Internet Protocol version 4

Claudio Cicconetti <c.cicconetti@iet.unipi.it>

International Master on Communication Networks Engineering 2006/2007

 Table of Contents

IP Addressing
Class-based IP addresses
IP subnets
IP routing
Methods of delivery
IP address exhaustion problem
Private IP addresses
Classless Inter-Domain Routing
IP datagram
IP fragmentation
 Internet Protocol (IP)

IP is a standard protocol with STD number 5

(see http://www.ietf.org/).

IP is the protocol that hides the underlying

physical network by creating a virtual
network view.

It is an unreliable, best-effort, and

connectionless packet delivery protocol.
 Internet Protocol (IP)
Best-effort means that datagrams may:
– be lost;
– arrive out of order;
– even be duplicated.

IP assumes that higher layer protocols (e.g., TCP)

will address these anomalies.

This makes IP a very robust network protocol. In

fact, the US DoD intended to deploy a network
that would still be operational if parts of the
country were destroyed.
 IP Addressing

In any network protocol, such as IP,

addressing is needed to allow any two
hosts to communicate between each other.

IP addresses are represented by a 32-bit

unsigned binary value, which is usually
expressed in a dotted decimal format
(e.g., 193.205.80.1) because the numeric
form (e.g., 3251458049) is hard to read.
 IP Addressing

The binary format of the 32-bit IP address

193.205.80.1 is:

193 205 80 1
11000001 11001101 01010000 00000001

An easier way to remember IP addresses is

by assigning to them a name (e.g.,
www.google.com), which is resolver
through the Domain Name System (DNS).
 IP Addressing

Strictly speaking, an IP address identifies an

interface that is capable of sending and
receiving IP datagrams. One system can
have multiple such interfaces.

Usually, hosts have only one interface (thus,

one IP address), whereas routers have
many interfaces (thus, many IP addresses).
 IP Addressing

IP datagrams (the basic data packets

exchanged between hosts) are transmitted
by a physical network attached to the host.

Each IP datagram contains a source IP

address and a destination IP address.
 Class-based IP addresses

There are five classes of IP addresses.

The prefix is referred to as the network part,
the suffix as the host part.
 Class-based IP addresses

The division of an IP address into two parts

also separates the responsibility for
selecting the complete IP address.

The network number portion of the address is

assigned by the Regional Internet Registries
(RIRs).

The host number portion is assigned by the

authority controlling the network.
 Class-based IP addresses

Not all suffixes are available:

all 0’s: this address refers to the network

itself, e.g., 10.0.0.0 means “the whole
network 10.*”.

all 1’s: this is the directed broadcast for this

network, e.g., 10.255.255.255 means “all
the hosts on network 10.*”.
 Class-based IP addresses

The Class A network 127.0.0.0 is defined as

the loopback network. Usually, the host
part is set to 1, which results in 127.0.0.1
being the loopback IP address of the host.

Addresses from that network are assigned to

interfaces that process data within the local
system. Thus, these loopback interfaces
never access a physical network.
 Class-based IP addresses

Class A addresses

These addresses use 7 bits for the 〈network〉

and 24 bits for the 〈host〉 portion of the IP
address.

This allows for 27-2 (126) networks (both ‘0’

and ‘127’ are reserved values) each with
224 - 2 (16,777,214) hosts, for a total of
2,113,928,964 available addresses.
 Class-based IP addresses

Class B addresses

These addresses use 14 bits for the 〈network〉

and 16 bits for the 〈host〉 portion of the IP
address.

This allows for 214 (16384) networks each with

216 - 2 (65,534) hosts, for a total of
1,073,709,056 available addresses.
 Class-based IP addresses

Class C addresses

These addresses use 21 bits for the 〈network〉

and 8 bits for the 〈host〉 portion of the IP
address.

This allows for 221 (2097152) networks each

with 28 - 2 (254) hosts, for a total of
532,676,608 available addresses.
 Class-based IP addresses

Class D addresses are reserved for

multicasting (a sort of broadcasting, but not
in a limited area, and only to hosts using the
same class D addresses).

Class E addresses are reserved for future use

(likely, they will never be used at all).
 Class-based IP addresses

A Class A address is suitable for networks

with an extremely large number of hosts.

Class C addresses are suitable for networks

with a small number of hosts.

This means that medium-sized networks

(those with more than 254 hosts or where
there is an expectation of more than 254
hosts) must use Class B addresses.
 Class-based IP addresses

However, the number of small- to medium-

sized networks has been growing very
rapidly.

It was feared that if this growth had been

allowed to continue unabated, all of the
available Class B network addresses would
have been used by the mid-1990s. This is
known as the IP address exhaustion
problem.
 IP subnets

An additional problem of the original IP

addressing scheme was that it required a
centralized authority, i.e., the RIR, to assign
network numbers.

This problem was then solved through IP

subnetting, which allows the network
administrator to locally partition her network
into several IP subnets.
 IP subnets

The host number part of the IP address is

subdivided into a second network number
(i.e., subnet) and a host number.

The entire network still appears as one IP

network to the outside world. Thus, a host
within a network that has subnets is aware
of the subnetting structure. A host in a
different network is not. This remote host
still regards the local part (subnet + host) of
the IP address as a host number.
 IP subnets

Any bits in the local portion can be used to

form the subnet. The division is done using
a 32-bit subnet mask, which is usually
written in dotted decimal form.

131 114 9 44
10000011.01110010.00001001.00101100
-----network----- -subnet- --host--

11111111.11111111.11111111.00000000
-----------network mask------------
 IP subnets
The special treatment of all bits zero and all
bits one applies to each of the three parts of
a subnetted IP address just as it does to
both parts of an IP address that has not
been subnetted.

There are two types of subnetting:

– static subnetting, i.e., all subnets have the
same prefix;
– variable subnetting, i.e., subnets may have
different prefixes.
 IP subnets

For example, assume that you have been

assigned an address pool (192.168.1.0/24)
to be subnetted so as to serve a set of 6
LANs with the following requirements:

LAN 1: 2 hosts LAN 2: 2 hosts

LAN 3: 2 hosts LAN 4: 2 hosts
LAN 5: 2 hosts LAN 6: 30 hosts
 IP subnets

An example of variable length subnetting is :

192.168.1.00100000/27 -> 30 hosts (30 needed)
.01000100/30 -> 2 hosts (2 needed)
.01001000/30 -> 2 hosts (2 needed)
.01001100/30 -> 2 hosts (2 needed)
.01010000/30 -> 2 hosts (2 needed)
.01010100/30 -> 2 hosts (2 needed)

With static subnetting you can have:

192.168.1.00100000/27 -> 30 hosts (30 needed)
.01000100/27 -> 30 hosts (2 needed)
.01001000/27 -> 30 hosts (2 needed)
.01001100/27 -> 30 hosts (2 needed)
.01010000/27 -> 30 hosts (2 needed)
.01010100/27 -> 30 hosts (2 needed)
 IP subnets

For example, assume that you have been

assigned the Class C network 193.205.82.0,
and you need to partition the networks so
that the following requirements are satisfied:

LAN 1: 50 hosts
LAN 2: 50 hosts
LAN 3: 50 hosts
LAN 4: 30 hosts
LAN 5: 30 hosts
 IP subnets

This cannot be achieved with static

subnetting, because you would require five
subnets, each allowing up to 64 hosts (i.e.,
255.255.255.192 masks).

However, you can accomplish your task with

variable length subnetting, by partitioning
the network into three 255.255.255.192
subnetworks, and two 255.255.255.224
subnetworks.
 IP routing

Whenever a host has a physical connection to

multiple networks or subnets, it is described
as being multi-homed. Typically, a multi-
homed host has different IP addresses
associated with all its network adapters,
each connected to a different subnet or
network.

Such a multi-homed host is usually employed

as router.
 IP routing

A router forwards incoming IP datagrams towards

a destination through a physical interface. Its
decisions are based on the datagrams’ destination
IP addresses, according to its routing table.
 IP routing

Four kinds of destinations:

– hosts or networks that are directly attached to
one of the physical networks to which the router
is attached (e.g., 192.168.1.0/255.255.255.0);
– hosts or networks for which the router has been
given explicit definitions (e.g.,
192.168.3.0/255.255.255.0);
– hosts or networks for which the router has
received an ICMP redirect message;
– a default for all other destinations (e.g., last
entry of the routing table above).
 IP routing

There are many Interior Gateway Protocols

(IGPs), such as:
– Open Shortest Path First (OSPF);
– Routing Information Protocol (RIP);
– Interior Gateway Routing Protocol (IGRP);
– Enhanced IGRP (EIGRP).

On the other hand, the most common Exterior

Gateway Protocol (EGP) is the Border
Gateway Protocol version 4 (BGP4).
 IP routing
If the destination host is attached to the same
physical network as the source host, IP
datagrams can be directly exchanged. This
is done by encapsulating the IP datagram in
the physical network frame. This is called
direct routing.

Indirect routing occurs when the destination

host is not connected to a network directly
attached to the source host. The only way to
reach the destination is via one or more IP
routers.
IP routing

lan0 192.168.1.64/26

Host A
192.168.1.66

Router R Host B
lan0: 192.168.1.65 lan1 192.168.1.130
lan1: 192.168.1.129 192.168.1.128/26

Host C
192.168.1.131
 IP routing

Host B communicates directly with Host B.

Host A communicates with Host B and Host C
indirectly via Route R.

Router R routing table:

Destination Netmask Interface Next-hop
192.168.1.64 255.255.255.192 lan0
192.168.1.128 255.255.255.192 lan1

Host A routing table:

Destination Netmask Interface Next-hop
192.168.1.64 255.255.255.192 lan0
192.168.1.128 255.255.255.192 lan0 192.168.1.65
IP routing
 IP routing

Even though Host A and Host B are

connected to the same physical network,
they cannot communicate directly, because
they belong to different logical subnetwork.

Router R routing table:

Destination Netmask Interface Next-hop
192.168.1.64 255.255.255.192 lan0
192.168.1.128 255.255.255.192 lan0

Host A routing table:

Destination Netmask Interface Next-hop
192.168.1.64 255.255.255.192 lan0
192.168.1.128 255.255.255.192 lan0 192.168.1.65
 IP routing

Without subnetting, IP uses the following

algorithm to route datagrams:

– Is the destination IP network address equal to

my IP network address?
– If so, send the IP datagram on local network.
– Otherwise send the IP datagram to the router
corresponding to the destination IP network
address.
 IP routing

With subnetting, IP uses the following

algorithm to route datagrams:

- Is (destination IP address & subnet mask)

equal to (my IP address & subnet mask)?
- If so, send the IP datagram on local network.
- Otherwise send IP datagram to router
corresponding to the destination IP
(sub)network address.
 Methods of delivery
The majority of IP addresses refer to a single
recipient, this is called a unicast address.

Unicast connections specify a one-to-one

relationship between a single source and a
single destination.

Additionally, there are three special types of

IP addresses used for addressing multiple
recipients: broadcast addresses, multicast
addresses and anycast addresses.
 Methods of delivery

Broadcast addresses are never valid as a

source address. They must specify the
destination address.

Different types of broadcast addresses:

– limited broadcast address;
– network-directed broadcast address;
– subnet-directed broadcast address;
– all-subnets-directed broadcast address.
 Methods of delivery

Limited broadcast address

Address 255.255.255.255 (all bits 1 in all

parts of the IP address), meaning all hosts
on the local subnet.

Routers do not forward this packet.

 Methods of delivery

Network-directed broadcast address

This is used in an unsubnetted environment,

with the host part being all 1’s, e.g.,
128.2.255.255, meaning all hosts on a
network.

Routers should forward these broadcast

messages.
 Methods of delivery

With multicasting, any host in the Internet can

associate to a multicast group, which is
identified by a Class D multicast address.

Packets sent to a multicast address are

forwarded only to the members of the
corresponding host group, which allows for
one-to-many connections.

E.g., used for audio/video brodcasting.

 Methods of delivery

With anycasting (not available in IPv4), hosts

can be grouped into anycast pools, which
are considered by the network to be
interchangeable.

When a remote host sends an IP datagram to

an anycast address, that datagram is
delivered to any of the hosts in the pool.

E.g., used for web proxies.

 The IP address exhaustion problem
The number of networks on the Internet has
been approximately doubling annually for a
number of years.

Nearly all of the new networks assigned in the

late 1980s were Class B, and in 1990 it
became apparent that if this trend
continued, the last Class B network number
would be assigned during 1994.

On the other hand, Class C networks were

hardly being used.
 The IP address exhaustion problem
To mitigate the exhaustion problem, Class B
networks have been only assigned to
organizations that:
– have a subnetting plan that documents more than 32
subnets within its organizational network;
– have more than 4096 hosts.

Otherwise, a consecutively numbered block of Class

C network numbers are assigned instead.

Any requirements for a Class A network would be

handled on an individual case basis.
 Private IP addresses

Another approach to conserve the IP address

space is to use private IP addresses.

Private IP addresses do not need to be

unique within the Internet.

However, hosts with private IP addresses

cannot communicate with hosts outside their
local networks.
 Private IP addresses
Three ranges of addresses have been
reserved for this purpose:

– 10.0.0.0, as a single Class A network;

– 172.16.0.0 through 172.31.0.0, as 16
contiguous Class B networks;
– 192.168.0.0 through 192.168.255.0, as 256
contiguous Class C networks.

These addresses can be used without

requesting authorization from the RIR.
 Classless Inter-Domain Routing

While subnetting and careful assignment of IP

addresses mitigated the IP address
exhaustion problem, a new issue arised: the
routing table explosion problem.

Assume an organization has 4000 hosts.

Then, it cannot be assigned a Class B
network, and is allocate 16 Class C
networks instead. This requires 16 entries in
the routing table of every router in the world
for the same organization.
 Classless Inter-Domain Routing

The solution to this problem is called

Classless Inter-Domain Routing (CIDR).

CIDR does not route according to the class of

the network number (hence the term
classless).

On the other hand, it is based solely on the

high order bits of the IP address.
 Classless Inter-Domain Routing

Each CIDR routing table entry contains a 32-bit

IP address and a 32-bit network mask, which
together give the length and value of the IP
prefix:
<IP_address> <network_mask>

For example, to address a block of eight Class

C addresses with one single routing table
entry, the following representation suffices:
<192.32.136.0> <255.255.248.0>
 Classless Inter-Domain Routing

This process of combining multiple networks

into a single entry is referred to as
supernetting.

Routing is based on network masks that are

shorter than the natural network mask of an
IP address.

This contrasts subnetting where the subnet

masks are longer than the natural network
mask.
 Classless Inter-Domain Routing

Supernetting example.
11000000 00100000 10001000 00000000 = 192.32.136.0 (class C address)
11111111 11111111 11111--- -------- 255.255.248.0 (network mask)
======== ======== ======== ======== logical_AND
11000000 00100000 10001--- -------- = 192.32.136 (IP prefix)

11000000 00100000 10001111 00000000 = 192.32.143.0 (class C address)

11111111 11111111 11111--- -------- 255.255.248.0 (network mask)
======== ======== ======== ======== logical_AND
11000000 00100000 10001--- -------- = 192.32.136 (same IP prefix)
 IP datagram

The unit of transfer in an IP network is called

an IP datagram. It consists of an IP header
and data relevant to higher level protocols.
IP datagram
 IP datagram

VER is the field that contains the IP protocol

version. The current version is 4. 5 is an
experimental version. 6 is the version for
IPv6.

HLEN is the length of the IP header in

multiples of 32 bits, without the data field.
The minimum value for a correct header is 5
(i.e., 20 bytes), the maximum value is 15
(i.e., 60 bytes).
 IP datagram

Service Type The service type is an

indication of the quality of service requested
for this IP datagram. It contains the following
information.

Precedence specifies the nature/priority:

000: Routine 001: Priority
010: Immediate 011: Flash
100: Flash override 101: Critical
110: Internetwork control 111: Network control
 IP datagram

TOS specifies the type of service value:

1000: Minimize delay

0100: Maximize throughput
0010: Maximize reliability
0001: Minimize monetary cost
0000: Normal service

The last bit is reserved for future use.

 IP datagram
Total Length specifies the total length of the
datagram, header and data, in octets.

Identification is a unique number assigned by the

sender used with fragmentation.

Flags contains control flags:

– the first bit is reserved and must be zero;
– the 2nd bit is DF (Do not Fragment), 0 means allow
fragmentation;
– the third is MF (More Fragments), 0 means that this is
the last fragment.
 IP datagram

Fragment Offset is used to reassemble the

full datagram. The value in this field
contains the number of 64-bit segments
(header bytes are not counted) contained in
earlier fragments. If this is the first (or only)
fragment, this field contains a value of zero.

TTL (Time to Live) specifies the time (in

seconds) the datagram is allowed to travel.
In practice, this is used as a hop counter to
detect routing loops.
 IP datagram
Protocol Number indicates the higher level
protocol to which IP should deliver the data in this
datagram. E.g., ICMP = 1; TCP = 6; UDP = 17.

Header Checksum is a checksum for the

information contained in the header. If the header
checksum does not match the contents, the
datagram is discarded.

Source/Destination IP Addresses are the 32-bit

source/destination IP addresses.
 IP datagram
IP Options is a variable-length field (there may be
zero or more options) used for control or
debugging and measurement. For instance:
– the loose source routing option provides a means for
the source of an IP datagram to supply explicit routing
information;
– the timestamp option tells the routers along the route
to put timestamps in the option data.

Padding is used to ensure that the IP header ends

on a 32 bit boundary. The padding is zero.
 IP fragmentation

IP provides fragmentation/reassembly of
datagrams. The maximum length of an IP
datagram is 65,535 octets.

When an IP datagram travels from one host to

another, it may pass through different
physical networks. Each physical network
has a maximum frame size, called
maximum transmission unit (MTU), which
limits the datagram length.
 IP fragmentation
A fragment is treated as a normal IP datagram while
being transported to their destination. Thus,
fragments of a datagram each have a header.

If one of the fragments gets lost, the complete

datagram is considered lost.

It is possible that fragments of the same IP datagram

reach the destination host via multiple routes.

Finally, Since they may pass through networks with a

smaller MTU than the sender’s one, they are
subject to further fragmentation.
 IP fragmentation

H IP datagram

1
nt #

2
gme

#
1

ent
#
nt

Fra
me

m
H
g
Fra

Frag
Fr

H
H

ag
m
en

H
t#
1
H
Fr t #2
ag en
m m
en ag
t #1

t #2 Fr
t #2
men

H
men
Frag

Frag
H

H IP datagram

H Fragment #1
H Fragment #2
IP fragmentation
 IP fragmentation
Fragmentation process:
– The DF flag bit is checked to see if
fragmentation is allowed. If the bit is set, the
datagram will be discarded and an ICMP
error returned to the originator.
– Based on the MTU value, the data field is
split into two or more parts. All newly created
data portions must have a length that is a
multiple of 8 octets, with the exception of the
last data portion.
– Each data portion is placed in an IP
datagram.
IP fragmentation
LEN = 1500
OFFSET = 0
DF = 0, MF = 1

DATA
(1500 bytes)
LEN = 3499
OFFSET = 0
DF = 0, MF = 0
LEN = 1500
OFFSET = 1500
DF = 0, MF = 1

MTU = 1500
DATA DATA
(3500 bytes) (1500 bytes)

LEN = 499
OFFSET = 3000
DF = 0, MF = 0

DATA
(499 bytes)
 IP fragmentation

Modification to the headers of fragments:

– the MF flag is set in all fragments except the
last;
– the fragment offset field is updated;
– if options were included in the original
datagram, they may be copied to all
fragment datagrams or only the first
datagram (depends on the option);
– the header length field is set;
– the total length field is set;
– the header checksum is re-calculated.
 IP fragmentation

At the destination host, data are reassembled

into the original datagram.

The identification field set by the sending host

is used together with the source and
destination IP addresses in the datagram.
Fragmentation does not alter this field.

In order to reassemble the fragments, the

receiving host allocates a storage buffer
when the first fragment arrives.
 IP fragmentation
The host also starts a timer. If the timer is
exceeded and fragments remain
outstanding, the datagram is discarded (in
linux this value, in seconds, is stored into
/proc/sys/net/ipv4/ipfrag_time).

When subsequent fragments of the datagram

arrive, data are copied into the buffer
storage at the location indicated by the
fragment offset field. When all fragments
have arrived, the original unfragmented
datagram is restored and passed to upper
layers, if needed.