers will know the
source & destination of the data. The data blocks are then passed to the Data Link layer
2

The DATA LINK layer packages the data blocks into individual frames and inserts a Frame Check
Sequence at the end of the frame, as well as a header that incorporates destination & source
addresses. The Data Link layer then passes the frames to the Physical layer
 The PHYSICAL layer just delivers the data to the cabling and sends it across the network
Once data arrives at the remote system, your request is unravelled, and the process is reversed.
Addressing through the layers
Addressing is a system for assigning unique ID numbers to each node on a network.
Data Link layer addresses:
 Fixed numbers are associated with the networking hardware, and are assigned at a factory
 Also called MAC (Media Access Control) addresses / physical addresses
 Guaranteed to be unique
 Block ID = the part of the MAC address that is unique to a particular vendor
 Device ID = the remaining 6 characters, based on the NIC’s model and manufacture date
 E.g. Block ID = 00608C, Device ID = 005499, so MAC address = 00608C005499
Network layer addresses:
 Follow a hierarchical addressing scheme in which subsets of data incrementally narrow down the
location of a node
 Also called logical addresses / virtual addresses
 Can be assigned through operating system software
 Useful to internetworking devices, like routers, because they make sorting data more logical
Network protocols
TCP/IP







Low cost
Ability to communicate between a multitude of dissimilar platforms
The standard on the Internet
The default protocol for NetWare 5.x and Windows 2000
Biggest advantage: Routable (i.e. carries network addressing information)
Flexible (can run on any combination of NOS or network media)
Disadvantage: Can require significant configuration (because of its flexibility)
Compared to the OSI model
OSI model
TCP/IP model
Application
Application
Presentation
Session
Transport
Transport
Network
Internet
Data Link
Network Interface
Physical
Protocols
FTP, TFTP, HTTP, SMTP, SNMP
DHCP, Telnet, POP, IMAP, NTP
TCP, UDP
IP, ICMP, ARP, RARP
Protocols:
IP (Internet Protocol)
 Provides information about how and where data should be delivered
 Unreliable, connectionless protocol (i.e. doesn’t guarantee delivery of data)
 Enables TCP/IP to internetwork (i.e. traverse more than one LAN segment or network)
 Subnets = individual networks (in an internetwork) that are joined together
 IP datagram = the IP portion of a data frame
 The IP datagram contains the necessary info for routers to transfer data between subnets


16
The length of the IP datagram cannot exceed 65 535 bytes (i.e. 2 )
The TTL (Time To Live) component of an IPv4 IP datagram header indicates the maximum time in
seconds that a datagram can remain on the network before it is discarded
3
TCP (Transmission Control Protocol)
 Provides reliable data delivery services
 Connection-oriented subprotocol (i.e. a connection must first be established between nodes)
 Checksum, flow control, and sequencing information compensate for IP’s reliability deficiencies
 TCP segment = the entity that becomes encapsulated by the IP datagram
UDP (User Datagram Protocol)
 Connectionless transport service
 Offers no assurance that packets will be received in the correct sequence
 Doesn’t guarantee that packets will be received at all
 Provides no error checking or sequence numbering
 Lack of sophistication makes it more efficient than TCP
 Useful when data must be transferred quickly, like live audio / video transmissions
ICMP (Internet Control Message Protocol)
 Notifies the sender when something goes wrong in transmission and packets aren’t delivered
 Simply reports which networks are unreachable and which packets have been discarded
 Doesn’t provide error control
 Used by diagnostic utilities like PING and TRACERT
ARP (Address Resolution Protocol)
 Obtains the MAC address of a node, then creates a local database that maps the MAC address to the
node’s IP address
 Works closely with IP
RARP (Reverse Address Resolution Protocol)
 A client sends a broadcast message with its MAC address and receives its IP address in reply
 The RARP server maintains a table of MAC addresses and their associated IP addresses
Telnet
 Terminal emulation protocol used to log on to remote hosts using TCP/IP
 Often used to connect two dissimilar systems
FTP (File Transfer Protocol)
 Used to send and receive files via TCP/IP
 The node running the FTP server portion accepts commands from one running the client portion
 Comes with a set of very simple commands that make up its user interface
TFTP (Trivial File Transfer Protocol)
 Like FTP, but it relies on UDP
 Connectionless and doesn’t guarantee reliable delivery of data
 A computer issues a read / write request to the remote node, which responds with an
acknowledgement, before the two computers begin transferring data
 Useful when you need to load programs on a diskless workstation
 Files that are transferred should have full read & write privileges for all remote users
SMTP (Simple Mail Transfer Protocol)
 Responsible for moving messages from one email server to another
 Uses a simple request-and-response mechanism to move messages
 Relies upon more sophisticated protocols, like POP to keep track of storing & forwarding messages
POP (Post Office Protocol)
 Provides centralised storage for email messages
 Relies on SMTP
 Doesn’t allow users to keep mail on the server after they retrieve it
IMAP (Internet Mail Access Protocol)
 Mail storage & manipulation protocol that also depends on SMTP’s transport system
 A more sophisticated alternative to POP
 Biggest advantage over POP: Users can store messages on the mail server, rather than always having
to download them to a local machine
4
Features:
 Users can retrieve all / a portion of a mail message
 Users can review messages & delete them while the messages stay on the server
 Users can create sophisticated methods of organising messages on the server
 Users can share a mailbox in a central location
 IMAP4 provides better security than POP because it supports authentication
Disadvantages:
 IMAP servers require more storage space & processing resources than POP servers
 If the IMAP server fails, users can’t access the mail left there
HTTP (Hypertext Transfer Protocol)
 The language that Web clients & servers use to communicate
 Forms the backbone of the Web
NTP (Network Time Protocol)
 Used to synchronise the clocks of computers on a network
 (If clocks aren’t synchronised, packets could expire prematurely, resulting in data loss)
 (Time is also used in routing to determine the most efficient path for data)
 Depends on UDP
 Time-sensitive, and can’t wait for the error checking that TCP would require
SNMP (Simple Network Management Protocol)
 Used to manage devices on a TCP/IP network
 Each device runs an agent that collects information about that device
 The collected information is transported to a central database
DHCP (Dynamic Host Configuration Protocol)
 Automatically assigns a unique IP address to devices on a network
Advantages:
 Reduces time & planning spent in IP address management
 Reduces potential for errors in assigning IP addresses
 Enables users to move their workstations & printers without changing their TCP/IP configuration
 Makes IP addressing transparent for mobile users
Addressing
First octet value:
0
1-126
127
128-191
192-223
224-239
240-254
255












Network class:
Additional info:
Class A
Share 1st octet
127.0.0.1
Share 1st two octets
Share 1st three octets
(Not for consumers)
(Not for consumers)
255.255.255.255
Class B
Class C
Class D
Class E
Purpose:
Broadcasts
Loopback information
Multicasting
Experimental use
Broadcasts
IP addresses are 32 bits in size, grouped into four 8-bit octets, separated by dots
Network portion: Indicates the network class
Host portion: Follows the network portion, and is unique to each device
Static IP address = one that is assigned manually
Dynamic IP address = one that is assigned automatically by DHCP
ICANN (Internet Corporation for Assigned Names and Numbers) are responsible for IP addressing and
domain name management
IANA (Internet Assigned Numbers Authority) are responsible for system administration
Individuals and businesses typically obtain IP addresses from their ISPs
If your network sits behind a firewall, you don’t need to adhere to ICANN standards
Each host requires a host name, e.g. student1
Each network requires a network name (domain name), e.g. sacc.tec.ca.us
Host name + domain name = FQDN (Fully Qualified Domain name) e.g. student1.sacc.tec.ca.us
5
IPX/SPX



Required for NetWare version 3.2 / lower
Can also be used with higher versions of NetWare, Windows 2000, and 98
Carries network addressing info, so it IS routable
Compared to the OSI model
OSI model
IPX/SPX model
Application
NCP, SAP, RIP
Presentation
Session
NetBIOS
Transport
SPX
Network
IPX
Data Link
Transmission media &
protocols like Ethernet
Physical
Core protocols
IPX (Internetwork Packet Exchange)
 Provides routing and internetwork services
 Uses datagrams to transport data
 Connectionless service (it doesn’t require a session to be established first)
 Doesn’t guarantee that data will be delivered in sequence or without errors
SPX (Sequenced Packet Exchange)
 Works with IPX to ensure that data are received whole, in sequence, and error free
 Connection-oriented protocol (it must verify that a session is established first)
 Can detect whether a packet was not received in its entirety
 If it discovers a packet was lost / corrupted, it resends the packet
SAP (Service Advertising Protocol)
 Devices advertise their availability through SAP
 Broadcasts occur every 60 seconds, by default
 Unnecessary traffic may be generated because of the broadcast mode used
 SAP servers maintain a database of device names correlated with their IPX addresses
 When a client needs a service, it queries the SAP database, which provides the IPX address
 Users don’t need to know the IPX addresses of other servers & workstations
NCP (Netware Core Protocol)
 Handles requests for services, like printing and file access, between clients & servers
 Acts as a translator between the workstation’s OS and the NetWare OS
 A client asks NCP to request a service  NCP notifies the server  NCP waits for acknowledgement
from the server  NCP allows the workstation to transmit data
 The request-and-response mechanism results in high reliability
 Extra traffic is generated that may lead to congestion
RIP (Routing Information Protocol)
 Factors in only the number of hops between nodes when determining a path between points
 Doesn’t consider things like network congestion or link speed
 Routers using RIP broadcast their routing tables every 30 seconds to other routers
 Broadcasting creates excessive network traffic
 Advantage of RIP: Stability
 RIP doesn’t work well in very large network environments with more than 15 routers
 RIP is slower and less secure than other routing protocols
Addressing
 IPX/SPX networks rely primarily on the MAC address for each workstation
 Network address (external network number) = the first part of an IPX address
 The network address is an 8-bit hex address, e.g. 000008A2
 Node address = the second part of an IPX address
 The node address is equal to the device’s MAC address, e.g. 0060973E97F3
 An example IPX address: 000008A2:0060973E97F3
6



Socket address = an address assigned to a process that needs to communicate on the network
Socket addresses (e.g. 456h) are appended to IPX addresses
Example of a complete IPX address for a socket: 000008A2:0060973E97F3:456h
NetBIOS & NetBEUI





NetBIOS doesn’t contain a Network layer so it can’t be routed
Used primarily with Windows-based systems
Not very secure
Not suitable for large networks
NetBEUI (‘Enhanced User Interface’) is an added application layer component on top of NetBIOS
Compared to the OSI model
OSI model
NetBIOS & NetBEUI
Application
Presentation
Session
NetBIOS
Transport
NetBEUI
Network
Data Link
Physical
Addressing
 NetBIOS (alphanumeric) names are assigned to each workstation
 The NetBIOS name can consist of up to 16 alphanumeric characters, e.g. MY_COMPUTER
 Once NetBIOS has found a node’s NetBIOS name, it will find its MAC address and use that
AppleTalk




An AppleTalk network is separated into logical groups of computers called AppleTalk zones
Each node can belong to only one zone
AppleTalk zones enable users to share file & printer resources on their Macintoshes
AppleTalk is unsuitable for large LANs or WANS
Compared to the OSI model
OSI model
AppleTalk
Application
AppleShare
Presentation
AFP
Session
ASP
Transport
ATP | NBP | RTMP | ZIP
Network
DDP
Data Link
Physical
Subprotocols
AppleShare
 Provides file sharing & print queuing services, password access to files or folders, and user accounting
information
AFP (AppleTalk Filing Protocol)
 Provides transparent access to files on both local and remote systems
ASP (AppleTalk Session Protocol)
 Establishes and maintains connections between nodes and servers
ATP (AppleTalk Transaction Protocol)
 Ensures reliable delivery of data
NBP (Name Binding Protocol)
 Translates human-readable node names into numeric AppleTalk addresses
7
RTMP (Routing Table Maintenance Protocol)
 Maintains a routing table of AppleTalk zones and their networks
 Uses ZIP to manage data in the routing table
ZIP (Zone Information Protocol)
 Updates zone information maps that tie zones to their networks for routing purposes
DDP (Datagram Delivery Protocol)
 Assigns an AppleTalk node’s address upon start-up
 Manages addressing for communications between AppleTalk nodes
Addressing
3 ways of identifying computers:
 Zone names can be plain words or numbers, and describe a department / group of users
 AppleTalk node IDs are 8 / 16-bit numbers that uniquely identify computers and that are randomly
chosen from a group of addresses
 AppleTalk network numbers are 16-bit numbers that identify the network to which a node is connected
so that nodes from different networks can communicate.
Transmission basics
Analogue signals







Voltage (= signal strength) varies continuously, appearing as a continuous wave
Advantage: can convey greater subtleties, because they are more variable than digital signals
Disadvantage: more susceptible to flaws (noise) than digital signals
Amplitude = a measure of a wave’s strength, or the height of the wave
Frequency = the number of times the amplitude cycles from and back to its starting point
Wavelength = the horizontal distance between extreme points on a wave’s cycle
Phase = the progress of a wave over time in relationship to a fixed point
Digital signals



Voltage turns on and off repeatedly
Advantage: more reliable than analogue signals, because noise affects transmission less
Disadvantage: many pulses are required to transmit the same amount of info as one wave can
Data modulation







A technology used to modify analogue signals in order to make them suitable for carrying data over a
communication path
A carrier wave is combined with a data wave, and is transmitted from one node to another
The purpose of the carrier wave is to help convey information - it does not represent information
The data wave is added to the carrier wave, modifying one property of the carrier wave (like frequency,
amplitude, or phase)
When the signal reaches the destination, the receiver separates the data from the carrier wave
Frequency modulation (FM) - the data must travel along a particular frequency
Amplitude modulation (AM) - the data must travel along a particular amplitude
Transmission direction
Simplex
 Signals may travel in only one direction
Half-Duplex
 Signals may travel in both directions, but in only one direction at a time
 One channel must be shared for multiple nodes to exchange information
Duplex
 Signals can travel in both directions over a medium simultaneously
 Multiple channels are used on the same medium
Multiplexing

A form of transmission that allows multiple signals to travel simultaneously over one medium
8
 Used to increase the amount of data that can be transmitted in a given time span
 To accommodate multiple signals, the single medium is logically separated into sub-channels
 A multiplexer (device that can combine many signals on a channel) is required at the sending end
 A demultiplexer (device that separates the combined signals) is required at the receiving end
Time division multiplexing (TDM)
 Divides a channel into multiple intervals of time
 Assigns a separate time slot to every node on the network
 If a node has no data to transmit, nothing will be sent during its time slot
 Disadvantage: Inefficient if some nodes on the network rarely send data
Statistical multiplexing
 Assigns slots to nodes according to priority and need
 More efficient than TDM because time slots are unlikely to remain empty
 Advantage: Allows networks to maximise available bandwidth
Wavelength division multiplexing (WDM)
 New technology used only with fibre-optic cable
 Each carrier signal is assigned a different wavelength, which is like its own separate sub-channel
 The wavelength of each carrier signal is then modulated with a data signal
 Multiplexed signals may be sent in both directions simultaneously
 At the transmitting end: A fibre-optic modem (FOM) creates a WDM wave
 At the receiving end: A FOM separates the multiplexed signals into individual ones
Relationships between nodes



Point-to-point transmission: One transmitter and one receiver
Broadcast transmission: One transmitter and multiple receivers
Webcasting: Broadcast transmission over the Web
Throughput and bandwidth




Throughput = the measure of how much data is transmitted during a given period of time
Throughput is expressed in bits per second
Bandwidth = a measure of the difference between the highest and lowest frequencies
Bandwidth is expressed in Hz / MHz
Baseband







A transmission form in which digital signals are sent through direct current pulses applied to wire
This direct current (DC) requires exclusive use of the wire’s capacity
As a result, baseband systems can transmit only one signal / one channel at a time
Every device on a baseband system shares the same channel
When one node is transmitting, all nodes must wait for transmission to end before they can send
Supports bidirectional signal flow: computers can both send & receive info on the same wire
Ethernet is an example of a baseband system
Broadband






Signals are modulated as radiofrequency (RF) analogue pulses that use different frequency ranges
The use of multiple frequencies enables access to several channels, so more data can be carried
Broadband transmission is used to bring cable TV to your home
Traditionally, signals travel in one direction, so you need a separate wire for transmission & receipt
More expensive than baseband transmission because of the extra hardware involved
Can span longer distances than baseband
Transmission flaws







Most noise is caused by electromagnetic interference (EMI) or radiofrequency interference (RFI)
EMI & RFI = waves that emanate from electrical devices / cables carrying electricity
Wireless transmission is typically more susceptible to noise than wireline transmission
Attenuation = the loss of a signal’s strength as it travels away from its source
To strengthen analogue signals: an amplifier increases the voltage of the signals
To strengthen digital signals: a repeater regenerates the signals
Amplifiers also amplify noise, whereas repeaters retransmit the signals in their original, pure form
9
Media characteristics
Cost
Factors that can influence the final cost of implementing a certain type of media:
 Cost of installation
 Cost of new infrastructure versus reusing existing infrastructure
 Cost of maintenance and support
 Cost of a lower transmission rate affecting productivity
 Cost of obsolescence
Noise immunity




You should install cabling well away from powerful electromagnetic forces
Thicker cables and cables coated with protective shielding are less susceptible to noise
You can use antinoise algorithms to protect data from being corrupted by noise
A metal conduit / pipeline can contain the cabling to protect it
Size and scalability
Maximum nodes per segment
 Depends on the attenuation: Each device added increases the signal’s attenuation slightly
Maximum segment length
 Limited because of attenuation: After a certain distance, a signal loses too much strength
Maximum network length
 Latency = the delay between transmission & receipt of a signal
 When you connect multiple segments, you increase the latency in the network
Things that can cause latency:
 The existence of intervening connectivity devices (like routers / hubs)
 A NIC in a PC (because data must travel through it)
 CAT5 wiring (The longer the wire, the greater the latency)
 (Note: A patch cable that is too short won’t cause latency because it is length that contributes)
Throughput



The most significant factor in choosing a transmission medium
Noise and devices connected to the transmission medium can limit throughput
(A noisy circuit spends more time compensating for the noise, and less time transmitting)
Connectors


Pieces of hardware that connect the wire to the network device
The type of connector affects the cost of installing & maintaining the network, ease of adding new
segments, and the technical expertise required to maintain the network
Twisted-pair cable
 Crosstalk occurs when signals travelling on nearby wire pairs infringe on another pair’s signal
 The twists in the wire help to reduce the effects of crosstalk
 Alien crosstalk occurs when signals from an adjacent cable interfere with transmission
 Alien crosstalk can occur when you bundle too many cables into small conduits
 STP can sustain the most bending without impairing transmission, compared to UTP, Thinnet, Thicknet
Standards that may be used on modern networks:
 CAT3 - contains 4 wire pairs and can carry up to 10 Mbps of data
 CAT5 - contains 4 wire pairs and supports up to 100 Mbps throughput (Most popular form of UTP!)
IEEE Physical layer networking standards:
 10BaseT - 10Mbps, CAT3+, UTP, star topology with central hub / repeater
 100BaseT - 100Mbps, CAT3+, UTP, star topology with central hub / repeater
* 100BaseTX - CAT5+, UTP, full duplexing
* 100Base T4 - uses all 4 pairs of UTP for unidirectional signalling, no full duplexing
 100BaseVG - efficient process to carry audio & video data, UTP, no full duplexing
Coaxial cable


Devices on a Thinnet network should be separated by at least 0.5m
Devices on a Thicknet network should be separated by at least 2.5m
10

Both Thinnet and Thicknet cable require 50-ohm resistors terminating either end of the network
Fibre-optic cable
 Fibre doesn’t conduct electricity like copper wire, so it doesn’t emit a current and is thus more secure
 Fibre can transmit data in only one direction at a time, so each cable must contain two strands
 Fibre’s high throughput makes it suitable for applications that generate a lot of traffic, like video & audio
 Multimode fibre uses multiple light paths and is typically used on LANs
 Multimode fibre is better suited for shorter distances than single-mode fibre
IEEE Physical layer networking standards:
 10BaseF - 10Mbps, 2 strands of multimode fibre, full-duplex, ST connectors, star topology
 100BaseFX - 100Mbps, 2+ strands of multimode, full-duplex, various connectors, star topology
Comparison of Networking Media:
Twisted pair
UTP
STP
1. Cost
2. Noise immunity:
Resistance to noise
3. Size & scalability
a) Max segment
length
b) Max nodes /
segment
c) Max network
length
Distance signals
can be carried
before amplification
is necessary
4. Throughput
5. Connectors
Cheapest
Least
resistant


100M
185M
500M
1024
30
100
Depends on type of
signalling used
550M
1500M
Standards
Topology
Fibre-optic
SingleMulti-mode
mode
Most expensive
Most resistant
100M
Depends on type of fibreoptic cable used
Shortest distance

Longest distance
Medium (100 Mbps)
Lowest (10 Mbps)
BNC-T &
Transceiver
BNC Barrel
(MAU) with
connectors
vampire taps.
AUI / n-series
connectors.
Highest (1000 Mbps)
SC connectors
ST connectors
RJ-45 connectors
Appearance
Coaxial
Thinnet
Thicknet
(RG-58A/U)
(RG-8)




Colour-coded pairs of
insulated copper wires,
twisted round each other
and all encased in a
plastic sheath
No
Surrounded
additional
by (foil)
shielding
shielding
CAT1 CAT7
10BaseT, 100BaseT,
100BaseVG
Star / star-hybrid
Black sheath
Yellow
sheath
Glass fibres at the centre,
surrounded by a layer of
glass (cladding)  a plastic
buffer  Kevlar strands 
a plastic sheath
Narrow
core
Wide core
10Base2
10Base5
10BaseF, 100BaseFX
Central copper core,
surrounded by an insulator,
then braiding, then a sheath
Bus
Star
Cable design and management

Cable plant = the hardware that makes up the enterprise-wide cabling system
Structured cabling
 TIA/EIA’s Commercial Building Wiring Standard
 Suggests how networking media can best be installed to maximise performance & minimise upkeep
 Specifies standards without regard for the type of media / transmission technology used
 Based on a hierarchical design that divides cabling into 6 subsystems:
1. Entrance facilities
 The point at which a building’s internal cabling plant begins
 Separates LANs from WANs
11
2. Backbone wiring
 Provides interconnection between ‘telco rooms’, equipment rooms, and entrance facilities
 Risers = vertical connectors between floors
 Usually fibre-optic / UTP cable
3. Equipment room
 Location where significant networking hardware, like servers and mainframe hosts, resides
 Cabling to equipment rooms usually connects telecommunications closets
4. Telecommunications closet
 Contains connectivity for workstations in its area, and cross connections to equipment rooms
 Punch-down block = a panel of data receptors into which cabling from the workstations is inserted
 Patch panel = a panel of receptors into which patch cables from the punch-down block are inserted
 Patch cables connect the patch panel to the hub / switch
 Good cooling and ventilation systems are important to maintain a constant temperature
5. Horizontal wiring
 Wiring that connects workstations to the closest telecommunications closet
 Three possible types for horizontal wiring: STP, UTP, or fibre-optic
 Maximum allowable distance = 100m
6. Work area
 An area encompassing all patch cables and horizontal wiring necessary to connect workstations,
printers, and other network devices from their NICs to the telco closet
 Patch cable = a short section of twisted-pair cabling with connectors on both ends that connects
network devices to data outlets
Atmospheric transmission media
Infrared transmission
 Infrared light signals transmit data through space
 Infrared pathways can carry data at rates that rival fibre-optic cable’s throughput
 Can function at 100 Mbps
 Can span distances up to 1000 m
Direct infrared transmission
 The transmitter and receiver must be within line of sight of each other
 Most often used for communications between devices in the same room
 Advantage: More secure than many other transmission methods, since signals are hard to intercept
 Disadvantage: The line of sight limitation prevents widespread use
Indirect infrared transmission
 Signals bounce off objects in the path
 Disadvantage: Not very secure because signals are not confined to a specific pathway
RF transmission










Relies on signals broadcast over specific frequencies, like radio / TV broadcasts
Advantage: RF can penetrate walls
Disadvantage: Easy interception of RF transmissions, so don’t use it if security is important
Disadvantage: Very susceptible to interference (RF signals can also interfere with each other)
Makers of RF components must obtain licenses for specific frequencies in different locations
The licensing procedure ensures that nearby systems won’t operate at the same frequencies
Narrowband = a RF technology which concentrates significant RF energy at a single frequency
Spread spectrum - uses a lower-level signal distributed over several frequencies simultaneously
Narrowband RF can be easily intercepted, but spread spectrum RF is quite secure
Both these types of RF offer a moderate throughput, ranging as high as 10 Mbps
Choosing the right transmission medium
Environmental factors to take into account:
Areas of high EMI or RFI
 Thick Ethernet and fibre-optic cable are the most noise-resistant media
Corners and small spaces
 STP and UTP are both very flexible
Distance
 Fibre-optic or wireless media are best for long stretches of transmission
12
Security
 Fibre-optic, direct infrared, and spread spectrum RF media are the most secure
Existing infrastructure
 The media you choose should be tailored to the previously installed equipment
Growth
 Find out how the organisation plans to expand and consider future applications, traffic, etc.
Simple physical topologies
Bus












A single cable connects all nodes, without intervening connectivity devices
Only one channel for communication, so every node share’s the bus’s total capacity
Most bus networks use coaxial cable
When a node wants to transmit data, it broadcasts an alert to the entire network
Each node passively listens for data directed to it - the destination node picks up the transmission
At the ends are 50-ohm resistors (terminators), which stop signals that reach the end of the wire
Signal bounce = when signals bounce between two ends of the network (if there are no terminators)
Advantage: Networks based on a bus topology are inexpensive to set up
Disadvantage: The more nodes, the more performance degrades and the slower the transmission
Disadvantage: Not practical for a network of more than 200 workstations
Disadvantage: Difficult to troubleshoot (intermediate errors are only picked up at the receiving point)
Disadvantage: Not very fault-tolerant (a break in the bus affects the entire network)
Ring
 Data are transmitted clockwise, in one direction, around the ring
 Each node accepts the packets addressed to it, and forwards the rest to the next node
 Most ring networks use twisted-pair or fibre-optic cabling
 No terminators, because there are no ends and data stops at the destination
 Disadvantage: A single malfunctioning workstation can disable the network
 Disadvantage: The more workstations, the slower the response time
Token passing:
1. A 3-byte packet (token) is transmitted from one node to another around the ring
2. A computer that wants to transmit picks up the token packet, adds control & data info & the destination
node’s address to transform the token into a data frame
3. The transformed token circulates around the network until reaching its destination
4. The destination node picks it up and returns an acknowledgment to the originating node
5. When the originating node gets the acknowledgement, it releases a new free token down the ring
 This ensures that only one workstation transmits data at any given time
 Active topology, because each workstation acts as a repeater for the transmission
 Advantage: High data reliability (no collisions), and efficient use of bandwidth
 Advantage: Doesn’t impose distance limitations on the LAN segment length (unlike CSMA/CD)
 Disadvantage: Generates extra network traffic
Star








Every node is connected through a central device, like a hub
Most star networks use twisted-pair or fibre-optic cabling
A cabling problem will affect two nodes at most
Devices transmit data to the hub, which retransmits the signal to the destination node
Advantage: More fault-tolerant than bus / ring networks
Advantage: Scalable (can easily be moved, isolated, or interconnected with other networks)
Disadvantage: A failure in the central connectivity device can take down a LAN segment
Disadvantage: More cabling and configuration is required than with bus / ring networks
Hybrid physical topologies
Star-wired ring


Physical layout of a star, with token-passing data transmission
Data are sent around the star in a circular pattern
13


Token Ring networks use this hybrid
Advantages: Fault tolerance of star topology and reliability of token passing
Star-wired bus




Groups of workstations are star-connected to hubs and networked via a single bus
You can cover longer distances and easily interconnect / isolate different network segments
Ethernet relies most often on this hybrid
Disadvantage: More expensive than using star or bus alone (more cabling & connectivity devices)
Daisy-chained




Hubs in star-wired bus / ring topologies can be daisy-chained
Little additional cost is required to expand a LAN this way
There is a limit as to how many hubs can be connected in sequence
If you daisy-chain a topology with limited bandwidth, you risk overloading the channel  errors
Hierarchical





You can separate hubs, switches, and routers for reasons of security, cost, scalability, reliability…
Layers separate devices based on their priority or function
Advantage: Ability to segregate (isolate) bandwidth among different groups
Advantage: Ease in adding / isolating different network groups
Advantage: Flexibility to interconnect different network types
Enterprise-wide topologies
Backbone networks

Backbones are capable of more throughput than the cabling that connects workstations to hubs
Serial backbone
 Simplest kind of backbone network
 Two or more hubs are connected to each other by a single cable
 Identical to daisy-chained networks of hybrid physical topologies
 Not suitable for large networks or long distances
 Rarely implemented for enterprise-wide networks
Distributed backbone
 A number of hubs are connected to a series of central hubs / routers in a hierarchy
 Allows for simple expansion and limited capital outlay for growth
 Network administrators can segregate workgroups and therefore manage them more easily
 You need to consider the maximum allowable distance between nodes and server
 Advantage: Implementation can be relatively simple, quick, and inexpensive
 Disadvantage: Central point of failure (The hub at the uppermost layer)
Collapsed Backbone
 A single router / switch is the central connection point (at the highest layer) for multiple subnetworks
 The router / switch must contain multiprocessors to handle the heavy traffic going through it
 Advantage: You can interconnect different types of subnetworks
 Advantage: You can centrally manage maintenance and troubleshooting chores
 Disadvantage: A failure in the central router / switch can bring down the whole network
 Disadvantage: Using routers can be slower, because they can’t move traffic as quickly as hubs
Parallel backbone
 The most robust enterprise-wide topology
 Like a collapsed backbone, but more than one connection to each network segment
 Advantage: Redundant links ensure connectivity to any area of the enterprise
 Disadvantage: More expensive, because more cabling is required
Mesh networks


Routers are interconnected with other routers, with at least two pathways connecting each router
Mesh networks typically contain several different backbone networks
14
WAN topologies
Peer-to-peer







Each site depends on every other site in the network to transmit and receive traffic
Each location is connected to another one through (usually) dedicated circuits
Suitable for only small WANs
Best for organisations with only a few sites and the capability to use dedicated circuits
Disadvantage: Doesn’t scale well, because all sites must participate in carrying traffic
Disadvantage: The addition of more sites can cause performance to suffer
Disadvantage: A single failure can take down communications between all sites
Ring






Each site is connected to two others, forming a ring pattern
Practical for connecting fewer than 4 or 5 locations
Advantage: A single cable problem won’t affect the entire network
Advantage: Routers at any site can redirect data to another route if one route is too busy
Disadvantage: Expanding ring-configured WANs can be difficult
Disadvantage: Expanding WANs is expensive because of the additional link
Star






A single site acts as the central connection point for several other points
This arrangement provides separate routes for data between any two sites
Advantage: More reliable that peer-to-peer or ring WANs
Advantage: Shorter data paths between any two sites
Advantage: Extending a star WAN is easy, and cheaper than with peer-to-peer or ring
Disadvantage: A failure at the central connection point can bring down the entire WAN
Mesh





Many locations are incorporated, so data can travel directly from origin to destination
If one connection has a problem, routers can redirect data easily and quickly
To reduce costs, you can implement a partial mesh, instead of a full mesh
Advantage: Very fault-tolerant, because of the multiple routes provided
Disadvantage: Expensive to connect every node to every other one
Tiered



WAN sites connected in star / ring formations are interconnected at different levels
Advantage: Easy expansion and inclusion of redundant links to support growth
Disadvantage: Flexibility means careful consideration of geography, usage patterns, and growth
Logical topologies
 The way in which data are transmitted, rather than the physical layout of the paths
 Most common logical topologies are bus (used by Ethernet) and ring (used by Token Ring)
Bus logical topology:
 Signals travel from one device to all other devices on a network
 Networks that use a bus / star / star-wired bus physical topology also use a bus logical topology
Ring logical topology:
 Signals follow a circular path between sender and receiver
 Networks that use a pure ring / star-ring hybrid physical topology also use a ring logical topology
Switching

Determines how connections are created between nodes
Circuit switching




A connection is established between two nodes before they begin transmitting data
Bandwidth is dedicated to this connection until the users terminate communication
E.g. Phone call
Good for live audio / videoconferencing, which can’t tolerate time delay of packet switching
15


Also used by: ISDN, T1 service, ATM
Disadvantage: Not economical, because bandwidth is monopolised
Message switching




A connection is established, information is transferred, and the connection is then broken
The information is stored and forwarded from one device to the next, until the destination
E.g. Email
Each device in the path must have enough memory and processing power to accept, store, and pass
on the information to the next node
Packet switching







Data is broken into packets before they are transported
Packets can travel any path on the network and needn’t arrive in the same sequence
The destination node reassembles the packets based on their control information
E.g. Ethernet, FDDI, the Internet
Fast and efficient mechanism for transporting data like word-processing files
Advantage: Doesn’t waste bandwidth by holding open a connection (like circuit switching)
Advantage: Doesn’t require devices in the path to process info (like message switching)
Ethernet


The most popular network technology used on modern LANs
Flexible (can run on a variety of media) and offers excellent throughput at a reasonable cost
Carrier Sense Multiple Access with Collision Detection (CSMA/CD)



1.
2.
3.
4.
5.
6.
7.
The network access method that all Ethernet networks have in common
“Carrier Sense”: Ethernet NICs listen and wait till sensing that no other nodes are transmitting
“Multiple Access”: Several Ethernet nodes can be connected and access the media simultaneously
When a node wants to transmit data it must first determine whether the channel is free
If the channel is not free, the node waits and checks after a random amount of time
When the channel is free, the node transmits its data
If two nodes see that the channel is free and begin to transmit, you get a collision
The network then carries out a collision detection routine
The NICs stop transmitting
Jamming follows: The NIC issues a 32-bit sequence that tells the rest of the nodes that its previous
transmission was faulty and that those frames shouldn’t be accepted as valid
8. The node waits, and retransmits its data when the line is available again
 Collision domain = an individual segment on an Ethernet network in which collisions can occur
 Repeaters repeat collisions, so connecting multiple parts of a network with repeaters results in a larger
collision domain
 Higher-layer connectivity devices, like switches and routers, can separate collision domains
 Data propagation delay = the time data take to travel from one point to another on a segment
 If the distance between two nodes is too far, CSMA/CD won’t work because collisions can’t be identified
accurately
Demand priority









The network access method used by 100BaseVG
Each device on a star / hierarchical network sends a request to transmit to the central hub
The hub grants the requests one at a time
Data travel from one device to the hub, and then to another device
Advantage: No collisions, because the hub determines which nodes transmit and when
Advantage: Secure, because data doesn’t pass by each node (only source & destination see it)
Advantage: The hub can prioritise transmission requests
Disadvantage: An intelligent hub is required (Some Ethernet networks don’t have one)
Disadvantage: The time a hub takes to process each request reduces overall performance
Switched Ethernet
Shared Ethernet:
 Fixed amount of bandwidth that must be shared by all devices on a segment
 Stations can’t send and receive data simultaneously
16
 Stations can’t transmit a signal when another one on the same segment is sending / receiving
Switched Ethernet:
 A switch separates a network segment into smaller ones, with each one supporting its own traffic
 Multiple nodes can simultaneously transmit & receive data over different logical network segments
 Increased bandwidth on a network segment, because of fewer workstations
Gigabit Ethernet



Can run over UTP cable, but performs better over multimode fibre
A fibre-based 1 Gigabit Ethernet network uses CSMA/CD transmission, the IEEE 802.3 frame type, and
is capable of full duplexing
Often encountered as part of a network’s backbone
Ethernet frame types






Frames are structured packages for moving data
Using frames reduces the possibility of errors / lost data, because of the built-in error checking
A node must be configured to expect one type of frame
If a node receives a different type of frame, it won’t be able to decode the data
Ethernet frame types have no relation to the topology / cabling characteristics of the network
Autosense = device driver software that senses what types of frames are running on a network
IEEE 802.3 (“Ethernet 802.2” / “LLC”)
 The default frame type for versions 4.x and higher of the Novell NetWare NOS
 The most popular Ethernet frame type for use with IPX/SPX traffic on most LANs
Preamble (8 bytes)
Destination address (6 bytes)
Source address (6 bytes)
Length (2 bytes)
LLC (DSAP, SSAP, Control field)
(46 to 1500 bytes)
Frame Check Sequence (4 bytes)
Signals that data are coming (Not included in frame’s total size)
MAC address identifies where data should be delivered
MAC address identifies where data originated
Length of the data field
Data plus padding. (Padding increases the size of the frame to
its minimum requirement of 46 bytes)
Ensures that data are received just as they were sent.
When a node transmits data, it performs a CRC (Cyclical
Redundancy Check): The values of the frame fields are used to
generate a unique number (the FCS), which the destination
node unscrambles and compares to the frame’s original form.
Logical Link Control (LLC) layer information:
Purpose: To distinguish among multiple clients on a network
Destination Service Access Point (1 byte) A Service Access Point (SAP) identifies a node that uses the
LLC protocol.
Source Service Access Point (1 byte)
Control field (1 byte)
Identifies the kind of LLC connection that must be established
(from unacknowledged to fully acknowledged)
Novell Proprietary 802.3 (“Ethernet 802.3”)
 The original NetWare frame type
 The default frame type for networks running NetWare versions lower than 3.12
 Supports only the IPX/SPX protocol
 Also called 802.3 Raw, because its data portion contains no control bits
 Rarely used on modern networks
Preamble (7 bytes)
SFD (1 byte)
Destination address (6 bytes)
Source address (6 bytes)
Length (2 bytes)
Data + padding (46 to 1500 bytes)
Frame Check Sequence (4 bytes)
Indicates the beginning of the addressing frame
Ethernet II
 The original Ethernet frame type developed by DEC, Intel, and Xerox
 The type field enables Ethernet II to support Novell IPX/SPX, TCP/IP, and AppleTalk protocols
17

The type field also compensates for the lack of LLC information
Preamble (8 bytes)
Destination address (6 bytes)
Source address (6 bytes)
Ethernet type (2 bytes)
Data + padding (46 o 1500 bytes)
Frame Check Sequence (4 bytes)
Identifies the upper-layer protocol contained in the frame
IEEE 802.3 SNAP
 SNAP = Sub-Network Access Protocol
 Compatible with IPX/SPX, TCP/IP, and AppleTalk protocols, but rarely used on modern LANs
Preamble (8 bytes)
Destination address (6 bytes)
Source address (6 bytes)
Length (2 bytes)
Data + padding: LLC (8 bytes),
(46 to 1500 bytes)
FCS (4 bytes)
Logical Link Control fields:
DSAP (1 byte)
SSAP (1 byte)
Control (1 byte)
OUI - Organisation ID (3 bytes)
Ethernet type (2 bytes)
Identifies the type of network on which the frame is running
(Like the Ethernet II frame)
Design considerations for Ethernet networks






Cabling - Ethernet networks can use coaxial / UTP cabling
Connectivity devices - Ethernet connectivity devices are cheaper than Token Ring / LocalTalk
Number of stations - Limit of 1024 on a 10BaseT / 100BaseTX Ethernet Network
Speed - Throughput of 10 Mbps, 100 Mbps, 1 Gbps and soon 10 Gbps
Scalability - You can easily expand Ethernet networks by adding connectivity devices on the bus
Topology - 10BaseT and 100BaseTX Ethernet networks use a star-wired bus hybrid topology
LocalTalk








A network access method for networking Macintosh computers
Only capable of 230 Kbps maximum throughput
Not easily supported by non-Macintosh devices
Requires twisted-pair wiring and uses the same type of cabling used for phone connections
Relies on a star or, more often, a bus physical topology
Maximum segment length = 305 metres
Up to 32 nodes may be connected to any single LocalTalk network before errors begin to occur
LocalTalk relies on the AppleTalk protocol, but may also support MacTCP (the Mac version of TCP/IP)
Carrier sense Multiple Access with Collision Avoidance (CSMA/CA)

A node signals its intent to transmit before it actually does so, so collisions are avoided
Token Ring








A network technology first developed by IBM
Biggest disadvantage: more expensive to implement than Ethernet networks
Reliability results in less downtime and lower network management costs than Ethernet
High-Speed Token Ring (HSTR) can use either twisted-pair / fibre-optic cable
Uses the token-passing routine and a star-ring hybrid physical topology
Active monitor = a workstation on the network, acting as the controller for token passing
Connectors that may be used on Token Ring cables: RJ-45, DB-9, or type 1 IBM
A media filter can enable different connectors and receptors to fit together
18
Multistation Access Unit (MAU)
 Token Ring’s equivalent of a hub
 In the star-ring hybrid topology, the MAU completes the ring internally with Ring In & Out ports
 You can easily expand a Token Ring network by connecting MAUs through by their In & Out ports
 Unused ports on a MAU have self-shorting data connectors that internally close the loop
 Advantage: The self-shorting feature of the MAU ports makes Token Ring highly fault-tolerant
Controlled Access Unit (CAU)
 A connectivity device similar to a MAU, but more flexible and manages nodes more easily
 Contain interchangeable modules that you can plug into the Ring In & Out connections
 With interchangeable modules, you can change easily from e.g. STP to fibre-optic cable
 You can connect to a CAU and determine what type of traffic is passing through a device
 More expensive than MAUs
 Plug in a Lobe Attachment Module (LAM) to a CAU if you want to connect more nodes
 LAMs allow up to 20 devices to plug into each CAU receptacle
Token ring switching


Switching helps networks better utilise limited bandwidth
More expensive and more difficult to manage than Ethernet switches
Token ring frames


Two types of frames: IEEE 802.5 and IBM Token Ring frame
Larger than Ethernet frames (Larger frame sizes result in more efficient data transmission)
Start delimiter (1 byte)
Access control (1 byte)
Frame control (1 byte)
Destination address (6 bytes)
Source address (6 bytes)
Data (0 - 16 000 bytes)
Frame Check Sequence (4 bytes)
End delimiter (1 byte)
Frame status (1 byte)
(Red = fields that make up the token)



Signifies the beginning of the packet
Contains info about the priority of the frame
Defines the type of frame; used in the FCS
Destination node’s MAC address
Originating node’s MAC address
Data, possibly also routing & management info
Results of CRC algorithm - used to check the integrity of the frame
Indicates the end of the frame
Provides acknowledgement that the frame was received whole
Note: Each frame type is unique and won’t interact with different frame types on a network
(I.E. Ethernet frames can’t interpret Token Ring frames and vice versa.)
You can however work with multiple protocols on a network while using only one frame type
Design considerations for Token ring networks






Cabling - STP or UTP
Connectivity devices - More expensive than Ethernet equipment
Number of stations - Limited, depending on cabling
Speed - Token ring networks can run at either 4, 16, or 100 Mbps
Scalability - You can easily daisy-chain Token ring MAUs to expand the network
Topology - Star-wired ring topology, which is highly fault-tolerant
Fibre Distributed Data Interface (FDDI)










A network technology that uses a double ring of fibre to transmit data at 100 Mbps
Developed in response to the throughput limitations and Ethernet & Token ring at the time
No longer popular, because Ethernet & Token ring now have faster transmission speeds
Based on ring physical & logical topologies, similar to a Token ring network
Also relies on the same token-passing routine
Unlike Token ring, FDDI runs on two complete rings
The secondary ring is used for transmission if the primary ring experiences Physical layer problems
Advantage: Reliable and secure, because of redundancy and fibre-optic cable
Advantage: Works well with Ethernet 100BaseTX technology
Disadvantage: High cost relative to Fast Ethernet
19
Asynchronous Transfer Node (ATM)

















A networking standard that describes both a network access method and a multiplexing technique
Relies on a fixed packet size to achieve data transfer rates up to 9953 Mbps
A cell (the fixed packet) contains 48 bytes of data and a 5-byte header
These small packet sizes need more overhead, but the cells’ efficiency makes up for that loss
Typically used on WANs
Very expensive
Very well suited to videoconferencing
A packet switching technology, because data is packaged into cells before transmission
(Circuit switching too, because of virtual circuits)
Relies on virtual circuits (connections between nodes that logically appear as direct links)
Advantage of virtual circuits: Efficient use of limited bandwidth
The reliable connection provided by virtual circuits allows ATM to guarantee a specific QoS
Quality of Service (QoS) = a standard that specifies that data will be delivered within a certain time
ATM cells can support multiple types of higher-layer protocols (TCP/IP, IPX/SPX, AppleTalk…)
LANE (LAN Emulation) enables ATM networks to be integrated with Ethernet / Token ring networks
LANE encapsulates incoming Ethernet / Token ring frames and converts them into ATM cells
Gigabit Ethernet (which is faster and cheaper) poses a threat to ATM
Network adapters
Types of network adapters
1. Expansion slot network adapters
 Connect to a PC’s bus
 Expansion board = a circuit board used to connect a device to the system board
 Expansion slots = openings into which the expansion board can be inserted
 Bus = the circuit used by the system board to transmit data to the computer’s components
 The type of expansion board (your NIC) must match the computer’s bus
Some PC bus types:
 Industry Standard Architecture (ISA)
 8 / 16-bit bus
 Can’t support 100-Mbps throughput
 May connect serial devices, like mice / modems
 MicroChannel Architecture (MCA)
 32-bit bus
 IBM’s bus for personal computers
 Extended Industry Standard Architecture (EISA)
 32-bit bus
 Compatible with older ISA devices because of the same length and pin configuration
 Peripheral Component Interconnect (PCI)
 32 / 64-bit bus
 Used for nearly all of today’s new PCs
 Shorter connector length, but faster data transmission
For externally attached network adapters:
2. USB network adapters
 Connect to a Universal Serial Bus port
 USB = a standard external bus that can be used to connect multiple types of peripherals
 Advantage of using a USB network adapter: simple installation
 Disadvantage: Throughput limitation of 12 Mbps
3. PC cards (‘credit card adapters’)
 PCMCIA (Personal Computer Memory Card International Association) interfaces provide a standard
interface for connecting any type of device to a portable computer
 PCMCIA devices are commonly known as PC cards
4. Parallel port network adapters
 Connect to the parallel port of a computer on one side, and to the network cable on the other side
 Designed primarily for use on laptops, but rarely used now
20
5. Wireless network adapters
 Use an antenna to exchange signals with a base station transceiver or another wireless NIC
 Expansion slot network adapters, PC card network adapters (especially!), and USB network adapters
can all be wireless
 Advantage: Well suited to environments where cabling can’t be installed
 Disadvantage: More expensive than wire-dependent network adapters
Installing network adapters
 To install a network adapter, first install the hardware, and then the software
 Sometimes a third step is necessary: Configuring the firmware (= combo of hard + soft-ware)
 The hardware component of firmware is an EEPROM chip (because data may be changed)
 Firmware contains info about the NIC’s transmission speed, preferred IRQ, I/O port address, etc.
Installing and configuring network adapter hardware
 On older expansion board NICs, you can use a jumper instead of modifying firmware settings
 Jumper = a small, removable piece of plastic that contains a metal receptacle
 The metal receptacle fits over a pair of pins on a circuit board to form a bridge that completes a circuit
between those two pins
 By moving the jumper you can modify the board’s circuit, thereby giving it different instructions
 DIP (Dual Inline Package) switch = a small, plastic toggle switch that can represent ‘on’ and ‘off’
 As with jumpers, they can be used to modify system settings and are rarely used on modern NICs
Installing and configuring network adapter software
 You must ensure that the correct device driver is installed for the network adapter
 Device driver = software that enables an attached device to communicate with the computer’s OS
 Some operating systems come with many built-in device drivers; others don’t
Variable settings:
IRQ (Interrupt Request)
 A message to the computer instructing it to stop what it’s doing and pay attention to something else
 When a device attached to the bus (e.g. keyboard) requires attention, it uses an interrupt request
 Interrupt = the wire on which a device issues voltage to signal this request
 IRQ number = a number that uniquely identifies a component to the main bus
 IRQ numbers range from 0 to 15, and many devices reserve the same number by default
 (Numbers 9 - 11 have no default assignment)
 You can view IRQ settings in the computer’s CMOS utility
 If two devices try to use the same IRQ, any of the following problems may occur:
 The computer could ‘hang’
 The computer could run much more slowly than usual
 Other devices (like serial / parallel ports) may stop working
 Video or sound card problems may occur
 The computer could fail to connect to the network
 The computer could experience intermittent data errors during transmission
Memory range
 Indicates the area of memory that the network adapter and CPU will use for exchanging data
 Some memory ranges are reserved for specific devices (e.g. the system board)
 Network adapters typically use a memory range in the high memory area (A0000 - FFFFF)
Base I/O port
 The base I/O port setting specifies which area of memory will act as a channel for moving data between
the network adapter and the CPU
Firmware settings
 When you change firmware, you are writing to the EEPROM chip on the network adapter
 To change a network adapter’s firmware, you need a bootable floppy disk containing the configuration
or DOS install utility that shipped with the network adapter
 To run the utility, you must start the computer with this floppy disk inserted
 The configuration utility allows you to view the IRQ, I/O port, base memory, and node address
 The utility also allows you to perform diagnostics (component and connectivity tests)
 If a network adapter fails its connectivity tests, it’s probably configured incorrectly
 If a network adapter fails a physical component test, it may need to be replaced
Choosing the right network adapter

The most critical factor is compatibility with your existing system
21

If the network adapter will be installed in a server, then performance is very important
Comparison of connectivity devices:
Time to transmit data:
Fastest  repeaters & hubs  bridges  routers  gateways  Slowest
Devices
Purpose
Repeaters
Regenerate a
digital
signal
Description
Hubs
Connect
switches,
servers, and
PCs etc. to
the network
A multi-port
repeater
Features
* Only
suited to
bus
topology
networks
* Support a
variety of
media and
transmission
speeds
* MIB stores
info
generated by
intelligent
hubs.
OSI layer
Ports
Physical
One input
One output
No
Physical
Multiple
Intelligence
Advantages
Disadvantages
* Allow you
to extend a
network
inexpensively
* Can’t
improve or
Passive /
intelligent
* Intelligent
hubs can
analyse data
Bridges
Analyse
incoming frames
and decide how
to direct them to
their destination
(Forward
packets based
on MAC
addresses)
Look like
repeaters
Switches
Subdivide a
network into
smaller logical
pieces
Routers
Integrate LANs
and WANs
running at
different
transmission
speeds and
using a variety
of protocols
Gateway
Connect
two
dissimilar
kinds of
networks
A multi-port
bridge
Combination of
hardware
& software
* Filtering
database stores
MAC addresses
* Don’t
distinguish
between
different
protocols
* Bridging
methods:
- Transparent
- Source-route
- Translational
Data link
One input
One output
Yes
* Each device
connected to a
switch
receives its
own dedicated
channel
* Often used
to ease traffic
congestion
Internal
processor,
memory &
power supply,
input & output
jacks,
management
console
interface
* Protocoldependent
* Can
determine
shortest,
fastest path
* Flexible
Data link
Multiple
Network
Multiple
Multiple
Yes
Yes
* Flexible: can
forward frames
from different
protocols with
equal speed and
accuracy
* Enhance
network
performance by
filtering traffic
* Can detect
and discard
flawed packets
* Extend the
max distance of
a network
* Limit the
number of
devices in a
collision
domain  ltd
potential for
collisions
* Secure
* Provide
separate
channels for
every device
* Intelligent
hubs report
* Can become
overwhelmed
22
* Not simple to
install
* Repackage
info so
that it
can be
read by
another
system
* More
ex-
correct a
bad signal
every time a
port detects a
lost
connection.
* Intelligent
hubs are
more
expensive
than passive
ones.
by continuous
heavy traffic
pensive
than
routers
Hubs
Elements shared by most hubs
 Ports = receptacles where patch cables connect devices to the hub
 Uplink port = the receptacle to connect one hub to another hub in a daisy-chain / hierarchy
 Port for management console = a receptacle used to connect some type of display that enables you to
view the hub’s management information
 Backbone port = the receptacle used to connect a hub to the network’s backbone
 Link LED = the light on a port that indicates whether it is in use
 Traffic LED = the light on a port that indicates that traffic is passing through the port
 Collision LED = the light (on Ethernet hubs only) that roughly indicates collisions by blinking
 Power supply = the device that provides power to the hub
 Ventilation fan = a device used to cool a device’s internal electronics

Intelligent hubs (‘managed hubs’) can provide network management information
Hub types
Description
Standalone
Serve a group of computers
that are isolated from the
rest of the network.
Connections
Can connect to another hub
by coax, fibre-optic, or
twisted-pair cable.
Typically not connected in a
hierarchy / daisy-chain.
Small, independent
departments, home offices,
or test lab environments.
Passive / intelligent.
Usually 4, 8, 12, or 24 ports
Hublet - 4 ports
Up to 200 connection ports
Environment
Intelligence
Ports
Modular
Provide a number of
interface options within
one chassis.
Adapters are inserted into
a system board.
Overkill for small networks.
Nearly always intelligent.
Usually 6, 12, or 24 ports
The network doesn’t
depend on a single hub (so
no single point of failure).
Advantages
Disadvantages
Stackable
Resemble standalone hubs,
but are physically designed
to be linked with other hubs
in a single telco closet.
Support a number of
different media connectors
and transmission speeds
A single hub for many
connections introduces a
single point of failure.
Choosing the right hub
Factors to consider:
23
More flexible than
standalone / stackable
hubs.
Highest reliability, because
you can attach redundant
components.
Allow for future growth by
providing expansion slots
for additional devices.
Can accommodate many
types of devices.
Most expensive type of
hub.






Performance - If it is a concern, rather use switches (which can sub-divide a LAN segment)
Cost - If your budget is tight and you don’t need much flexibility, use a passive standalone hub
Size & growth - You need to determine how many devices will connect to each hub
Security - If your network carries very sensitive data, rather use switches, routers, or firewalls
Management benefits - An intelligent hub can provide management info
Reliability - If your network can’t tolerate any downtime, choose a modular hub with redundant power
supplies and redundant connections to the backbone
Bridges
Methods for forwarding / filtering packets:
Transparent bridging
 Used on many Ethernet networks
 A bridge polls a network to learn about its physical topology as soon as it is installed
 When a bridge receives a packet from an unknown source, it adds the location of that source to its
filtering database
 Disadvantage: On LANs with many bridges, each one may recognise a different path to one node
 When this happens, data could get bounced among the bridges and never reach the destination
 Spanning tree algorithm = a routine that can detect circular traffic patterns and modify the way multiple
bridges work together to avoid such patterns
Source-route bridging
 Used on most Token Ring networks
 A bridge polls the network to determine what path is the best to get from point A to B
 The bridge then adds this info to the data packet
 The forwarding info is part of the data, so you don’t get circular traffic problems
 Very well-suited to WANs, where multiple bridges and long routes are common
Translational bridging
 Connects networks that use different logical topologies
 The bridge forwards packets, and also translates packets between different logical topologies
Switches
Two methods of switching used by LAN switches:
1. Cut-through mode







The switch reads a frame’s header and decides where to forward the data before receiving the entire
packet
The switch can’t verify data integrity from the FCS because it isn’t read before transmitting
Switches can detect runts (packet fragments with a certain type of data flaw), and will wait to transmit
that packet until it determines its integrity
Switches can’t detect corrupt packets
Biggest advantage: Speed (because the switch doesn’t stop to read the entire data packet)
Time-saving advantages become insignificant if the switch is flooded with traffic
Cut-through switches are best suited to small workgroups where speed is important and the low
number of devices minimises the potential for errors
2. Store and forward mode




A switch reads the entire data frame into its memory and checks it for accuracy before transmitting
More time-consuming than cut-through mode, but data is transmitted more accurately
More appropriate for larger LAN environments because data errors are not propagated
Switches can also transfer data between segments running different transmission speeds
Using switches to create VLANs (Virtual LANs)





VLAN = a logically separate network within a network, created by grouping a number of ports into a
broadcast domain
Broadcast domain = a combination of ports that make up a Layer 2 segment and must be connected by
a Layer 3 device (like a router or a Layer 3 switch)
The ports don’t have to reside on the same switch or even on the same network segment
A VLAN can include servers, workstations, printers, routers… (devices you can connect to a switch)
Advantage of VLANs: Ability to link distant users and create small workgroups from large LANS
24
Higher layer switches







Layer 3 switch = one capable of interpreting Layer 3 data
Layer 4 switch = one capable of interpreting Layer 4 data
These higher-layer switches are also called routing switches / application switches
These switches can perform advanced filtering, statistics keeping, and security functions
They can also transmit data more rapidly than a router and are easier to install
They can’t translate between Token Ring and Ethernet networks, encapsulate protocols, or prioritise
traffic, so you still do need routers
Higher-layer switches can cost three times more than Layer 2 switches
Routers


Modular router = one with multiple slots that can hold different interface cards or other devices
Routers can’t forward packets as quickly as bridges can because they operate at layer 3 and therefore
take time to interpret logical addressing information
Basic functions:
 Connect dissimilar networks
 Interpret Layer 3 information
 Determine the best path for data to follow from point A to point B
 Reroute traffic if a primary path is down but another path is available
Optional functions:
 Filter out broadcast transmissions to alleviate network congestion
 Prevent certain types of traffic from getting to a network
 Support simultaneous local and remote connectivity
 Provide high network fault tolerance through redundant components
 Monitor network traffic and report statistics to a MIB
 Diagnose internal or other connectivity problems and trigger alarms
Static routing
 A network administrator programs a router to use specific paths between nodes
 Not optimal, because it doesn’t account for congestion, failed connections, or device moves
Dynamic routing
 Automatically calculates the best path between two nodes and stores this info in a routing table
 If failures affect the network, the router can reroute data through a different path
Brouter (bridge router)
 A combined router and bridge
 Advantage: You can forward non-routable protocols, like NetBEUI, and connect multiple network types
through one device
 Offers support at both Layers 2 & 3 of the OSI model
 Intelligently handles any packets that contain Layer 3 addressing info, and simply forwards the rest
Routing switch
 A combined router and switch
 Can interpret data from Layers 2 & 3 of the OSI model
 Not as fully featured as a true router, so routing switches haven’t gained wide acceptance
Routing protocols
 Used to determine the best path between nodes
 Used to collect data about current network status and contribute to selecting the best path
 Convergence time = the time it takes for a router to recognise a best path in the event of a change
 Bandwidth overhead = the burden placed on the network to support the routing protocol
RIP (Routing Information Protocol)
 For IP and IPX (see notes above in the IPX/SPX section)
OSPF (Open Shortest Path First)
 For IP
 Makes up for some of RIP’s limitations, and can co-exist with RIP on a network
 Uses a more complex algorithm for determining best paths
 Under optimal network conditions, the best path is the most direct path
 If there is too much traffic in the way, the router can determine the most efficient path
 Each router maintains a database of the other routers’ links, and if it is notified of a failure, it can rapidly
compute an alternate path
25
 Advantage: Network bandwidth is kept to the minimum
 Disadvantage: This approach requires more memory and CPU power on the routers
EIGRP (Enhanced Interior Gateway Routing Protocol)
 For IP, IPX, and AppleTalk
 Fast convergence time and low network overhead
 Easier to configure & less CPU intensive than OSPF
 Supports multiple protocols and limits unnecessary network traffic between routers
 Accommodates very large & heterogeneous networks, but is only supported by Cisco routers
BGP (Border Gateway Protocol)
 For IP
 The routing protocol of Internet backbones
 The most complex of the routing protocols
Gateways
Popular types of gateways
E-mail gateway
 Translates messages from one type of email system to another
IBM host gateway
 Establishes and manages communication between a PC and an IBM mainframe
Internet gateway
 Allows and manages access between LANs and the Internet
 Can restrict the kind of access LAN users have to the Internet, and vice versa
LAN gateway
 Allows segments of a LAN running different protocols / network models to communicate
 A router, single port on a router, or a server may act as a LAN gateway
WANs




Send data over public communications links
Mostly take the form of mesh or partial-mesh networks
WAN link = a connection between two WAN sites
WAN links typically connect one site to only one other site (point-to-point)
PSTN (Public Switched Telephone Network)
 The network of typical phone lines
 Originally composed of analogue lines and developed to handle voice-based traffic
 Now uses digital transmission through fibre-optic & twisted-pair cable, microwave and satellite
 A modem must still convert a computer’s digital pulses into analogue signals for PSTN
 Currently used for most dial-up connections to LANs
 Uses circuit switching
 Offers only marginal security because it is a public network, so you must encrypt data for security
 Advantages: Ubiquity, ease of use, low cost
 Disadvantage: Inability to ensure quality (The digital lines are more reliable than the analogue ones)
 Disadvantage: Limited throughput (PSTN modems have a connection speed of 56 Kbps)
POP (Point Of Presence)
 POP = the place where two phone systems meet (a long-distance carrier with a local phone company,
or a local carrier with an ISP’s data centre)
1. When you dial into your ISP, your signal travels to the local phone company’s POP
2. At the POP, your signal is converted back to digital pulses and transmitted to your ISP’s POP through a
digital backbone
3. The ISP’s POP connects to its ISP through a digital link
4. Your request for info enters the Internet and the process is reversed to bring you a Web page

Each time your transmission travels through a POP, or is converted, it loses a little throughput
X.25 and Frame Relay
26
X.25





An analogue, packet-switched technology designed for long-distance data transmission
Maximum throughput of 2.048 Mbps
Originally developed as a more reliable alternative to the voice phone system for connecting mainframe
computers and remote terminals
Ensures data reliability over long distances by verifying the transmission at every node
Slow and unsuitable for time-sensitive applications (audio/video) because of this verification
Frame relay





An updated, digital version of X.25 that also relies on packet switching
Data is separated into frames, which are then relayed from one node to another without any verification
or processing
Supports higher bandwidth than X.25
Maximum of 1.544 or 45 Mbps throughput
Something in common with ATM connections: both rely on virtual circuits










X.25 & frame relay are depicted as clouds, because of the indeterminate nature of their traffic patterns
X.25 and frame relay may be configured as switched / permanent virtual circuits
SVC = a connection established when parties need to transmit and dismantled when complete
PVC = a connection established before transmission and maintained afterwards
CIR (Committed Information Rate) = the amount of bandwidth guaranteed by the service provider
Advantage: Leasing a frame relay circuit means you pay for only the amount of bandwidth required
Advantage: Frame relay is much cheaper than the newer WAN technologies, like ATM
Advantage: Frame relay follows an established worldwide standard
Disadvantage: Variable traffic patterns, because X.25 and frame relay use shared lines
Disadvantage: Frame relay circuits are not as private as dedicated circuits
ISDN (Integrated Services Digital Network)
 An international standard for transmitting data over digital lines
 Uses the phone carrier’s lines and either a dial-up or dedicated connection
 Can carry up to two voice calls and one data connection simultaneously
 B (bearer) channel: employs circuit-switching techniques to carry voice, video, audio, etc
 D (data) channel: employs packet switching to carry information about the call (like session info)
 Each ISDN connection uses only one D channel and the number of B channels varies
 Disadvantage: ISDN can span up to only 18 000 feet before a repeater is needed
BRI (Basic Rate Interface)
 The most economical type of ISDN connection
 Uses 2 B channels and 1 D channel (2B + D)
 The two B channels are treated as separate connections by the network
 The B channels can carry voice and data simultaneously and separate from each other
 Bonding = when the two 64 Kbps B channels are combined to achieve 128 Kbps throughput
 A NT1 (Network Termination 1) device at the customer’s site connects the twisted-pair wiring with the
ISDN terminal equipment
 TE (Terminal Equipment) may include cards / devices used to connect computers to the ISDN line
 A TA (Terminal Adapter) converts digital signals into analogue ones for the analogue equipment
PRI (Primary Rate Interface)
 Used by businesses that need more throughput
 Uses 23 B channels and 1 D channel (23B + D)
 Uses the same kind of equipment as BRI links, but needs an extra termination device: NT2
 Network Termination 2 is necessary for handling the multiple ISDN lines
T-carriers


Transmission uses time division multiplexing over two wire pairs (one for transmitting and one for
receiving) to divide a single channel into multiple channels
Each channel may contain data, voice, or video signals
27

Common WAN links for use in large businesses
Types of T-carriers








A T1 circuit can carry 24 voice / data channels, with max throughput of 1.544 Mbps
A T3 circuit can carry 672 voice / data channels, with max throughput of 44.746 Mbps
Speed depends on a T-carrier’s signal level (= Physical layer signalling characteristics)
DS0 (Digital Signal, level 0) is the equivalent of one data / voice channel
A T3 provides 28 times more throughput than a T1
Leasing several T1s can be cheaper than leasing a single T3
A fractional T1 lease allows you to use (and pay for) only some of the T1 channels
Only businesses with extraordinary bandwidth requirements should consider using T3s
T-carrier connectivity


Wiring




Every T-carrier line requires connectivity hardware at the customer site and the local carrier’s POP
The specialised connectivity hardware can’t be used with other WAN transmission methods
T1 technology can use UTP / STP copper wiring (i.e. telephone wire)
Repeaters must regenerate the signal approximately every 6000 feet
For multiple T1s, coaxial cable, microwave, or fibre-optic cabling should be used
For T3s, microwave or fibre-optic cabling is necessary
CSU/DSU (Channel Service Unit / Data Service Unit)
 The connection point for a T1 line at the customer’s site
 The CSU provides termination for the digital signal and ensures connection integrity
 The DSU converts the digital signal used by bridges, routers, and multiplexers into the digital signal
sent via the cabling
 The CSU/DSU box connects the incoming T1 with the multiplexer
Multiplexer
 A device that combines multiple voice / data channels on one line
 Terminal equipment = the devices that connect to the multiplexer
 Multiplexers can take input from a variety of terminal equipment (bridges, routers…)
 In some network configurations, the multiplexer is integrated with the CSU/DSU
Routers and bridges
 On a typical T1-connected network, the terminal equipment will consist of bridges / routers / both
 With the T1 connection, the bridge / router typically integrates two types of networks: the Internet and
an Ethernet / Token Ring LAN at the customer’s site
DSL (Digital Subscriber Line)









Suited to the local loop portion of a WAN link, since repeaters are necessary after short distances
Can support multiple data and voice channels over a single line
A dedicated service, so a connection is always available for use
Uses advanced data modulation techniques to achieve extraordinary throughput over phone lines
Cheaper installation and faster throughput than T1s
More secure than cable
Advantages: High throughput, low cost, ease of installation
Disadvantage: Not available in all areas
Disadvantage: Relative newness has led to a backlog in installations, so subscribers must wait…
Types of DSL
 There are at least 8 types of DSL
 Downstream transmission = data travelling from the carrier’s POP to the customer
 Upstream transmission = data travelling from the customer to the carrier’s POP
Asymmetrical communications
 Downstream throughput is usually much higher than upstream throughput
 Well suited to users who pull more info off the network than they send
 E.g. ADSL (Asymmetric DSL), G.Lite (a version of ADSL), VDSL (Very high bit-rate DSL)
28
Symmetrical communications
 Provides equal capacity for data travelling both upstream and downstream
 Well suited to users who upload and download significant amounts of data
 E.g. SDSL (Symmetric / Single-line DSL), HDSL (High bit-rate DSL)
DSL types
 Vary in terms of their capacity and maximum line length
 Vary according to how they use the PSTN
 ADSL is the most popular form of DSL
DSL connectivity

Voice signals on the PSTN use a small range of frequencies, leaving higher, inaudible frequencies
unused and available for carrying data
 ADSL, G.Lite and VDSL use the same pair of wires that carry voice signals, but modulate data on the
higher frequencies
 ADSL: A splitter must be installed at the carrier & customer’s premises to separate the data signal from
the voice signal before it reaches the terminal equipment
 G.Lite: Eliminates the splitter, but requires a filter to prevent high-frequency DSL signals from reaching
the telephone
 HDSL, SDSL: Can’t use the same wire pair that is used for voice signals, but use the extra pair of wires
contained in a telephone cable (that are typically unused)
DSL modem
 Found at the customer’s office, to demodulate the signal and pass the info to the computer
 May contain a splitter (for ADSL) to separate the line into multiple channels for voice & data
 May be external to the computer and connect to a computer’s Ethernet NIC or USB port
 May connect to a connectivity device (hub / router) if bandwidth is to be shared on a LAN
DSLAM (DSL Access Multiplexer)
 Found at the other end of the line, at the carrier’s POP
 Aggregates multiple DSL subscriber lines and connects them to a larger carrier / Internet backbone
 (At the POP, there might be a splitter between the incoming line and the phone & data switches)
Cable










An asymmetrical technology
You need a special cable modem
Traditional cable TV supplies the infrastructure for downstream (but not upstream) communication
To provide Internet access through its network, the cable company must upgrade its equipment to
support bidirectional, digital communications:
The wiring must be replaced with HFC (Hybrid Fibre-Coax) - an expensive fibre-optic link
The HFC connects the cable company’s offices to a node location near the customer
Fibre-optic / coax may connect the node to the customer via a connection called a cable drop
The head end (the cable company’s central office) can connect to the Internet in different ways
Advantage: Provides a dedicated connection that doesn’t require dialling up a service provider
Disadvantage: Many subscribers need to share the same line, affecting security and throughput
SONET (Synchronous Optical Network)













Data transfer rates from 64 Kbps  39.8 Gbps
Uses the same TDM technique used by T-carriers
Was developed to link different phone systems around the world
Emerged as the best choice for linking WANs between America, Europe, and Asia, because it can work
directly with the different standards used in different countries
Internationally known as SDH (Synchronous Digital Hierarchy)
Integrates well with T-carriers, making it good for connecting WANs & LANS over long distances
Often used to aggregate multiple T1s or T3s
Also used as the underlying technology for ATM transmissions
Depends on fibre-optic media to achieve its extraordinary QoS and throughput
Uses multiplexers and terminal equipment to connect at the customer’s end (Like T-carriers)
A typical SONET network takes the form of a ring topology, similar to FDDI
One ring acts as the primary route for data and a second one acts as a backup
Self-healing = when SONET automatically reroutes traffic along the backup ring (very reliable!)
29



The data rate of a particular SONET ring is indicated by its Optical Carrier (OC) level
SONET technology is not implemented by small / medium businesses, because of its high cost
Particularly suited to audio, video, and image data transmission
WAN implementation
1. Speed

Speeds range from 56 Kbps for a PSTN dial-up connection to 39.8 Gbps for SONET
2. Reliability



Not very reliable:
Sufficiently reliable:
Very reliable:
PSTN dial-up
ISDN, T1, fractional T1, T3, DSL, cable, X.25, frame relay
SONET
3. Security




Provide encryption measures, and secure connectivity devices, like firewalls
Enforce password-based authorisation for LAN and WAN access
Enforce a security policy for users
Maintain restricted access to network equipment rooms and data centres
Virtual Private Networks (VPNs)







WANs logically defined over public transmission systems, isolating the organisation’s traffic from other
users of the same public lines
Provide a way of constructing a WAN from existing public transmission systems
Inexpensive solution for long distance WANs, because you don’t have to lease a full T1 circuit
Security techniques ensure that data can be interpreted only at the WAN’s nodes
The security techniques may be purely software-based or may include hardware, like a firewall
The software required to establish VPNs is usually inexpensive
A company might choose to implement a VPN to lower its WAN transmission costs
Remote connectivity
Three ways for a remote user to connect to a LAN:
1. Dial directly to the LAN, with a modem
 Remote access server = a combination of hardware & software that provides a central access point for
multiple users to dial into a LAN / WAN
 The LAN treats the direct-dial remote client like any other client on the LAN
 Remote node = the computer dialling into the LAN
 Advantage: Can provide the best security
 Advantage: The transmission speed doesn’t suffer when the Internet becomes congested
 Disadvantage: The most complex access method to configure
2. Dial directly to a workstation, with a modem
 Remote control = when the remote computer takes over a LAN computer, using special software
running on both computers
 Advantage: Not as difficult to configure and confers the same security and throughput benefits as
directly dialling into a remote access server
 Advantage: Provides the best performance for processing-intensive applications, like databases
 Disadvantage: Allows only one connection to the LAN at any given time
3. Use an Internet connection with a Web interface
 Requires some setup steps on client and server, but isn’t as complex as direct-dial configuration
 Advantage: A Web interface is very simple to use and widely available
 Advantage: A nearly unlimited number of remote users can simultaneously access the LAN
 Disadvantage: Security and throughput can’t be controlled as thoroughly as direct-dial, because the
remote user’s connection isn’t dedicated


A popular method for gaining remote access to a LAN is by using an ICA (Independent Computing
Architecture) client to connect with a remote access server
The simplest dial-in server is the RAS (Remote Access Service), which comes with Windows 2000
server
30
Serial Line Internet Protocol (SLIP) and Point-to-Point Protocol (PPP)
SLIP
PPP
Communications protocols that enable a workstation to connect to a server using a serial connection
(like a modem)
Necessary to transport Network layer traffic over serial interfaces, which belong to the Data Link layer
Encapsulate higher-layer networking protocols in their lower-layer data frames
Less popular
More popular for dial-up connections to the
Internet because it doesn’t require as much
configuration on the client side as SLIP does
You have to specify the IP address for client and
PPP can automatically obtain this information as it
server
connects to the server
Can carry only IP packets
Can carry many different types of Network layer
packets (IPX, AppleTalk…)
Supports only asynchronous data transmission
Supports both asynchronous and synchronous
transmission
Asynchronous
 Data being transmitted and received don’t have to conform to any predetermined schemes that specify
when they can transmit data.
 A node can transmit at any time, and the destination node must accept the data as it comes
 Start and stop bits ensure that the receiving node knows when it has received a complete frame
 Transmission occurs in random stops and starts
Synchronous
 The data being transmitted & received must conform to a timing scheme
 A receiving node knows it should be receiving data by looking at the time on the clock
 Start & stop bits are not necessary, because the clock indicates where transmission begins & ends
TCP/IP addressing & name resolution
Subnetting


The process of subdividing a single class of network into multiple, smaller networks
More efficient use of IP addresses because you don’t need to request another class of addresses when
you exceed your node limit
 You can use one class of addresses for several network segments
 A subnetted address includes network, subnet, and host information
 E.g. If you use a Class B network, you can use the third octet to subdivide the network into the
functional equivalent of 254 Class C networks
 Extended network prefix = the combination of an address’s network + subnet information
 If you don’t use subnetting, the extended network prefix = just the network portion of the IP address
 Devices external to a subnetted LAN don’t need to interpret the LAN’s subnetting information
Subnet masks
 Special 32-bit numbers that, when combined with a device’s IP address, inform the rest of the network
about the network class to which the device is attached
 Composed of 4 octets - 255 represents part of the extended network prefix; 0 = host information
 If you don’t specify a subnet mask, the default is 255.0.0.0 for a Class A network, etc.
 If you are using subnetting, the subnet mask for a class A network is 255.255.0.0
When not to use subnetting:
 If you have a very small network
 If you never want to connect to the Internet (and don’t need to follow IP addressing standards)
 If your organisation has more IP addresses than it can ever use
Gateways



A combination of hardware & software than enables 2 different network segments to exchange data
Facilitates communication between different subnets
Because one device can’t send data directly to a device on another subnet, a gateway must intercede
and hand off the information
31






Every device on a TCP/IP network has a default gateway that first interprets its outbound requests to
other subnets, and then interprets its inbound requests from other subnets
If your network has only one segment and you don’t connect to the Internet, your devices wouldn’t need
a default gateway because traffic wouldn’t need to cross the network’s boundary
Often, a default gateway isn’t a separate device, but a network interface on a router
When a router is used as a gateway, it must maintain routing tables as well
Gateways on the Internet handle a relatively small amount of addressing information, which is used to
forward data to another gateway that knows more about the data’s destination
Core gateways = the gateways that make up the Internet backbone
NAT (Network Address Translation)




The process of assigning an IP address to the client’s transmission when it reaches the IP gateway
IP gateways can hide the IP numbers assigned within an organisation
Hiding IP addresses protects your identity and gives you more flexibility in assigning addresses
Clients behind the gateway may use any (unrecognised) IP addressing scheme
Sockets & ports












Socket = a logical address assigned to a specific process running on a host computer
A socket forms a virtual connection between the host and client
The socket’s address combines the host computer’s IP address with the port number associated with a
process (The port number follows a colon after the IP address), e.g. 10.43.3.87:23
Primary advantage of sockets: They enable clients & servers to communicate more expeditiously
Default port numbers for commonly used TCP/IP processes have values lower than 255
Well-known ports = port numbers in the range of 0 to 1023
The use of port numbers simplifies TCP/IP communications and ensures that data are transmitted to
the correct application
You can configure port numbers through software
Ports 20 & 21: FTP
Port 23:
TELNET
Port 25:
SMTP
PORT 80:
HTTP
Host names & Domain Name System (DNS)
Domain names
 Every host is a member of a domain (= a group of computers belonging to the same organisation with
part of their IP addresses in common)
 Domain names are associated with a company / organisation, e.g. ibm.com
 ICANN has established conventions for domain naming in which certain suffixes apply
Host files
 The Internet used to rely on one text file (HOSTS.TXT) to associate names with IP addresses
 The growth of the Internet made the file impossible to maintain
 Within a company you may still find this older system of ASCII text files
 Each host is matched by one line identifying the host’s name & IP address, and an alias
 On a UNIX-based computer, a host file is called hosts and is located in the /etc directory
 On a Windows 9x computer it is called lmhosts and is located in the c:\windows directory
DNS



A hierarchical way of tracking domain names and their addresses
A TCP/IP service that belongs to the Application layer
The DNS database doesn’t rely on one file / server, but is distributed over several key computers
across the Internet to prevent failure if a few computers go down
 To route traffic more efficiently, the DNS is divided into 3 components:
1. Resolvers
 Any hosts on the Internet that need to look up domain name information
2. Name servers
 Servers that contain databases of names and their associated IP addresses
 Supply the resolvers with the information they require
 If a name server can’t resolve the IP address, the query passes to a higher-level name server
32
 Each name server manages a group of devices, collectively known as a zone
 Root server = a name server maintained by ICANN, at the highest level in the hierarchy
 ICANN acts as the ultimate authority and maintains 13 root servers around the world
3. Name space
 Refers to the actual database of Internet IP addresses and their associated names
 Every name server holds a piece of the DNS name space
 Resource record = a single record that describes one piece of info in the DNS database
 Resource record fields: name, type, class, time to live, data length, and the actual record data
BOOTP (Bootstrap Protocol)


1.
2.
3.
4.



A service to facilitate IP address management
Uses a central list of IP addresses and their associated MAC addresses to dynamically assign IP
addresses to clients
A client connects to the network and sends a broadcast asking to be assigned an IP address
This broadcast message includes the MAC address of the client’s NIC
The BOOTP server looks up the client’s MAC address in its BOOTP table
It then supplies: The client’s IP address, the IP address of the server, the host name of the server, and
the IP address of a default router
Thanks to BOOTP, a client doesn’t have to remember its own IP address
An ideal situation for diskless workstations (with no hard disks) i.e. network computers
A BOOTP table needs manual intervention and can be difficult to maintain on large networks
DHCP (Dynamic Host Configuration Protocol)
DHCP leasing
 A device borrows an IP address while attached to the network and relinquishes it when logging off
 A network administrator can configure the duration of the lease to be as short or as long as needed
1. When the client starts up, it broadcasts a DHCP discover packet to the DHCP/BOOTP server port
2. Every DHCP server connected to the same subnet as the client receives the broadcast request
3. Each DHCP server responds with an available IP address, while withholding it from other clients
4. The client accepts the first IP address that it receives
5. The client responds with a broadcast confirming to the DHCP that it wants to accept the address
6. All other DHCP servers see this confirmation and return their IP addresses to the pool
7. When the DHCP server receives the confirmation, it broadcasts an acknowledgement message
8. It also provides more info, such as DHS or gateway addresses that the client might have requested
Terminating a DHCP lease
 A DHCP lease can be manually terminated or expire based on the period established for it
WINS (Windows Internet Naming Service)







Used exclusively with systems that use NetBIOS
Provides a means of resolving NetBIOS names to IP addresses
A computer’s NetBIOS name and its TCP/IP host name are different entities
The WINS / NetBIOS relationship is just like the DNS / TCP/IP relationship
Unlike DNS, WINS is an automated service that runs on a server
Maintains a database on the server that accepts client requests to register with a NetBIOS name
WINS doesn’t assign names / IP addresses, but keeps track of which NetBIOS names are linked to
which IP addresses
 The WINS server can’t use a dynamic IP address; one must be assigned to it manually
Advantages:
 Guarantees that a unique NetBIOS name is used for each computer
 Support for DHCP
 Better network performance because there’s no broadcast traffic
Addressing in IPv6
Advantages over IPv4:
 More efficient header
 Inclusion of the IPSec security technique
 Better support for QoS
 Automatic IP address configuration
 Biggest advantage: Billions of additional IP addresses through the new addressing scheme
33
Differences:
IPv4
IPv6
Four 8-bit octets  32 bits
Eight 16-bit fields  128 bits
Binary numbers, separated by a dot:
Hex numbers separated by a colon:
123.45.67.89
F:F:0:0:0:0:3012:0CE3, or FF::3012:0CE3
Loopback address = 127.0.0.1
Loopback address = 0:0:0:0:0:0:0:1 or ::1
Each node has its own IP address
Each interface on a device has its own IP address
No distinction is made between an address Each address contains a format prefix at the
that represents one device / interface and
beginning of the address that indicates what type of
an address that represents multiple
address it is (Unicast / anycast prefix = FEC0 /
devices / interfaces
FE80; Multicast prefix = FF0x)
 Unicast address = one that represents a single interface on a device
 Multicast address = one that represents multiple interfaces (so there is no broadcast address)
 Anycast address = one that represents any one interface from a group of interfaces
TCP/IP troubleshooting
ARP





ARP table = the database that lists the associated MAC and IP addresses
Dynamic ARP table entries are created when a client makes an ARP request that can’t be satisfied by
data already in the ARP table
Static ARP table entries are those that someone entered using the ARP utility
The ARP utility provides a way of obtaining info from and manipulating a device’s ARP table
The ARP utility can be a valuable troubleshooting tool
PING (Packet Internet Groper)

A utility that can verify that TCP/IP is installed, bound to the NIC, configured correctly, and
communicating with the network
 Often used to determine whether the host is responding
 Uses ICMP to send echo request & reply messages that determine the validity of an IP address:
1. An echo request is sent out to another computer
2. The other computer rebroadcasts the signal in the form of an echo reply, to the sender
 Pinging = the process of sending this signal back and forth
 You can ping either an IP address or a host name
 By pinging the loopback address you can determine whether your TCP/IP services are running
Netstat
 Displays TCP/IP statistics and details about TCP/IP components and connections on a host
Information that can be obtained:
 The port on which a particular TCP/IP service is running
 Whether or not a remote node is logged into a host
 Which network connections are currently established for a client
 How many packets have been handled by a network interface since it was activated
 How many data errors have occurred on a particular network interface
Nbtstat



NetBIOS isn’t routable, but when encapsulated in another protocol, like TCP/IP, it is
On networks that run NetBIOS over TCP/IP, the nbtstat utility can provide info about NetBIOS statistics
and resolve NetBIOS names to their IP addresses
I.e. If you know the NetBIOS name of a workstation, you can use nbtstat to determine its IP address
Nslookup



Allows you to query the DNS database from any computer on the network
You can find the DNS host name of a device by specifying its IP address, or vice versa
Useful for verifying a host is configured correctly or for troubleshooting DNS resolution problems
Tracert (Traceroute)



Uses ICMP to trace the path from one node to another, identifying all intermediate hops in between
Useful for determining router / subnet connectivity problems
Useful for diagnosing network congestion / network failures
34
1. To find the route, traceroute transmits a series of UDP datagrams to a specified destination
2. The first three datagrams have their TTL set to 1 (which expire as they hit the first router)
3. When they expire, they are returned to the source (i.e. the node that began the traceroute)
4. In this way, traceroute obtains the identity of the first router
5. Then traceroute transmits datagrams with a TTL of 2 etc… until the destination node is reached
6. Traceroute also returns the time taken for the datagrams to reach each router in the path
Disadvantages:
 Traceroute can’t detect router configuration problems
 Traceroute can’t detect whether a router uses different send & receive interfaces
 Routers may not decrement the TTL value correctly at each stop in the path
Ipconfig





The TCP/IP administration utility for use with Windows NT, 2000, and XP
Command-line based
Provides info about a network adapter’s IP address, subnet mask, and default gateway
Useful for checking a computer’s TCP/IP configuration and managing TCP/IP settings
Also enables you to release & renew DHCP-assigned addresses
Winipcfg


Performs the same TCP/IP configuration & management as ipconfig, but for Windows 9x & Me
Supplies the user with a GUI
Ifconfig

The TCP/IP configuration & management utility used on UNIX systems
Applications & services
WWW

Unqualified host name = a host name minus its prefix and suffix, e.g. ‘weather’
E-mail



If a user can’t retrieve email, you must verify his TCP/IP settings
If a department can’t retrieve email, you must investigate problems with the department’s gateway
If an organisation can’t retrieve email, you must troubleshoot your mail server & Internet connection
FTP & TFTP (already covered)
Gopher






A text-based utility that allows you to navigate through a series of menus to find & read files
Not sophisticated enough to interpret document formatting commands, like HTML
Allows you to transfer files from one host to another by connecting with FTP
The first Internet interface to provide links from one host to another that are transparent to the user
You need a local gopher client and a gopher server
Rarely used today because Web servers and browsers have made it obsolete
Newsgroups





Messages are distributed to a wide group of users at once
Require news servers and, on the client side, email programs capable of reading newsgroups
Instead of using SMTP, messages are transported by the NNTP
Network News Transport Protocol supports reading messages, posting new messages, and transferring
news files between servers
News servers are organised hierarchically, similarly to DNS servers
E-commerce

One of the fastest growing sectors of the Internet
VoIP (Voice over IP)


The provision of telephone service over a TCP/IP network
Internet telephony = when VoIP is carried over by Internet
35

VoIP over private lines is effective and economical, and congestion can be easily controlled, resulting in
better sound quality than the Internet can provide
 Advantage: Low cost
 Disadvantage: Voice transmissions are distorted more than data by the wire’s quality of service
 Disadvantage: Voice transmissions are subject to distortion if the connection is too noisy
 To prevent delays, disorder, and distortion, a voice connection requires more dedicated bandwidth
 Technical obstacles have prevented Internet telephony from becoming widespread
 Billing network-based calls is a challenge because their costs depend on many factors
 Three categories of VoIP technology:
Phone-to-phone
1. A user picks up his phone to make a call
2. His phone is connected to a local telephone switch, which handles call routing for his business
3. The phone switch accepts his voice signals and passes them on to a gateway
4. The gateway connects the PSTN with a TCP/IP network (like the Internet)
5. The gateway digitises the analogue signals, compresses the data, & assembles them into packets
6. The packets traverse the network and are accepted by another gateway at the receiving end
7. The receiving gateway reverses what the transmitting gateway did
8. The voice signal is passed to another telephone switch, to which the other phone is connected
 Both gateways perform their functions simultaneously, so VoIP is full-duplex
PC-to-phone
1. A PC with a microphone, speaker, & appropriate software initiates calls over its network connection
2. Voice signals are converted to digital data by the PC’s software, and transmitted through the PC’s NIC
and over the network, like any other data
3. At the receiving end, the gateway accepts the data and translates it into voice signals
4. The voice signals are transmitted to a telephone switch to which a traditional phone is connected
PC-to-PC
1. A PC with a microphone, speaker, & appropriate software transforms voice signals into digital data
2. The data are transmitted over the network
3. On the other end, another PC with the same setup accepts the call from the network
 During the call, users speak into the microphone and listen via the speakers
 An IP address essentially becomes a phone number
Integrity and availability



Integrity refers to the soundness of a network’s programs, data, services, devices, and connections
Availability refers to how consistently and reliably a file / system can be accessed
Things that can compromise integrity & availability: security breaches, natural disasters, power flaws,
human error, malicious intruders
Guidelines for protecting your network:
 Prevent anyone other than the network administrator from opening / changing the system files
 Monitor the network for unauthorised access / changes
 Record authorised system changes in a change management system
 Install redundant components
 Perform regular health checks on the network
 Monitor system performance, error logs, and the system log book regularly
 Keep backups, boot disks, and emergency repair disks current and available
 Implement and enforce security and disaster recovery policies
Viruses


Programs that replicate themselves so as to infect more computers
Can damage files / systems, annoy users with flashing messages, cause the computer to beep, or may
cause no harm and remain unnoticed forever
Types of viruses
Virus type
Boot sector
viruses
Description
Reside on the boot
sector of a floppy disk
How it spreads
When you start the computer
with an infected floppy disk, the
virus is transferred to the boot
36
Damage
Inability of the file system to
access some workstation files
sector on the hard disk
Can travel between computers
that use different operating
systems
Macro viruses
Word-processing /
spreadsheet program
macro
File-infected
viruses
Attach themselves to
executable files
When the infected file runs, the
virus copies itself to memory,
and later attaches itself to other
files
Network virus
Designed to take
advantage of network
vulnerabilities
Worms
Technically not viruses,
but programs that run
independently and
travel between
computers
Propagate via network
protocols, commands,
messaging programs, and data
links
May be transmitted by any type
of file transfer, including email
Trojan horses
Technically not viruses,
but programs that claim
to do something useful
and instead harm your
computer
By downloading .exe files
Missing options from
application menus, damaged /
changed / missing files…
(Don’t affect performance or
behaviour)
Damaged files, file size
increases, changed icons,
strange messages…
You need to disinfect your
computer and reinstall all
software
Any type of anomaly, ranging
from strange pop-up messages
to file damage
Don’t alter other programs in
the same way that viruses do,
but they may carry viruses.
Symptoms: any type of
anomaly, ranging from strange
pop-up messages to file
damage
E.g. a file that’s supposed to
be a game could reformat your
hard disk instead
Virus characteristics
Encryption
 Virus-scanning software may search files for a recognisable string of characters that identify a virus
 If the virus is encrypted, it may thwart the antivirus program’s attempts to detect it
Stealth
 Some viruses hide themselves to prevent detection
 Typically, stealth viruses disguise themselves as legitimate programs or replace part of a legitimate
program’s code with their destructive code
Polymorphism
 Viruses change their characteristics every time they are transferred, making them harder to identify
 May use complicated algorithms and incorporate nonsensical commands to achieve change
 The most sophisticated and potentially dangerous type of virus
Time-dependence
 ‘Time bombs’ are programmed to activate on a particular date
Virus protection
Antivirus software
 Virus symptoms: increases in file sizes, slow programs, unusual error messages, unexpected loss of
memory, fluctuations in display quality…
Functions that antivirus software should perform:
 Detect viruses through signature scanning (comparing a file’s content with known virus signatures)
 Detect viruses through integrity checking (comparing current characteristics of files & disks against an
archived version to discover any changes)
 Detect viruses by monitoring unexpected file changes / virus-like behaviours
 Receive regular updates and modifications from a centralised network console
 Consistently report only valid viruses, rather than false alarms. Heuristic scanning is a fallible scanning
technique
 Popular antivirus packages: (McAffe’s) VirusScan, (Norton’s) AntiVirus, etc…
37
Antivirus policies
 Every computer should have virus detection & cleaning software that regularly scans for viruses
 Users shouldn’t be allowed to alter / disable the antivirus software
 Users should know what to do if their antivirus program detects a virus
 Have an antivirus team that focuses on maintaining the antivirus measures in place
 Users should be prohibited from installing any unauthorised software on their systems
 Organisations should impose penalties on users who don’t follow the antivirus policy
Virus hoaxes

Typically demand that you pass a virus alert to everyone you know, propagating the rumour
Fault tolerance
 The capacity for a system to continue performing despite an unexpected malfunction
 Failure = when something doesn’t work as planned
 Fault = when the malfunction of a system component is involved (Can result in a failure)
 The goal of fault-tolerant systems is to prevent faults from progressing to failures
Ways of achieving fault tolerance:
 An uninterruptible power supply provides high fault tolerance
 Two servers can mirror each other, taking over for the partner if it should fail
 (Fail-over = when one component assumes the duties of an identical one)
 Provide redundant elements to compensate for faults in critical components
Environment


Analyse the physical environment (heat, moisture, break-ins, natural disasters…)
Make sure the telco closets and equipment rooms are air-conditioned and at the right humidity
Power
Power flaws
Surge
 A momentary increase in voltage due to distant lightning / electrical problems
 Several surges can degrade a computer’s power supply
Line noise
 A fluctuation in voltage levels caused by other devices on the network or EMI
 E.g. if you turn on the printer and the lights dim
 Over time, electrical line noise will damage your PC’s internal circuit boards
Brownout
 A momentary decrease in voltage (also known as a sag)
 Can be caused by an overtaxed electrical system
Blackout
 A complete power loss
 May or may not cause significant damage to the network
Uninterruptible power supply (UPS)
 A battery-operated power source directly attached to devices and to a power supply
 Prevents undesired features of the wall outlet’s A/C power from harming the device
Standby UPS
 Provides continuous voltage by switching immediately to the battery when loss of power is detected
 Upon restoration of the power, it switches the device back to using A/C power again
 Disadvantage: In the brief amount of time that it takes the UPS to discover that the power has faltered,
a sensitive device (like a server) may have already detected the power loss and shut down
 Advantage: Cost significantly less than online UPSs
Online UPS
 Uses the A/C power from the wall outlet to continuously charge its battery, while providing power to a
network device through its battery
 Advantage: Offers the best kind of power redundancy available
 Advantage: No risk of momentarily losing service, because the server need never switch power
 Advantage: Can deal with noise, surges, and sags before the power reaches the device
Factors to consider when choosing a UPS:
 Amount of power needed (The more power required, the more powerful the UPS needed)
 Period of time to keep a device running (Most UPSs can support a device for 15 to 20 minutes)
38

Line conditioning (Surge suppression protects against surges and line conditioning; Filtering guards
against line noise)
 Cost
Generators
 Use if your organisation can’t withstand a power loss of any duration
 Can be powered by diesel, liquid propane gas, natural gas, or steam
 Don’t provide surge protection, but provide clean (free from noise) electricity
Topology


A mesh topology offers the best fault tolerance
SONET rings can easily recover from a fault because they are self-healing
Connectivity





‘Hot swappable’ refers to identical components (in the same device) that automatically assume the
functions of their counterpart if one suffers a fault
E.g. If a router’s processor fails, the redundant processor will automatically take over processing
Faults may also affect the connecting links
E.g. If your T1 connection fails, it doesn’t matter whether your router has redundant NICs
Load balancing = an automatic distribution of traffic over multiple links / processors to optimise
response
Servers
 Critical servers often contain redundant NICs, processors, and hard disks
Server mirroring
 A fault-tolerance technique where one server duplicates the transactions & data storage of another
 The servers involved must be identical machines using identical components
 You need a link between the servers, and software running on both servers that allows them to
synchronise their actions continually (and let one server take over in the event of a failure)
 Advantage: The servers involved can stand side by side (or be geographically far apart)
 Disadvantage: The time it takes for a mirrored server to assume the functionality of the failed server
 Disadvantage: The toll on the network as data are copied between sites
Server clustering
 A fault-tolerance technique that links multiple servers together to act as a single server
 Clustered servers share processing duties and appear as a single server to users
 More cost-effective than mirroring, because servers can perform services independently of others
 To detect failures, clustered servers regularly poll each other on the network
 If they don’t receive a response, the clustering software initiates the fail-over
 Unlike with mirroring, users won’t notice the switch (transparent fail-over and recovery processes)
 Advantage: Each server can perform its own data processing, improving performance
 Disadvantage: The clustered servers must be geographically close
Storage
RAID (Redundant Array of Inexpensive Disks)
 The multiple disks in a RAID drive appear as a single logical drive
 Advantage: a single failure won’t cause a catastrophic loss of data
 Typically used on servers, but not on workstations because of its cost
RAID level 0 - disk striping
 Data are written in blocks across all disks in the array
 Not fault-tolerant, because if one disk fails, the data in it is accessible
 Improves performance by utilising multiple disk controllers
 Multiple disk controllers allow several instructions to be sent to the disks simultaneously
RAID level 1 - disk mirroring
 Provides redundancy: data from one disk are copied to another disk
 If one disk fails, the disk array controller will switch to the disk that was mirroring the failed one
 Advantages: Simplicity, and automatic & complete data redundancy
 Disadvantage: Costly, because it requires two identical disks
 Disadvantage: Not too efficient, because it relies on software to do the mirroring, taxing the CPU
RAID levels 2 & 4
 Rarely used, because they are less reliable / efficient than the other levels
39
RAID level 3 - disk striping with parity
 ECC = error correction code
 Parity = the mechanism used to verify the integrity of data
 When data are written, a parity bit is assigned to each data byte
 When data are read, the parity is checked, and if it doesn’t match, damage is assumed
 Parity error checking = comparing the parity of data read with the type of parity used by the system
 A RAID level 3 system can automatically correct parity errors
 Advantage: High data transfer rate when reading / writing
 Disadvantage: Parity info appears on a single disk, representing a single point of failure
RAID level 5 - disk striping with distributed parity
 The most popular, highly fault-tolerant data storage technique used today
 Data are written in small blocks across several disks, and parity error checking info is distributed
 Advantage: Can write data more rapidly, because the parity info can be written by any one of the
several disk controllers in the array
 Advantage: Uses several disks for parity information, making it more fault-tolerant
 Advantage: You can replace failed disks with good ones without any interruption of service
Network Attached Storage (NAS)
 A specialised storage device (or group) that provides centralised fault-tolerant data storage
 Differs from RAID in that it maintains its own interface to the LAN, rather than relying on a separate
server to connect it to the network and control its functions
 Advantage over a file server: a NAS device contains its own file system that is optimised to save and
serve files. Because of this optimisation, NAS reads & writes faster than other types of servers
 Advantage: Can be easily expanded without interrupting service
 Although NAS is a separate device with its own file system, it still can’t communicate directly with
clients on the network:
1. When using NAS, the client requests a file from its usual file server over the LAN
2. The server then requests the file from the NAS device on the network
3. The NAS device retrieves the file and transmits it to the server, which transmits it to the client
 Appropriate for small or medium sized enterprises that require fault tolerance and fast data access
 Also appropriate for organisations that use a mix of different operating systems
Storage Area Networks (SAN)
 Distinct networks of storage devices that communicate directly with each other and other networks
 Multiple storage devices are connected to multiple, identical servers
 If one storage device within a SAN suffers a fault, data is automatically retrieved from elsewhere
 If one server in a SAN suffers a fault, another server steps in to perform its functions
 Extremely fault-tolerant and fast
 Fibre Channel = a network transmission method that relies on fibre-optic media and its own, proprietary
protocol
 Fibre Channel connects devices within the SAN and also connects the SAN to other networks
 Because it depends on Fibre Channel, a SAN is not limited to the speed of the client/server network for
which it provides data storage
 Since the SAN doesn’t belong to the client/server network, it doesn’t have to contend with the normal
overhead of that network (like broadcasts and acknowledgements)
 A SAN frees the client/server network from the traffic-intensive duties of backing up & restoring data
 Advantage: Highly scalable (You can easily add further storage and devices)
 Advantage: Faster than NAS devices
 Disadvantage: High cost
 Disadvantage: More complex than NAS or RAID systems
 Best suited to environments with huge quantities of data that must always be quickly available
 Typically used to house multiple databases
Data backup
Tape backups




The most popular method for backing up networked systems, because it’s simple and cheap
On small networks, standalone tape drives can be attached to each server
On large networks, one large centralised tape backup device can manage all subsystem backups
Extremely large environments may require robots to retrieve and circulate tapes from a vault
40
Online backups




Usually, online backup providers require you to install their client software
Online backups implement strict security measures to protect the data in transit
Disadvantage: The cost of this service can vary widely
Disadvantage: It may be difficult to verify that your data has been backed up successfully
Backup strategy







Full backup: All data on all servers are copied to a storage medium
Incremental backup: Only data that have changed since the last backup are copied
Differential backup: Only data that have changed since the last backup are copied, and that information
is then marked for subsequent backup, regardless of whether it has changed
Backup rotation scheme = a plan that specifies when and how often backups will occur
Grandfather-father-son scheme = a popular backup rotation scheme which uses daily (son), weekly
(father), and monthly (grandfather) backup sets
You should ensure that backup activity is recorded in a backup log
From time to time you should attempt to recover some critical files from your backup media
Disaster recovery
 The process of restoring your data after an outage that affects more than a single system
 In a disaster recovery plan, consider the worst-case scenarios, rather than minor outages
 The plan should outline multiple contingencies, in case the best options won’t do
Issues that the data recovery team should address:
 Contact names for emergency coordinators who will execute the disaster recovery response
 Details on which data and servers are being backed up and how data can be recovered
 Details on network topology, redundancy, and agreements with national service carriers
 Regular strategies for testing the disaster recovery plan
 A plan for managing the crisis, including regular communications with employees and customers
Project management
The project plan
 Project management software: Microsoft Project, Plan View, PrimaVera Project Planner
Elements of a project plan:
Task breakdown
 Large tasks are broken down into smaller subtasks
Dependencies
 You must specify which tasks depend on the completion of other ones before you can begin them
 Predecessor = a task that must be completed before another task can begin
Timeline
 Identify how long each task will take, with priorities
 Allow extra time for especially significant tasks
 A Gantt chart depicts when projects begin and end along a horizontal timeline
 One technique for making the project fit into a tight time frame is to work backwards
Resources
 Staff, materials, and money
 Owner = the person(s) responsible for tasks, who must ensure the project is completed on time
Milestones
 Reference points that mark the completion of major tasks
 Useful in large projects because they provide a quick indication of a project’s relative success
Project participants




Project sponsors = people in authority who supervise everyone involved in the project
Sponsors can ask for budget increases, deadline extensions, and help negotiate vendor contracts
A sponsor may be the person who originated the idea for the project
Stakeholder = any person who may be affected by the project
41
Funding

A project budget is usually set at the beginning, and approved by a hierarchy of managers
Communications
Reasons why communications are critical:
 To ensure that a project’s goals are understood by participants, stakeholders, and sponsors
 To keep a project’s timeline and budget on track
 To encourage teamwork among participants
 To allow you to learn from previous mistakes
 To prevent finger pointing if a task is not completed correctly / on time
 To avoid duplication of errors
 To prepare stakeholders for the effects of the change
Methods of communication:
 Weekly status meetings and daily status briefings for each team
 Weekly messages to stakeholders about the project’s progress
 Monthly reports that compare the anticipated and actual spending & timelines
 Distribution lists to share email correspondence
 A Web page containing an archive of meeting minutes
Processes


Process management = planning for and handling the steps needed to accomplish a goal in a
systematic way
Processes that can be managed: change, support, training, delegation, problem resolution…
Contingency planning
 The process of identifying steps that will minimise the risk of (endangering) unforeseen events
 Analyse the organisation’s history to identify potential threats
Possible measures to take:
 Order more hardware components than you think you need
 Ensure that your vendors have extra components on hand
 Document each piece of hardware & software ordered
 Rely on a pilot network to test your project’s goals
 If the technology required to implement the project is new to participants, ask a consulting company
with expertise in that technology to be available for questions in case you need help
Testing and evaluation
 Once you have reached a project milestone, you must verify that you are on the right path
 Establish a testing plan that includes relevant methods and criteria
A test plan should address at least the following issues:
 Was the change nominally successful?
 Did the change fully accomplish its purpose?
 If the change didn’t fully accomplish its purpose, did it partially accomplish its goal?
 Did the change result in unexpected consequences?
 Did the change point to a need for additional changes?
Managing network implementation
Implementation steps
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
Determine whether the proposed change is feasible, given time, resource and budget constraints
If a change is feasible and desirable, identify specific goals for the project
Assess the current state of the network, including physical and logical topology, protocols…
Assess the requirements as expressed by stakeholders
Create a project plan that includes tasks and subtasks, dependencies, resource allocation, timelines,
and milestones, and specify necessary hardware & software purchases
If possible, build a pilot network based on your recommendations
If the pilot network shows promise, begin to implement the changes on a larger scale
If possible, release the changes to a group of users who will evaluate the success of the changes
If the evaluation indicates that the changes were successful, release the changes to all users
Update your network baseline documentation to reflect the changes
42
Determining project feasibility

A feasibility study outlines the costs & benefits of the project and predicts if it will be a success
Setting project goals



Begin with a broad goal, then narrow it down into specific goals
The feasibility study should help determine whether you can achieve the project goals on time
A lack of well-defined goals can result in misunderstandings, lack of focus, lack of proper resource
allocation, and an uncertainty about whether the project’s outcomes constituted success
Baselining




Baselining = the practice of measuring and recording a network’s current state of operation
Includes keeping a history of performance measurements, like response times & no. of collisions
Also involves tracking the physical & logical topology, no. of devices, OSs and protocols in use…
Critical to network implementations because it provides the basis for determining which changes may
improve the network and also for later evaluating how successful those improvements were
Assessing needs and requirements
 Needs assessment = the process of clarifying the reasons underlying a proposed change
 People must be interviewed to compare perceptions to factual data
Some questions that can be asked:
 Is the expressed need valid, or does it mark a different need?
 Can the need be resolved?
 Is the need important enough to allocate resources to its resolution?
 If fulfilled, will the need result in additional needs? Will fulfilling it satisfy other needs?
 Do users affected by the need agree that change is a good answer?
User requirements
 A good technique for beginning to clarify user requirements is user interviews
 Sort out which needs have a greater priority, and how to address those in the minority
Some questions to ask:
 What makes you think this need should be addressed?
 How quickly do you think this need must be addressed?
 Can you suggest ways we can meet this need?
 What kind of priority would you place on this need?
 Are you willing to ignore other needs to have this need met?
Performance requirements
 Take the same approach as with interviewing users about their needs
Some questions to ask:
 Where do current performance bottlenecks exist?
 What kind of performance is optimal?
 What priority would you assign to improving performance?
 What measures can bring current performance levels to your recommended level?
 How will performance improvements affect access, availability, customer needs, security…
 How will you ensure that measures taken to improve performance are successful?
Availability requirements
 Interview technical staff (to find out how availability can best be achieved) and management staff (to
find out what types of availability are most important)
Questions to ask the technical staff:
 Where do current availability flaws / vulnerabilities exist?
 What kind of availability is acceptable? (99.5%? 99.9%?)
 What priority would you assign to improving availability?
 What measures can boost current availability to your recommended percentage?
 How will availability improvements affect access, performance, customer needs, security…
Questions to ask the management staff:
 What is the cost of one hour of down time during business hours?
 What is the cost of one hour of down time during off-hours?
 What is your ideal availability percentage?
 What part of the application or access is most important to keep available?
 What priority would you assign to improving availability?
 How much are you willing to spend to ensure that the network remains available?
43
Integration & scalability requirements
 Consider how the proposed change might affect the network’s integration and ability to grow
 Integration & scalability requirements are less likely to represent the primary reason for changes than
are customer, performance, or security needs
Questions to ask the technical staff:
 How and where is the network’s growth currently limited?
 What needs to change to accommodate growth or new hardware / software?
 In what ways do you expect the network to grow over the next two years?
 How will improving scalability and integration affect customers, performance, security, availability?
 How would you prioritise your suggested measures for accommodating growth?
Questions to ask the management staff:
 In what ways do you expect the network (and organisation) to grow over the next 1  5 years?
 Which of these growth directions is your top priority?
 What type of hardware and software do you expect to adopt in coming months and years?
 How much are you willing to spend to optimally position the network and systems for growth?
 Would you place a higher priority on positioning the network for growth / facilitating better systems
integration or on improving security, availability, usability, or performance?
Security requirements
 Security needs are typically identified by the technical staff
 Ask management staff how they would prioritise security improvements and how much they would be
willing to pay to improve network / systems security
Questions to ask the technical staff:
 What type of security must be improved (hardware, software, user, facilities)?
 Why does security need to be improved?
 To what extent does security need to be improved?
 Will the improvement require extra staff, hardware, software, or consulting services?
 What is the priority of security improvements?
 How will security improvements affect network access, performance, or scalability?
Using a pilot network




Pilot network = a small-scale network that stands in for the larger network
It shouldn’t be connected to your live network, in case it inadvertently causes harm
Document what you learn about the new technology’s features and idiosyncrasies
As you evaluate your results against your predefined test criteria, note where your results show
success or failure
Tips for creating a more realistic and useful pilot network:
 Include at least one of each type of device that might be affected by the change
 Use the same transmission methods and speeds as employed on your network
 Try to emulate the number of segments, protocols, and addressing schemes in your network
 Always implement the same server & client software and configurations as in your current network
 Test the pilot network for at least two weeks to verify that its characteristics meet your criteria
Preparing users

You must notify users of impending changes so that if something goes wrong they won’t be caught off
guard
Things to explain to users:
 How their access to the network will be affected
 How their data will be protected during the change
 Whether you will provide any means for users to access the network during the change
 Whether the change will require users to learn new skills
Highlighted facts = those that appeared in the October 2004 exam.
44