Internet Protocol (IP)

The Network Layer

EE450: Introduction to Computer Networks

Professor A. Zahid

EE450, USC, Zahid

Network Layer

On sending side, it encapsulates

TCP Segments (or UDP
Datagrams) into Packets

On receiving side, it delivers TCP

Segments (or UDP Datagrams) to
corresponding Transport layer
Network layer protocols are
implemented in every host and
router across the Network
Router examines header fields in
all IP packets passing through it
and forward according to routing
tables
EE450, USC, Zahid

Packet Switched Network (Review)

No call setup at Network layer

Routers: no state about end-to-end connections
Packets may arrive out-of-order

Packets forwarded using destination IP address

EE450, USC, Zahid

Routing vs. Forwarding

routing algorithm determines
end to-end route through network
forwarding table determines
local forwarding (to next hop)
at this router

EE450, USC, Zahid

Router Architecture (EE555)

two key router functions:

Run Routing algorithms/protocols (RIP, OSPF, BGP)

forwarding packets from incoming to outgoing link

EE450, USC, Zahid

Network Layer at
Source/Router/Destination

EE450, USC, Zahid

The IP Protocol

EE450, USC, Zahid

The Internet Protocol

IP is a connection-less, unreliable network layer

protocol designed to be used in a connection-less
packet switched network such as the Internet.
IP provides best effort services in the sense
There is no guarantee of delivery of error-free
packets
There is no guarantee of ordered delivery of
packets
There is no guarantee of delivery of packets, i.e.
some packets may be lost, some packets may be
duplicated
IP relies on upper layer transport protocols (TCP)
to take care of these problems
EE450, USC, Zahid

IP Packet Format

EE450, USC, Zahid

The Protocol Field

The protocol field (8-bits) defines the protocol

that is using the services of IP. It defines the
final destination protocol the packet should be
delivered to. This is important since several
protocols could be multiplexed over IP
ICMP : 1, IGMP: 2, TCP: 6, EGP: 8, UDP: 17,
IPv6: 41, OSPF: 89, etc...

EE450, USC, Zahid

10

Fragmentation

IP packet may travel over different networks

(LANs and WANs)
A router de-capsulate an IP packet from the frame
it receives, process it, and encapsulate it in
another frame
Frame size and format varies depend on the data
link protocol used by the physical network through
which the frame is traveling
MTU (Maximum Transmission Unit) is the maximum
size of the data field (payload) in the frame
If Packet size > MTU, Need for Fragmentation
EE450, USC, Zahid

11

MTU: Maximum Transfer Unit

Protocol
MTU (Octets)
Ethernet
1500
Token Ring (4 Mbps)
4464
Token Ring (16 Mbps)
17914
FDDI
4352
X.25
576
PPP
296
EE450, USC, Zahid

12

Fragmentation (Continued)

Each fragment has its own header (most of fields

are copied, some will change, including the total
length, the Flags and the fragmentation offset
fields)

A fragmented datagram may itself be fragmented

if it encounters a network with smaller MTU
A packet can be fragmented by a source host or by
any router in the path. Re-assembly of the packet
must be done at the destination host because
those fragments become independent packets and
may travel different routes
EE450, USC, Zahid

13

Fragmentation and Reassembly

Network links have MTU

([Link] size) largest possible link-level
frame.

fragmentation
of one large
packet into 3
smaller packets

Different link types,

Different MTUs

large IP datagram
fragmented within net

reassembly

One packet becomes

several packets
(fragments)
Reassembled only at final
destination
IP header bits used to
identify, order related
fragments

EE450, USC, Zahid

14

Fields related to Fragmentation

Identification (16 bits): All fragments of a packet

has the same ID number which is the same as that
of the original packet. The R/x knows that all
fragments having the same ID should be assembled
into one packet

Flags (3 bits):

D M

D: Do not Fragment
M: More Fragment

Fragmentation Offset (13 bits): Relative position

of the fragment to the whole packet measured in
units of 8 Bytes
EE450, USC, Zahid

15

Fragmentation Example (cont.)

EE450, USC, Zahid

16

IP v4. Addressing

The Internet is made of combination of LANs

and WANs connected via routers
A host needs to be able to communicate with
another host without worrying about which
physical network must be passed through
Hosts must therefore be identified uniquely
and globally at the network layer

For efficient and optimum routing, routers

must also be identified uniquely and globally at
the network layer
EE450, USC, Zahid

17

IP v4. Addressing (Cont.)

IPv4 address is a 32-bit address, implemented

in software, is used to uniquely and globally
identify a host or a router on the Internet

A device can have more than one IP address if

it is connected to more than one network
(multi-homed)
An IP address have two parts, the netid and
the hostid. They have variable lengths
depending on the class of the address
All devices on the same network have the same
netid
EE450, USC, Zahid

18

IP Addresses (Example)

IP address: 32-bit identifier

for host, router interface
Interface: connection between
host/router and physical link
routers typically have
multiple interfaces
host typically has one or
two interfaces (e.g., wired
Ethernet, wireless 802.11)
IP addresses associated with
each interface
EE450, USC, Zahid

19

Classful IP Addressing

EE450, USC, Zahid

20

Number of Blocks/Size of Blocks in

Classful IP Addressing

In classful addressing, a large part of the

available addresses are wasted. It is hence
being replaced by Classless IP Addressing
EE450, USC, Zahid

21

Special IP Addresses
Special Address

Netid

Hostid

Source/Destination

Network Address

Specific

All 0's

None

Direct Broadcast
Address
Limited Broadcast
Address
This host on this
network
Specific host on
this network
Loopback address

Specific

All 1's

Destination

All 1's

All 1's

Destination

All 0's

All 0's

Source

All 0's

Specific

Destination

127

Any

Destination

EE450, USC, Zahid

22

Private IP Addressing

One of the problems in IP network address allocation

is that many hosts do not require access to hosts in
other networks
Assigning Globally unique public IP
addresses for such hosts may be wasteful
IETF proposed the use of Private IP addresses that
are not advertised outside the private network.
Private IP Addresses
are non-routable

EE450, USC, Zahid

23

NAT: Network Address Translation

NAT is a protocol that maintains a translation

table for mapping an internal private IP address to
a globally unique IP addresses and vice versa
May be static (one-to-one) or dynamic (from a pool
of global IP addresses)
Implemented in the border (access) router
separating the private network and the public
Internet or as a stand-alone multi-homed Server
NAT is a special type of Proxy Server. It was
introduced with Windows 2000
EE450, USC, Zahid

24

NAT Implementation

EE450, USC, Zahid

25

NAT Address Translation

EE450, USC, Zahid

26

NAT & PAT

2: NAT router
changes packet
source addr from
[Link], 3345 to
[Link], 5001,
updates table
2

NAT translation table

WAN side addr
LAN side addr

1: host [Link]
sends datagram to
[Link], 80

[Link], 5001 [Link], 3345

S: [Link], 3345
D: [Link], 80

S: [Link], 5001
D: [Link], 80

[Link]

[Link]

[Link]

[Link]
S: [Link], 80
D: [Link], 5001

3: Reply arrives
dest. address:
[Link], 5001

S: [Link], 80
D: [Link], 3345

[Link]

4: NAT router
changes datagram
dest addr from
[Link], 5001 to [Link], 3345
EE450, USC, Zahid

27

Motivation for NAT

Local network uses just one IP address as far as

outside world is concerned:
Range of addresses not needed from ISP: just one
IP address for all devices

can change addresses of devices in local network

without notifying outside world
can change ISP without changing addresses of
devices in local network

devices inside local net not explicitly addressable

or visible by outside world (a security plus)
EE450, USC, Zahid

28

Subnetting

Subnetting is the process of creating multiple

segments within a single IP network address
space
From the perspective of a node outside the
network, all nodes on any of the subnetworks
appear to be on the original single network
Internet routing tables are not affected by
subnetting, I.e. routing tables need not be
overloaded with information about routes to all
internal subnets, just information to the
access router/gateway
EE450, USC, Zahid

29

Subnetting (Continued)

Classes A, B and C in IP addressing are designed

with two levels of hierarchy (netid & hostid)
Problem: An organization with a class B address
can not have more than network and all 216 hosts
are attached to that network
A nightmare in
managing this network, single broadcast domain,
security issues, etc

Subnetting create another level of hierarchy

(netid, subnetid and hostid). Delivery of IP
packets involves three steps; delivery to the site
router, delivery to the subnet router, delivery to
the host
EE450, USC, Zahid
30

Example of Subnetting
Network consisting of 3 Subnets
[Link]
[Link]
[Link]
[Link]

[Link]
[Link]

[Link]
[Link]

subnet
[Link]

[Link]

EE450, USC, Zahid

31

Example of Subnetting (How many?)

Answer:

EE450, USC, Zahid

32

Subnet Masking

Subnetting is achieved by stealing some bits

from the hostid field to represent the subnet
portion of the address
Those bits that are used for the subnetid are
identified through the use of a subnet mask
Masking is the process of extracting the address
of the physical network (if subnetting is not used)
or the subnet address (if subnetting is used) from
an IP address

A subnet mask is a 32-bit pattern having a 1 in

every netid and subnetid locations and a 0 in
every hostid locationEE450, USC, Zahid

33

Subnet Masking (Continued)

Subnet masking is performed (both at the host

and at the router) by applying bit-wise-AND
operation between the IP address and the
subnet mask
Example 1: Class B network without subnetting

[Link]
10001101.00001110.00000010.00010101
[Link] 11111111.11111111.00000000.00000000
Bit-wise and 10001101.00001110.00000000.00000000
IP address
[Link]

Mask
[Link]

EE450, USC, Zahid

Network address
[Link]

34

Example of Subnetting
Mask: [Link]
Subnet #: [Link]

[Link]

[Link]

Mask: [Link]
Subnet #: [Link]

[Link]

[Link]

Mask: [Link]
Subnet #: [Link]

[Link]

[Link]
[Link]

[Link]

[Link]

[Link]

[Link]

Internet

Internal routers
use subnet IDs
to route
[Link]

External routers do not

know about subnet IDs,
only net IDs

EE450, USC, Zahid

35

Subnet Router Routing Table

Partial Table in Router [Link]
Subnet Number
[Link]

Subnet Mask
[Link]

Next Hop
Left Router

[Link] [Link]

Left Interface

[Link]

Right Interface

[Link]

for each table entry do

if (DestAddr & Subnet Mask) = SubnetNumber
if NextHop is an Interface
deliver packet directly to DestAddr
else
deliver packet to Router
EE450, USC, Zahid

36

Classless IP Addressing

To overcome IP Address depletion and allow

more organizations ability to access the
Internet
Get rid of classes and assign addresses in
Blocks. The size of the block (The # of IP
addresses in the block) depends on the
organization demand
There are some restrictions on these
blocks
EE450, USC, Zahid

37

Restrictions of Address Blocks

The number of addresses in a block

must be a power of 2 (1, 2, 4, 8, etc)
The addresses in the block are
contiguous, i.e. one after the other
The first address in the block (in
decimal) must be divisible by the size
of the block
EE450, USC, Zahid

38

CIDR Notation
CIDR: Classless Inter Domain Routing

Network and Subnet portions of address

of arbitrary length
address format: a.b.c.d/x, where x is #
bits in network and subnet portion of
address
[Link]/23

Subnet part
subnet mask: /23 or
[Link]

EE450, USC, Zahid

Host part
39

Example: A block of Size 16

In IPv4 addressing, a block of addresses can be defined as

w.x.y.z /n in which w.x.y.z defines one of the addresses and the /n
defines the mask. The number of address in the block is 232-n
The first address in the block can be found by setting the
rightmost 32 n bits to 0s. Hence the first address in the block
represents the Network/Subnet address of the organization
The last address in the block can be found by setting the rightmost
32 n bits to 1s. Hence the last address in the block represents
the broadcast address in the organization network
EE450, USC, Zahid

40

Example of Classless IP Addressing

An ISP gets a block of IP addresses. The Block is

[Link]/20 . The size of the block is 212 = 4096.
The ISP has 8 customers (organizations), each
requiring a block of size 512

EE450, USC, Zahid

41

Example of Classless IP Addressing

Organization was assigned a block [Link]/26

(Size of block is 64).Organization has three
departments. Three subnets of sizes 32, 16 and 16

EE450, USC, Zahid

42

Hierarchical Addressing
Hierarchical addressing allows efficient
advertisement of routing information
Organization 0

[Link]/23
Organization 1

Send me anything
with addresses
beginning
[Link]/20

[Link]/23
Organization 2

[Link]/23
Organization 7

.
.
.

.
.
.

Fly-By-Night-ISP
Internet

[Link]/23
ISPs-R-Us

EE450, USC, Zahid

Send me anything
with addresses
beginning
[Link]/16

43