CIS4445 DATA COMMUNICATIONS

1
CIS-4445 DATA COMMUNICATIONS

• Chapter 6
• Local Area Networking

2
HOMEWORK - CHAPTER 6
Review Questions 3, 4, 6, 13, 16, 33
Exercises 1, 2, 3, 4, 6, 14, 18, 19, 21,
3
Introduction

• Network Topologies Review
• Ethernet IEEE Standard 802.3
• Token Ring IEEE Standard 802.5
• Token Bus IEEE Standard 802.4
• Interconnecting LANs
• Case Study Novell Netware

4
Network Topologies Review

• Remember there are four basic topologies plus a
  hybrid topology
• Connections between units (Segments)
• All of the topologies should support the OSI
  layered approach

5
Fully-Connected Topology
6
Common Bus Topology
Server
Printer
Workstation
Workstation
Workstation
Mainframe
Workstation
7
Star Topology
Printer
User
Communications Computer
Server
8
Ring Topology
9
Combined Topology
Server
Workstation
Local Area Network
BRIDGE
BRIDGE
Local Area Network
Workstation
Mainframe
Workstation
Printer
10
Network Topologies Review

• Remember there are four basic topologies plus a
  hybrid topology
• Connections between units (Segments) are normally
  made via
• Twisted Pair
• Coaxial Cable
• Optical Fiber
• All of the topologies should support the OSI
  layered approach

11
Segments

• The physical medium that makes it happen
• Limited by distance, speed, cost, and quality of
  link
• Choose type based on needs and cost
• Signaling technique used depends on medium used
  and distance required
• Pulses of photons for fiber
• Manchester Encoding
• Modulation Approaches

12
Network Topologies Review

• Remember there are four basic topologies plus a
  hybrid topology
• Connections between units (Segments) are normally
  made via
• All of the topologies should support the OSI
  layered approach
• Network topologies will for the most part focus
  on the lowest three layers of the OSI model

13
The OSI Model - 7 Layers
Logical Communications between layers
Application
Application
Presentation
Presentation
Session
Session
Transport
Transport
Network
Network
Data Link
Data Link
Physical
Physical
Physical transmission of data
14
The OSI Model - Overview
Application

• Network Layer
• Determines the best routes between the two points
  in the network - consider speed, cost, security,
  traffic etc.
• Should all the data go along the same route or
  should parts be transferred independently?
• Billing/Accounting information (who pays, how
  much based on time of day etc.)

Presentation
Session
Transport
Network
Data Link
Physical
15
Network Layer

• Transport Layer depends on the Network Layer to
  establish end-to-end communications
• Finds the best route between two points
• Considers cost, speed, availability
• Tries to find the quickest and cheapest route.
• Algorithms exist for this task e.g. Shortest
  Path algorithm
• For datagram circuit switching, must consider
  these things for each packet sent.
• For virtual circuit, only consider these when the
  connection is initially established.
• Accounting Billing
• Who must pay for the use of the network?
• Keeps stats on usage.

16
The OSI Model - Overview

• Data Link Layer
• Supervises the reliable transfer of data between
  two directly-connected nodes in the network
• Error detection and correction - request
  retransmission if error occurs
• Determines amount of information sent (frame
  size)
• Determines format of frames.
• Sends error-free frames to Network Layer

Application
Presentation
Session
Transport
Network
Error Detection/Correction
Data Link
Contention
CSMA/CD
Physical
Collision Detection
Token Passing
17
Data Link Layer - Contention

• It's when two or more devices try to use the same
  transmission medium at the same time.
• Collision results (data sent is garbled)
• Detecting collisions (Bus networks)
• Nodes listen for collisions while transmitting
• If collision occurs, stops transmitting and
  retries later.
• CSMA/CD
• Reduces collisions - does not eliminate them.

18
Data Link Layer - Contention

• Preventing collisions (e.g. Ring or Bus networks)
  
• Token passing
• A token is a special (unique) string of bits
• Only the node "with the token" may transmit
  anything
• Token is passed on, for example, either
• with a packet sent, or
• when the node has nothing more to send, or
• after it's had the token for more than a certain
  time

19
Data Link Layer - Error Detection/Correction

• Errors (flipped bits) can result from
• Bad connections
• Faulty lines
• Electrical interference
• Error detection is determining whether an error
  has occurred.
• Error correction is setting "damaged" bits to
  their correct state.
• Parity bit (one simple solution)
• Can detect whether a single bit is in error
• Even parity number of 1's (including the parity
  bit) is even.
• I.e. if frame has odd number of 1's, parity bit
  is 1, else it's 0.
• Problem if more than one bit is wrong it may not
  work.

20
The OSI Model - Overview

• Physical Layer
• Handles the physical (e.g. electrical or optical)
  aspects
• Sends frames received from Data Link Layer on
  communication medium without regard to their
  meaning or format
• Retrieves frames without regard to meaning or
  format, passing them to Data Link Layer for
  analysis

Application
Presentation
Session
Transport
Network
Data Link
Circuit Switching
Packet Switching - datagram - virtual
circuit
Message Switching
Physical
21
Physical Layer

• Transmission Media
• Determines how signals are sent
• Twisted wire pairs
• Coaxial cable
• Optical fibre
• Satellites
• Microwave and Radio towers
• Considers properties such as
• Analogue v/s digital transmission
• Bandwidth
• Signal-to-noise ratios

22
Physical Layer

• Connection Strategy
• How to connect two nodes
• Not which of the available routes is best (which
  is the job of the network layer
• Various strategies
• Circuit Switching
• Message Switching
• Packet Switching
• Datagram packet switching
• Virtual Circuit packet switching

23
Introduction

• Network Topologies Review
• Ethernet IEEE Standard 802.3
• Token Ring IEEE Standard 802.5
• Token Bus IEEE Standard 802.4
• Interconnecting LANs
• Case Study Novell Netware

24
Ethernet IEEE Standard 802.3

• Based on a Bus Topology Network
• Uses CSMA/CD contention protocol
• Commonly used throughout PC world
• Splits the Data Link Layer of the OSI model into
  2 sublayers
• Logical Link Control - IEEE 802.2 Standard
• Flow control
• Error detection
• Medium Access Control - IEEE 802.3 Standard
• Framing
• Controlling access to the media

25
Ethernet Modified Data Link Layer
Ethernet Implementation
OSI Model
Network
Network
Logical Link Control (LLC)
IEEE 802.2
Data Link
Medium Access Control (MAC)
Physical
IEEE 802.3
Physical
26
Ethernet IEEE Standard 802.3

• 802.3 is a MAC protocol
• Specifies
• Max. backbone length (including repeaters)
• Max. stations which can connect (500)
• Min. and max. distance between stations
• Cable type
• Data rates
• Frame structure (incl. min/max frame size)

27
Ethernet (IEEE 802.3) Components

• Cable (coax, fibre)
• Terminators (prevent signal echo)
• Transceiver (interface to PC)
• transmits/receives bits on medium
• listens to media (for CSMA/CD)

NIC
28
Ethernet (IEEE 802.3) Components

• Network interface card (NIC). Logic for
• buffering data
• sending/receiving frames to/from transceiver
• error checking
• framing
• retransmissions
• communication with the rest of the PC

NIC
29
Sequence of Events for Sending

• The sending PC uses network software that puts a
  packet of information in the PCs memory. It then
  signals the NIC via internal bus that a packet is
  waiting to go
• The NIC gets the packet and creates the correct
  frame format, storing the packet in the frames
  data field. It then waits for a signal from the
  transceiver, which is monitoring the bus segment
  waiting for a chance to send.
• When the transceiver detects a quiet cable it
  signals the NIC, which then sends the frame to
  the transceiver. The transceiver transmits the
  bits onto the cable, listening for any
  collisions. If none, it assumes the transmission
  was successful. If a collision occurs, the
  transceiver notifies the NIC. The NIC executes
  the exponential binary backoff.

30
Sequence of Events for Receiving

• The transceiver monitors the cable traffic. It
  copies frames from the cable and routes them to
  the NIC.
• The NIC performs CRC error check. If no error,
  NIC checks the destination address in the frame.
  If the frame is destined for its PC, the NIC
  buffers the frame in memory and generates an
  interrupt informing the PC that a packet has
  arrived.
• The PC uses network software to determine if the
  packet can be accepted according to flow control
  algorithms (like HDLC). If it can, the PC gets
  the packet from memory for further processing. If
  not, the network software responds according to
  the protocols at the next higher layer.

31
Cable Basics

• 10 BASE 2 ? means 10 Mbps, Baseband Signaling,
  200 meters maximum segment length
• Baseband cables
• All available bandwidth used to get single high
  bit rate.
• (Asynchronous TDM)
• Manchester encoding
• Broadband cables
• Bandwidth divided into sub-channels of smaller
  bandwidth on single cable.
• FDM
• Differential phase shift keying

32
Manchester Encoding
T, Duration of 1 bit
V
Magnitude
Time
-V
0
0
0
0
1
1
1
1
33
Cable Types for IEEE 802.3
34
Cable Basics

• Need to make more trade-offs
• Trade cost vs speed or
• Trade cost vs distance or
• Trade speed vs distance
• All the cable types for IEEE 802.3 do just that
• Provide the user with a choice that can
  approximate an optimal solution for that
  particular problem

35
Cable Basics

• 10BaseT
• Based on twisted wire pairs common in
  pre-computer office buildings
• TWP needs to have a minimum quality factor
• CAT5
• Use HUB approach
• looks like STAR topology but is really BUS
• HUB acts as repeater
• Limited to 100 meters
• Good for many applications

36
Cable Basics

• 100BaseT
• Basically a faster version of 10BaseT
• Still based on twisted wire pairs
• TWP still needs to have a minimum quality factor
• CAT5
• Still uses HUB approach
• Still limited to 100 meters
• UGH?!?
• How can this data rate be supported with
  basically the same hardware as 10BaseT?

37
Cable Basics

• 1000BaseT?? - ie Gigabit Ethernets
• Will that ever happen??
• Can TWP support gigabit data rates??

38
IEEE 802.3 Frame Format

• "Octet" 8 bits
• Preamble (to synchronise) - 7 Octets of 01010101
• "Start of frame" delimiter - 10101011
• Destination address - 0 is unique, 1 is group
  address
• Source address - where the frame came from
• Length - specifies the number of octets in the
  data and pad fields
• Min. 46 octets (frame512 bits) for collision
  detection
• Max. 1500 octets to limit access to medium.
• Data padding (if necessary)
• Frame Check Sequence - CRC-32 checksum

39
Ethernet Frame Format
7
1
2 or 6
2 or 6
Preamble
Start of Frame Delimiter
Destination Address
Source Address
2
4
46 - 1500
Data Field Length
Data Pad
Frame Check Sequence
40
Ethernet Frame Format

• Preamble - alternating pattern of 0s and 1s
  used for synchronization
• Start of Frame delimiter - Special pattern
  10101011 indicates the start of a frame
• Destination Address - if first bit is a 0,
  specifies a specific station. If a 1, specifies a
  group address

41
Ethernet Frame Format

• Source address - Specifies where the frame comes
  from
• Data length field - Specifies the number of
  octets in the combined data and pad fields
• Data field - self-explanatory
• Pad field - extra octets added to make 46 if
  necessary
• Frame check sequence - Error checking using
  32-bit CRC

42
Efficiency

• Define efficiency as the average amount of time
  to make a successful transmission
• Define T as the maximum amount of time to
  detect a collision which is also the maximum
  round trip propagation time between the two
  farthest points on the network

Data frames collide at dDistance/2
d
d
Distance
43
Efficiency

• What are the chances that a station will
  successfully send a frame during a time slot?
• Depends on N - the total number of stations
• Depends on probability of a given station sending
  during the slot ps (0 lt ps lt 1)
• pns 1- ps

Probability successful
44
Efficiency

• What conditions will allow the largest number of
  frames to be sent successfully?
• Find the largest value for P
• Take derivative and set equal to zero to find

Probability successful
Deriv
Set 0
45
Efficiency

• P will be largest when ps 1/N!
• As N increases, the likelihood of success
  decreases
• What will happen to P for large LANs?

46
Efficiency

• How many time slots before a successful
  transmission?
• Call this the Contention Period
• Define C as weighted average
• The number of slots a station waits times the
  probability of waiting that many slots

47
Efficiency

• Recognize that the probability that a frame is
  not sent successfully during one time slot is
  (1-P)
• Then the probability of no successful
  transmissions for i slots is (1-P)i

In the limit as i approaches infinity
48
Efficiency

• This says that as the probability of success goes
  up the contention period goes down
• How do you interpret this for large N?

All of this assumes a probability of sending that
changes as the number of stations changes
49
Efficiency

• Define percent utilization (U) as the amount of
  time spent on transmitting a frame as a
  percentage of the total time spent on contending
  and transmitting

Define R Transmission rate
F number of bits in a frame
T slot time
Total contention time is then TC TxC
So percent utilization
50
Efficiency

• So what the ideal utilization?

R Transmission rate 10 Mbps
T Slot time 512/10M 51.2 ?sec
F number of bits in a frame 512 Bytes
N 500 stations
Total contention time is then TC TxC
51
Efficiency
So percent utilization
52
Efficiency

• This assumes that the probability of sending is
  1/N1/500 0.002
• What about ps .2?
• Now P success is going to change

53
Efficiency
So real percent utilization
54
Introduction

• Network Topologies Review
• Ethernet IEEE Standard 802.3
• Token Ring IEEE Standard 802.5
• Token Bus IEEE Standard 802.4
• Interconnecting LANs
• Case Study Novell Netware

55
Token Ring
NIC
NIC
Token circulating around the ring
NIC
NIC
NIC
56
Token Ring (IEEE 802.5)

• MAC protocol (as with Ethernet) sitting between
  the Logical Link Control and the Physical layer
  in the OSI model
• Token passing between neighboring stations
• IBM claims 4 Mbps, 16 Mbps, or 100 Mbps
• Standard defined for 1 Mbps or 4 Mbps
• Differential Manchester Encoding
• Uses a deterministic approach for allowing
  transmissions to occur therefore no collisions
  occur

57
Token Ring

• Special frame called the token controls access
• Station can send only when it has the token
• Token is passed on if nothing to send
• Sending
• Attach the data to the end of a token frame
• Mark the token as a "data frame"
• Send frame along

58
Differential Manchester Encoding
T, Duration of 1 bit
Signal level at start of transmission
V
Magnitude
Time
-V
0
1
0
1
1
0
1

• 0 - the signal changes value
• at the start of the interval

• 1 - the signal remains where it was
• at the end of the previous interval

59
Token Ring (IEEE 802.5)

• Problem A failure in the traditional Token Ring
  results in a network failure
• Solution Use a Wire Center
• Layout similar to Star topology
• Keeps logical ring structure (send to neighbour)
• Relay switches route data between stations
• If a station fails the wire center relays switch
  to bypass it

60
Token Ring
Token circulating around logical ring
NIC
NIC
Wire Center
NIC
NIC
NIC
61
Token Ring (IEEE 802.5)

• Frame doesn't always fit on ring if the ring is
  not long enough
• Token is 24 bits long
• Delays are added if necessary

62
Token Ring (IEEE 802.5)

• Example
• 20 stations, 10 meters apart, 200 meters total
  length around ring
• At 4Mbps, one bit every 0.25 ?sec
• Propagation speed is 200 m/?sec
• Can only fit 4 bits on entire ring
• Use monitor station to add 20 bit delays so that
  token logically fits on ring

63
Token Format
3 Octets
SD
AC
ED
SD (Starting Delimiter) J K 0 J K 0 0 0
AC (Access Control) p p p t m r r r
p p p Priority bits
t token bit
m monitor bit
r r r reserevation bits
ED (Ending Delimiter) J K 1 J K 1 I E
64
Token Frame Format
1
1
1
2 or 6
2 or 6
0 - 5000
4
1
1
SD
FC
Destination Address
Source Address
data...
Frame check sequence
AC
ED
FS
FC (Frame Control) f f z z z z z z
f f frame type bits
z z z z z z control bit
FS (Frame Status) a c x x a c x x
a address recognized bit
c frame copied bit
x undefined bit
65
Token Frames

• Each token has a start delimiter octet and an
  ending delimiter octet that designate the tokens
  boundaries
• The SD starting delimiter has a specialsignal
  pattern (JK0JK000)
• 0s are binary 0s as defined by differential
  Manchester encoding
• J and K correspond to special signals
• J starts out as a zero but has no transition in
  the middle
• K starts out like a one but has no transition in
  the middle

66
SD Octet Encoding using Differential Manchester
Encoding
Signal level at start of transmission
V
Magnitude
Time
-V
K
J
0
K
J
0
0
0

• The J signal starts out like a zero but
• does not transition in the middle

• The K signal starts out like a one but
• does not transition in the middle

67
Token Frames

• The ED ending delimiter has a specialsignal
  pattern also (JK1JK1IE)
• 1s are binary 1s as defined by differential
  Manchester encoding
• J and K correspond to special signals
• J starts out as a zero but has no transition in
  the middle
• K starts out like a one but has no transition in
  the middle
• I is the intermediate frame bit which equals 1
  unless it is the last frame
• E is an error bit and is set to 1 whenever an
  error is detected (ie. Failed check)

68
ED Octet Encoding using Differential Manchester
Encoding
Signal level at start of transmission
V
Magnitude
Time
-V
K
J
1
K
J
1
I0
E0

• The J signal starts out like a zero but
• does not transition in the middle

• The K signal starts out like a one but
• does not transition in the middle

69
Token Frames

• The AC access control octet handles frame type,
  ring maintenance, and token reservation
• t token bit (0 if token, 1 if data frame)
• n used for ring maintenance
• ppp priority bits
• rrr reservation bits
• FS (Frame Status) octet - tells the sending
  station if the destination station is on the ring
• a address recognised (destination saw the
  packet)
• sending station sets to 0
• c frame copied ('ACK' for frame)
• sending frame sets to 0

70
Reserving/Claiming Tokens

• Initially Simple solution
• Pass token to next station
• If data to send and token arrives, use it
• Need priorities for time-sensitive applications
• Stations have priorities
• Token has priority (ppp bits of AC)
• Claim token only if station priority gt token
  priority, else pass it on.

71
Token Ring
Station A Priority1
Token circulating around the ring with priority
ppp and reservation rrr
NIC
NIC
Station D Priority2
NIC
NIC
Station B Priority5
Station C Priority0
72
Reserving/Claiming Tokens

• Reserving the Token
• Reserve the token for future use when priority to
  send is not high enough
• Data frame to send?
• Check priority -
• lower? - claim the token and send data
• too high? - Check reservation and try to reserve
  token
• Check reservation
• lower? - place station priority in reservation
  bits to place reservation
• too high? - cant do anything, send on

73
Reservation System

• Consider two distinct cases
• Data Frame arrives
• Token arrives

74
Data Frame arrives

• if frame originated at this station
• Remove frame from ring
• examine rrr of incoming frame -
• if frame rrr gt frame ppp
• Create token
• new ppp rrr from incoming frame
• new rrr 0
• push old and new ppp onto stack
• become the stacking station
• Send token DONE

75
Data Frame arrives

• else if this is the frame's stacking station
• Create token
• new ppp max( rrr, top_of_stack)
• for rrr, replace the current priority on the top
  of stack with it
• for top_of_stack, pop stack
• If stack empty, no longer tracking station
• Send token DONE
• else
• Create token
• retain ppp and rrr from data frame
• Send token DONE

76
Data Frame arrives

• Else frame not originating from this station
• Reserve frame if data to send and rrr lt station
  priority.
• Continue sending frame DONE

77
Token arrives

• If this is the token's stacking station
• Create token
• ppp max( rrr from token, top_of_stack)
• for rrr, push rrr from token onto stack
• for top_of_stack, pop stack
• If stack empty, no longer tracking station
• rrr retained

78
Token arrives

• If data frame to send
• if Station priority gt ppp
• ppp station priority
• rrr 0
• Send data frame DONE
• else
• reserve token if rrr lt station priority
• Send token

79
Ring Maintenance

• Orphaned frames (not removed by sender)
• One station is the monitor station
• Sets "monitor" bit of AC from 0 to 1 as frame
  passes
• If bit arrives set to 1, must have already passed
  
• gt Remove it replace it with a token

80
Ring Maintenance

• No Token (station crashed before sending it)
• Timer set (depends on size of ring, max frame
  size, number of stations)
• If timer expires before token seen, generate new
  token

81
Ring Maintenance

• Node entering the ring
• Node sends itself a frame
• FC field set as "Duplicate Address Test"
• frame circulated check "a" field of FS
• if 1, other station with this address exists

82
Ring Maintenance

• Electing a new monitor
• Determining address of previous station
• Detecting position of cable breaks
• Ensuring token fits on the ring
• Who fixes damaged frames?

83
Token Ring Control Frames
84
Slotted Ring
NIC
NIC
NIC
NIC
Token circulating around the ring
NIC
85
Slotted Ring

• "Slots" (similar to tokens) circulate the ring
• As empty slot passes station can put data in it
• When slot returns
• May not be used immediately
• Must wait for the next free one
• Means that free slots are available often
• Enough bits can fit onto the ring (add delays if
  necessary)

86
Introduction

• Network Topologies Review
• Ethernet IEEE Standard 802.3
• Token Ring IEEE Standard 802.5
• Fiber Distributed Data Interface
• Token Bus IEEE Standard 802.4
• Interconnecting LANs
• Case Study Novell Netware

87
FDDI (Fiber Distributed Data Interface)

• Similar to Token Ring over optic fiber
• 100 Mbps, 200 km ring, 100 stations
• 2 rings
• Both can carry tokens/frames
• Create single ring if one node fails
• Frame format similar to token ring

88
FDDI (Fiber Distributed Data Interface)

• Not differential Manchester encoding
• NRZI (nonreturn-to-zero-invert)
• constant signal while 0 sent
• changes if 1 sent
• 4-of-5 code
• See table 6.4
• Limits max. number of consecutive 0's
• 25 overhead acceptable

89
FDDI (Fiber Distributed Data Interface)

• Token-claiming method different
• Token sent immediately after data sent
• Any station can send data as token passes it
• CDDI (Copper DDI)
• 100 Mbps over copper wire use FDDI specs

90
FDDI (Fiber Distributed Data Interface)

• FDDI - II
• Response to the growing need of multimedia
  applications
• Incorporates a Hybrid mode that allows for
  circuit switching
• similar to telephone systems in use now

91
Introduction

• Network Topologies Review
• Ethernet IEEE Standard 802.3
• Token Ring IEEE Standard 802.5
• Fiber Distributed Data Interface
• Token Bus IEEE Standard 802.4
• Interconnecting LANs
• Case Study Novell Netware

92
Token Bus
E
A
F
NIC
NIC
NIC
NIC
NIC
NIC
Token
C
D
B
Token circulates A-B-C-D-E-F
93
Token Bus

• All stations are numbered
• Physical bus, logical ring
• Token passed along via a shared bus in order of
  number
• Each node must know predecessor and successor
• Difficult to add/remove stations
• If one station hangs?
• Deterministic - theoretical limit on max time
  required for token to return

94
Token Bus

• Baseband/broadband
• Modulation schemes
• continuous FSK - Baseband
• phase coherent FSK - Baseband
• AM/PSK - Broadband
• Frame structure much like FDDI ones

95
Token Bus
96
Removing Stations

• Must inform predecessor/successor
• Detecting unresponsive system
• Station A sends frame to B
• A listens afterwards does B send anything?
• If not (timeout), try again
• If fail again, assume B not responding
• Find B's successor (send "who follows" frame).
• B's successor (C) responds
• A and C update themselves for new state
• If leaving ring, station informs predecessor and
  successor

97
Introduction

• Network Topologies Review
• Ethernet IEEE Standard 802.3
• Token Ring IEEE Standard 802.5
• Fiber Distributed Data Interface
• Token Bus IEEE Standard 802.4
• Interconnecting LANs
• Case Study Novell Netware

98
Interconnecting LANs

• Reasons for choosing specific LAN type depend on
  goals it must fulfil.
• Machines on different LANs may need to
  communicate
• Hardware/software depends on how compatible the
  LANs are.
• Identical LANs easy
• Different LANs difficult (depends on type)

99
Interconnecting Networks
B
Token ring
Wide area network
connector
connector
connector
File Server
A
Texas
Ethernet LAN
Ethernet LAN
New York
C
File Server
File Server
100
Interconnecting LANs

• Connecting at various OSI levels
• Layer 1 (physical) gt repeater
• Layer 2 (data-link) gt bridge
• Layer 3 (network) gt router
• Layer 7 (application) gt gateway convert entire
  protocol to another

101
The OSI Model - 7 Layers
Gateway
Application
Application
Presentation
Presentation
Session
Session
Transport
Transport
Router
Network
Network
Bridge
Data Link
Data Link
Repeater
Physical
Physical
Physical transmission of data
102
Interconnecting at Layer 1

• Use repeater (connect at physical layer)
• Receives frames travelling on one LAN
• Buffers them (if necessary)
• Retransmit
• On all other LANs it's connected to
• Same protocol
• Same frame format
• Simply extends distance LAN covers

103
Interconnecting LANs with Repeaters
A
E
B
LAN 1
LAN 4
Repeater
Repeater
LAN 2
Repeater
D
LAN 3
C
104
Interconnecting at Layer 1

• Disadvantages
• Packets sent to stations on same LAN still sent
  to everyone else (wasted transmissions lower
  performance)
• Security (everyone can see everything sent)

105
Interconnecting at Layer 2

• Use a Bridge to connects at data-link layer
• Executes a subset of a protocol
• Error detection/correction
• Frame formatting
• Frame routing
• Frames forwarded depending on their destination
• if only path to destination is through the bridge

106
Interconnecting at Layer 2

• Advantages over repeaters
• Increased efficiency (less unnecessary traffic)
• Stations on same LAN can communicate without
  affecting stations on other LANs
• Increased security
• Bridges can prevent frames reaching destination
  other than the ones intended

107
Interconnecting LANs with Bridges
A
D
LAN 1
LAN 4
Bridge B1
B
Bridge B4
LAN 2
Bridge B2
Bridge B3
LAN 6
LAN 3
LAN 5
F
E
C
108
Problems bridging different types of LANs

• Different bit rates (e.g. fast to slow)
• Buffering may be required
• Delays at bridges affect flow-control time-outs
• Different frame formats
• Not always simple rearrangement of data
• Some fields discarded, others added
• e.g. priorities Token Ring to Ethernet

109
Problems bridging different types of LANs

• (Maximum) Frame sizes
• e.g. Token Ring frames gt Ethernet frames
• Use only small frames, or
• Split frames
• Must send correct ACKs back!
• Requires functionality above data-link layer
  (properties of a router)
• Brouter
• A device that maintains bridge function and
  assumes some router functions as well

110
Using a Brouter to Interconnect Different Types
of LANs
A
D
B
Ethernet LAN
Token Bus LAN
5000 byte frame
Repeater
Brouter
Four frames each less than 1528 bytes
Ethernet LAN
C
111
Routing Data using Bridges

• Process of deciding which ones and where is
  called Bridge Routing
• How does bridge know whether to forward or ignore
  frame?
• How to detect whether station moved adapt?
• Bridges don't have global view of network

112
Fixed-Routing Bridges

• Routing Table stored on each bridge
• Destination source/dest LAN
• For each frame
• Consider destination address and LAN it arrived
  on.
• Use table to get destination LAN (if any) and
  retransmit.

113
Bridge Routing Tables
Bridge B1
Bridge B2
Bridge B3
Bridge B4
114
Fixed-Routing Bridges

• Routing tabled fixed can't handle
• Machines moving location
• Addition/removal of machines/LANs
• Choices
• Reprogram routing table after each change
• Allow bridges to adapt dynamically

115
Transparent Bridges

• Create and update their own routing tables
• IEEE 802.1d
• Start with empty table
• Need to "learn" the layout
• Plug them in and they work immediately regardless
  of the topology and the stations locations

116
Route Learning

• For each frame going past, consider source
  address and LAN it came from
• Source must be accessible from this LAN
• Update routing tables if necessary
• If station moves and doesn't send anything, table
  never updated
• Timer associated with each entry in table
• Set each time entry updated
• If expires, entry discarded (may be inaccurate)

117
Route Learning

• What to do if frame arrives and no entry for
  destination in routing table
• Flooding Algorithm
• Send frame on all other LANs
• Guarantees delivery
• Allows other bridges to update tables to source
• Used to initialize empty table

118
Bridge Routing Table Changesusing Route Learning
Bridge B1
Bridge B1
Changes to table
Before D moves to LAN 1
After D moves to LAN 1
119
Frame Propagation Problems

• Problems with certain topologies
• if more than 1 bridge connected to LAN (in case
  of failure)
• Can result in endless propagation of frames
• Complex topologies can present many possible
  cases for endless loop propagation

120
Propagating Frames because of Multiple Bridges
A
LAN 1
Bridge B2
Bridge B1
LAN 2
B
If B doesnt reply to a frame, the two bridges
will continue to forward the frame
121
Frame Propagation Solutions

• Use Spanning Tree Algorithm
• Don't use duplicate bridges unless failure occurs
• Use a minimal set of edges based on cost for
  each bridge-to-LAN connection
• Bridges elect a root bridge from which all
  connection costs are calculated
• Result
• Spanning Tree connects all of the LANs even
  though some of the bridges are not used

122
Executing the Spanning Tree Algorithm

• Elect a root bridge
• Each bridge determines its root port
• Designate a bridge for each LAN

123
Multiple LANs with Loops
A
LAN 1
Cost 4
Bridge B1
Cost 4
Cost 6
Cost 2
Bridge B5
Bridge B6
LAN 2
Cost 5
Cost 1
Cost 2
Bridge B3
Cost 6
Cost 3
Bridge B2
LAN 3
Cost 6
Cost 4
Bridge B4
Cost 5
LAN 4
B
124
Multiple LANs with Loops - Graphical
Representation Before Determining Root Ports
L1
4
B5
4
B1
6
1
B6
2
B3
L3
2
3
L2
6
5
6
B4
5
4
L4
B2
125
Multiple LANs with Loops - Graphical
Representation After Determining Root Ports
L1
4
B5
Root
4
B1
6
1
B6
2
B3
L3
2
3
L2
6
5
6
B4
5
4
L4
B2
Denotes Cheapest route
Indicates Root Port
126
Resulting Spanning Tree
L1
B5
Root
4
B1
B6
2
B3
L3
2
3
L2
6
B4
4
L4
B2
Bottom Line Eliminates Loops
127
Multiple LANs with Loops Taken Out
A
LAN 1
Cost 4
Bridge B1
Cost 2
Bridge B5
Bridge B6
LAN 2
Cost 2
Bridge B3
Cost 6
Cost 3
Bridge B2
LAN 3
Cost 4
Bridge B4
LAN 4
B
128
Frame Propagation Solutions

• Source Routing Bridges
• Stations (not bridges) determine best route
• Network software at the sending station
  determines route information and stores it in
  each frame
• Bridge looks at frame and sees if the route
  designator contains its ID
• If it does, bridge accepts the frame and forwards
  it to the next LAN specified in the next
  designator
• Determine route by Route Discovery

129
Sending a Frame From A to B
A
A designates the route L1-B5-L3-B4-L4
LAN 1
Bridge B1
Bridge B5
Bridge B6
LAN 2
Bridge B3
Bridge B2
LAN 3
Bridge B4
LAN 4
B
130
Introduction

• Network Topologies Review
• Ethernet IEEE Standard 802.3
• Token Ring IEEE Standard 802.5
• Fiber Distributed Data Interface
• Token Bus IEEE Standard 802.4
• Interconnecting LANs
• Case Study Novell Netware

131
Case Study Novell NetWare

• Network Operating System.
• User know many machines make up network
• User can log onto remote machines, and can access
  files etc.
• Each machine runs own operating system
• Protocols to allow communication between PC's,
  printers, Apple Mac's, etc.
• Ethernet, Token Rings and ArcNet.
• Users can share printers, data, software.
• Various versions

132
Configuration

• One/more File Servers
• Stores shared data
• Runs all NetWare protocols
• Security (login, privileges)
• Collects stats on network use
• Bridge/router functionality
• Dedicated (only runs NetWare software)
• Non-Dedicated (runs NetWare DOS apps)

133
Configuration

• One/more Print Servers
• Allows users to access printers
• Queues, priorities
• Client PC's (workstations)
• Licensing agreement limits max number of
  workstations.
• NetWare's Modular design
• Load/unload modules (NLMs) to change
  configuration
• e.g. adding print servers, drivers, protocols
  (e.g. TCP/IP), email, etc

134
Configuration

• Commands from user
• Pass through NetWare software (NETX.COM under
  DOS)
• If DOS service, pass it on to DOS
• If NetWare service
• Create packet for the request
• Give packet to IPX (internet packet exchange)
• IPX attaches source/dest addresses sends it to
  server
• Server extracts request and performs appropriate
  action.

135
Configuration

• IPX maintains connection between devices
• IPX doesn't guarantee delivery SPX does (using
  ACKs)
• Network driver
• Converts IPX packets into frames
• IPX independent of underlying network type.
• ODI (Open Data Interface)
• Allows software to use multiple protocols using
  one network card

136
Security

• Passwords at login to fileserver
• Rights to resources based on trustee groups

137
Data Integrity

• Transaction Tracking System (TTS)
• Regard multiple transactions as a unit
• "Rollback" in event of failure
• Disk Duplexing
• Disk Mirroring

138
End Chapter 6
139
Problem 18 - Build Routing Tables
D
F
LAN L4
LAN 4
Bridge B5
Bridge B4
LAN L2
B
Bridge B2
Bridge B3
LAN L3
LAN L5
E
Bridge B1
C
LAN L1
A
140
Bridge Routing Tables
Bridge B1
Bridge B2
Bridge B3
Bridge B4
141
Problem 6.31 - Bridge B1 fails
A
LAN 1
Cost 4
Bridge B1
Cost 4
Cost 6
Cost 2
Bridge B5
Bridge B6
LAN 2
Cost 5
Cost 1
Cost 2
Bridge B3
Cost 6
Cost 3
Bridge B2
LAN 3
Cost 6
Cost 4
Bridge B4
Cost 5
LAN 4
B
142
Executing the Spanning Tree Algorithm

• Elect a root bridge - its now B2
• Each bridge determines its root port
• Designate a bridge for each LAN

143
Root Bridge B1 Fails - Need to Reconfigure
L1
4
B5
X
4
B1
6
1
B6
2
B3
L3
2
3
L2
6
5
6
B4
5
4
L4
B2
144
Bridge B2 Elected Root Bridge
Denotes Cheapest route
Indicates Root Port
L1
4
?
B5
X
B1
6
Cost 3
1
B6
?
B3
L3
2
3
L2
Cost 5
6
Cost 2
5
6
B4
5
4
L4
B2
Root
Cost 5
145
Resulting Spanning Tree
L1
4
B5
X
B1
1
B6
B3
L3
2
3
L2
6
B4
4
L4
B2
Root
Designated Bridges B5 for L1 B2 for L2 and
L4 B3 for L3
146
Resulting Network Configuration
A
LAN 1
Cost 4
X
Cost 4
Bridge B1
Cost 2
Bridge B5
Bridge B6
LAN 2
Cost 2
Cost 1
Cost 6
Bridge B3
Cost 3
Bridge B2
Root
LAN 3
Bridge B4
Cost 4
LAN 4
B
147
Efficiency

• Define percent utilization (U) as the amount of
  time spent on transmitting a frame as a
  percentage of the total time spent on contending
  and transmitting

Define R Transmission rate
F number of bits in a frame
T slot time
Total contention time is then TC TxC
So percent utilization
148
Efficiency

• So what the ideal utilization?

R Transmission rate 10 Mbps
T Slot time 512/10M 51.2 ?sec
F number of bits in a frame 512 Bytes
N 500 stations
Total contention time is then TC TxC
149
Efficiency
So percent utilization
150
Efficiency

• This assumes that the probability of sending is
  1/N1/500 0.002
• What about ps .2?
• Now P success is going to change

151
Efficiency
So real percent utilization