A view of the Internet as an interconnected
collection of many networks

European
backbone network

High-speed
transatlantic
line



Host A


Commun.
server
M
M
Teleph. line (SLIP
or PPP prot.)
Your home computer
in Lefkosa
US
backbone


METU campus
network
Ethernet LAN
of EMU

Regional
network
Host B
Token
Ring LAN
of Cornell
Univ.
M – modem
R – router
Features:
1. Connectionless datagram service at network layer.
2. This structure is based on IP protocol.
3. IP packets may have different size in different
networks.
4. Speeds of data transmission are different in
different networks.
5. There can be many possible routers between hosts.
1
Addressing in TCP/IP networks
0
Bit order
IP address
netid
Class
0
7 8
31
netid
hostid
7
24
01 2
Class B
hostid
32 bits
0 12
Class A
15 16
10
31
netid
hostid
14
16
012 3
Class C
31
23 24
110
netid
hostid
21
8
0123 4
Class D
1110
31
31
Multicast address
24
Examples:
IP addr.=118.6.20.250 ; 11810=0111 01102
118 6
20
250
class A
32 bits
IP addr.=193.17.40.78 ; 19310=1100 00012
class C
2
Subnet addressing (an example)
134.220.1.10
134.220.1.15
Server DNS
Organization:
netid=134.220
class B
Subnet 134.220.1.0
134.220.1.1
Router
134.220.2.5
!!!!
134.220.2.2
Subnet 134.220.2.0
2 addresses
for the
router
3
Subnetworking in TCP/IP networks
Subnetworking is the division of the normal range of
host id’s into a number of subnetworks, with the
reduced number of hosts on each subnetwork.
An example:
A company wants to install 400 TCP/IP hosts
(computers).
So class B network is necessary.
Max.
Desired
Utilization=
number of number of (400/65536)*100% =
hosts
hosts
0.6%
Class B 65 536
400
Using a subnet mask:
IP address with masking = (original IP address) AND (subnet mask)
In our example:
Binary
Original
IP
address
Subnet
mask
Network
ID
Host ID
Dotted
Decimal
10000001.10000010.01001111.01010101 129.130.79.85
11111111.11111111.11111110.00000000 255.255.254.0
10000001.10000010.01001110.00000000 129.130.78.0
00000000.00000000.00000001.01010101 0.0.1.85
Or 129.130.79.85/23
Number of 1’s in network mask
Network ID = Orig. IP address AND Subnet mask
Host ID = Orig. IP address AND NOT Subnet mask
4
Network byte order
The purpose: to define a standard representation of
data.
Two types of ordering:
The lowest memory address
Byte 0
Big endian
computers
(Motorola 68000, ..)
Most
significant
byte
Byte 3
Little endian
computers
(Intel 286/386/..)
The most sign.
byte first
Host A
3 2 1
1
0
Local representation
of an integer in A
0 1
2 3
Network byte order
(Big endian)
2
3
Integer
2
1
0
Integer
The lowest
memory address
Host B
3 2 1
0
Local representation
of an integer in B
5
Mapping IP addresses to physical address
1. Each host in the TCP/IP internet has a physical
address on some network.
2. Each host . . . has an IP address (or a few IP
addresses).
3. Physical address  IP address.
4. Two hosts on a physical network can communicate
only if they know each other’s physical network
address.
A
Phys. Address PA
IP address IA
B
PB, IB
Network 1
C
Network 2
Two phys. Addresses,
two IP addresses
The goal: higher-level programs must
work only with IP address.
6
IP address and physical address
FCS
Frame
from
data link
layer
Data from
appl. layer
Error control
TCP
segment
IP packet
TCP
header
IP header
Data link
(Frame)
header
Sequence of
bits in phys.
layer
(nonstructured)
IP address
Phys.
addresses
7