Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                

Ipv4 Addressing and Subnetting

Download as docx, pdf, or txt
Download as docx, pdf, or txt
You are on page 1of 24

IPv4 Addressing and Subnetting v1.

41 – Aaron Balchunas 1

- IPv4 Addressing and Subnetting -


Hardware Addressing
A hardware address is used to uniquely identify a host within a local
network. Hardware addressing is a function of the Data-Link layer of the
OSI model (Layer-2).
Ethernet utilizes the 48-bit MAC address as its hardware address. The
MAC address is often hardcoded on physical network interfaces, though
some interfaces support changing the MAC address using special utilities. In
virtualization environments, dynamically assigning MAC addresses is very
common.
A MAC address is most often represented in hexadecimal, using one of two
accepted formats:
00:43:AB:F2:32:13
0043.ABF2.3213
The first six hexadecimal digits of a MAC address identify the
manufacturer of the physical network interface. This is referred to as the
OUI (Organizational Unique Identifier). The last six digits uniquely
identify the host itself, and are referred to as the host ID.
The MAC address has one shortcoming – it contains no hierarchy. MAC
addresses provide no mechanism to create boundaries between networks.
There is no method to distinguish one network from another.
This lack of hierarchy poses significant difficulties to network scalability. If
only Layer-2 hardware addressing existed, all hosts would technically exist
on the same network. Internetworks like the Internet could not exist, as it
would be impossible to separate my network from your network.
Imagine if the entire Internet existed purely as a single Layer-2 switched
network. Switches, as a rule, will forward a broadcast out every port. With
billions of hosts on the Internet, the resulting broadcast storms would be
devastating. The Internet would simply collapse.
The scalability limitations of Layer-2 hardware addresses are mitigated
using logical addresses, covered in great detail in this guide.

***
All original material copyright © 2013 by Aaron Balchunas (aaron@routeralley.com),
unless otherwise noted. All other material copyright © of their respective owners.
consent of the owner of the above copyright. Updated material may be found at
http://www.routeralley.com.
Logical Addressing
Logical addressing is a function of the Network layer of the OSI Model
(Layer-3), and provides a hierarchical structure to separate networks.
Logical addresses are never hardcoded on physical network interfaces, and
can be dynamically assigned and changed freely.
A logical address contains two components:
 Network ID – identifies which network a host belongs to.
 Host ID – uniquely identifies the host on that network.
Examples of logical addressing protocols include Internetwork Packet
Exchange (IPX) and Internet Protocol (IP). IPX was predominantly used
on Novell networks, but is now almost entirely deprecated. IP is the most
widely-used logical address, and is the backbone protocol of the Internet.

Internet Protocol (IP)


In the 1970’s, the Department of Defense developed the Transmission
Control Protocol (TCP), to provide both Network and Transport layer
functions. When this proved to be an inflexible solution, those functions
were separated - with the Internet Protocol (IP) providing Network layer
services, and TCP providing Transport layer services.
Together, TCP and IP provide the core functionality for the TCP/IP or
Internet protocol suite.
IP provides two fundamental Network layer services:
 Logical addressing – provides a unique address that identifies both
the host, and the network that host exists on.
 Routing – determines the best path to a particular destination
network, and then routes data accordingly.
IP was originally defined in RFC 760, and has been revised several times.
IP Version 4 (IPv4) was the first version to experience widespread
deployment, and is defined in RFC 791. IPv4 will be the focus of this
guide.
IPv4 employs a 32-bit address, which limits the number of possible
addresses to 4,294,967,296. IPv4 will eventually be replaced by IP Version 6
(IPv6), due to a shortage of available IPv4 addresses. IPv6 is covered in
great detail in another guide.
IPv4 Addressing
A core function of IP is to provide logical addressing for hosts. An IP
address provides a hierarchical structure to both uniquely identify a host,
and what network that host exists on.
An IP address is most often represented in decimal, in the following format:
158.80.164.3
An IP address is comprised of four octets, separated by periods:
First Octet Second Octet Third Octet Fourth Octet
158 80 164 3

Each octet is an 8-bit number, resulting in a 32-bit IP address. The smallest


possible value of an octet is 0, or 00000000 in binary. The largest possible
value of an octet is 255, or 11111111 in binary.
The above IP address represented in binary would look as follows:
First Octet Second Octet Third Octet Fourth Octet
10011110 01010000 10100100 00000011

Decimal to Binary Conversion


The simplest method of converting between decimal and binary is to
remember the following table:
128 64 32 16 8 4 2 1
To convert a decimal number of 172 to binary, start with the leftmost
column. Since 172 is greater than 128, that binary bit will be set to 1. Next,
add the value of the next column (128 + 64 = 192). Since 172 is less than
192, that binary bit will be set to 0.
Again, add the value of the next column (128 + 32 = 160). Since 172 is
greater than 160, that binary bit will be set to 1. Continue this process until
the columns with binary bits set to 1 add up to 172:
Decimal 128 64 32 16 8 4 2 1
Binary 1 0 1 0 1 1 0 0
Binary to Decimal Conversion
Converting from binary back to decimal is even simpler. Apply the binary
number to the conversion table, and then add up any columns with binary
bits set to 1.
For example, consider the binary number of 11110001:
Decimal 128 64 32 16 8 4 2 1
Binary 1 1 1 1 0 0 0 1

By adding 128 + 64 + 32 + 16+ 1, it can be determined that 11110001


equals 241.

The Subnet Mask


Part of an IP address identifies the network. The other part of the address
identifies the host. A subnet mask is required to provide this distinction:
158.80.164.3 255.255.0.0
The above IP address has a subnet mask of 255.255.0.0. The subnet mask
follows two rules:
 If a binary bit is set to a 1 (or on) in a subnet mask, the corresponding
bit in the address identifies the network.
 If a binary bit is set to a 0 (or off) in a subnet mask, the corresponding
bit in the address identifies the host.
Looking at the above address and subnet mask in binary:
IP Address: 10011110.01010000.10100100.00000011
Subnet Mask: 11111111.11111111.00000000.00000000
The first 16 bits of the subnet mask are set to 1. Thus, the first 16 bits of the
address (158.80) identify the network. The last 16 bits of the subnet mask are
set to 0. Thus, the last 16 bits of the address (164.3) identify the unique host
on that network.
The network portion of the subnet mask must be contiguous. For example, a
subnet mask of 255.0.0.255 is not valid.
The Subnet Mask (continued)
Hosts on the same logical network will have identical network addresses,
and can communicate freely. For example, the following two hosts are on
the same network:
Host A: 158.80.164.100 255.255.0.0
Host B: 158.80.164.101 255.255.0.0
Both share the same network address (158.80), which is determined by the
255.255.0.0 subnet mask. Hosts that are on different networks cannot
communicate without an intermediating device. For example:
Host A: 158.80.164.100 255.255.0.0
Host B: 158.85.164.101 255.255.0.0
The subnet mask has remained the same, but the network addresses are now
different (158.80 and 158.85 respectively). Thus, the two hosts are not on
the same network, and cannot communicate without a router between them.
Routing is the process of forwarding packets from one network to another.
Consider the following, trickier example:
Host A: 158.80.1.1 255.248.0.0
Host B: 158.79.1.1 255.248.0.0
The specified subnet mask is now 255.248.0.0, which doesn’t fall cleanly on
an octet boundary. To determine if these hosts are on separate networks, first
convert everything to binary:
Host A Address: 10011110.01010000.00000001.00000001
Host B Address: 10011110.01001111.00000001.00000001
Subnet Mask: 11111111.11111000.00000000.00000000
Remember, the 1 (or on) bits in the subnet mask identify the network portion
of the address. In this example, the first 13 bits (the 8 bits of the first octet,
and the first 5 bits of the second octet) identify the network. Looking at only
the first 13 bits of each address:
Host A Address: 10011110.01010
Host B Address: 10011110.01001
Clearly, the network addresses are not identical. Thus, these two hosts are on
separate networks, and require a router to communicate.
IP Address Classes
The IPv4 address space has been structured into several classes. The value
of the first octet of an address determines the class of the network:
Class First Octet Range Default Subnet Mask
Class A 1 - 127 255.0.0.0
Class B 128 - 191 255.255.0.0
Class C 192 - 223 255.255.255.0
Class D 224 - 239 -
Class A networks range from 1 to 127. The default subnet mask is 255.0.0.0.
Thus, by default, the first octet defines the network, and the last three octets
define the host. This results in a maximum of 127 Class A networks, with
16,777,214 hosts per network!
Example of a Class A address:
Address: 64.32.254.100
Subnet Mask: 255.0.0.0
Class B networks range from 128 to 191. The default subnet mask is
255.255.0.0. Thus, by default, the first two octets define the network, and the
last two octets define the host. This results in a maximum of 16,384 Class B
networks, with 65,534 hosts per network.
Example of a Class B address:
Address: 152.41.12.195
Subnet Mask: 255.255.0.0
Class C networks range from 192 to 223. The default subnet mask is
255.255.255.0. Thus, by default, the first three octets define the network,
and the last octet defines the host. This results in a maximum of 2,097,152
Class C networks, with 254 hosts per network.
Example of a Class C address:
Address: 207.79.233.6
Subnet Mask: 255.255.255.0
Class D networks are reserved for multicast traffic. Class D addresses do
not use a subnet mask.
CIDR (Classless Inter-Domain Routing)
Classless Inter-Domain Routing (CIDR) is a simplified method of
representing a subnet mask. CIDR identifies the number of binary bits set to
a 1 (or on) in a subnet mask, preceded by a slash.
For example, a subnet mask of 255.255.255.240 would be represented as
follows in binary:
11111111.11111111.11111111.11110000
The first 28 bits of the above subnet mask are set to 1. The CIDR notation
for this subnet mask would thus be /28.
The CIDR mask is often appended to the IP address. For example, an IP
address of 192.168.1.1 and a subnet mask of 255.255.255.0 would be
represented as follows using CIDR notation:
192.168.1.1 /24

Address Classes vs. Subnet Mask


Remember the following three rules:
 The first octet on an address dictates the class of that address.
 The subnet mask determines what part of an address identifies the
network, and what part identifies the host.
 Each class has a default subnet mask. A network using its default
subnet mask is referred to as a classful network.
For example, 10.1.1.1 is a Class A address, and its default subnet mask is
255.0.0.0 (/8 in CIDR).
It is entirely possible to use subnet masks other than the default. For
example, a Class B subnet mask can be applied to a Class A address:
10.1.1.1 /16
However, this does not change the class of the above address. It remains a
Class A address, which has been subnetted using a Class B mask.
Remember, the only thing that determines the class of an IP address is the
first octet of that address. Likewise, the subnet mask is the only thing that
determines what part of an address identifies the network, and what part
identifies the host.
Subnet and Broadcast Addresses
On each IP network, two host addresses are reserved for special use:
 The subnet (or network) address
 The broadcast address
Neither of these addresses can be assigned to an actual host.
The subnet address is used to identify the network itself. A routing table
contains a list of known networks, and each network is identified by its
subnet address. Subnet addresses contain all 0 bits in the host portion of
the address.
For example, 192.168.1.0/24 is a subnet address. This can be determined by
looking at the address and subnet mask in binary:
IP Address: 11000000.10101000.00000001.00000000
Subnet Mask: 11111111.11111111.11111111.00000000

Note that all host bits in the address are set to 0.


The broadcast address identifies all hosts on a particular network. A packet
sent to the broadcast address will be received and processed by every host on
that network. Broadcast addresses contain all 1 bits in the host portion of
the address.
For example, 192.168.1.255/24 is a broadcast address. Note that all host bits
are set to 1:
IP Address: 11000000.10101000.00000001.11111111
Subnet Mask: 11111111.11111111.11111111.00000000
Broadcasts are one of three types of IP packets:
 Unicasts are packets sent from one host to one other host
 Multicasts are packets sent from one host to a group of hosts
 Broadcasts are packets sent from one host to all other hosts on the
local network
A router, by default, will never forward a multicast or broadcast packet
from one interface to another.
A switch, by default, will forward a multicast or broadcast packet out every
port, except for the port that originated the multicast or broadcast.
Subnetting
Subnetting is the process of creating new networks (or subnets) by stealing
bits from the host portion of a subnet mask. There is one caveat: stealing bits
from hosts creates more networks but fewer hosts per network.
Consider the following Class C network:
192.168.254.0
The default subnet mask for this network is 255.255.255.0. This single
network can be segmented, or subnetted, into multiple networks. For
example, assume a minimum of 10 new networks are required. Resolving
this is possible using the following magical formula:

2n
The exponent ‘n’ identifies the number of bits to steal from the host portion
of the subnet mask. The default Class C mask (255.255.255.0) looks as
follows in binary:
11111111.1111111.1111111.00000000
There are a total of 24 bits set to 1, which are used to identify the network.
There are a total of 8 bits set to 0, which are used to identify the host, and
these host bits can be stolen.
Stealing bits essentially involves changing host bits (set to 0 or off) in the
subnet mask to network bits (set to 1 or on). Remember, network bits in a
subnet mask must always be contiguous - skipping bits is not allowed.
Consider the result if three bits are stolen. Using the above formula:
2n = 23 = 8 = 8 new networks created
However, a total of 8 new networks does not meet the original requirement
of at least 10 networks. Consider the result if four bits are stolen:
2n = 24 = 16 = 16 new networks created
A total of 16 new networks does meet the original requirement. Stealing four
host bits results in the following new subnet mask:
11111111.11111111.11111111.11110000 = 255.255.255.240
Subnetting (continued)
In the previous example, a Class C network was subnetted to create 16 new
networks, using a subnet mask of 255.255.255.240 (or /28 in CIDR). Four
bits were stolen in the subnet mask, leaving only four bits for hosts.
To determine the number of hosts this results in, for each of the new 16
networks, a slightly modified formula is required:
2n – 2
Consider the result if four bits are available for hosts:
2n – 2 = 24 – 2 = 16 – 2 = 14 usable hosts per network
Thus, subnetting a Class C network with a /28 mask creates 16 new
networks, with 14 usable hosts per network.
Why is the formula for calculating usable hosts 2n – 2? Because it is never
possible to assign a host an address with all 0 or all 1 bits in the host portion
of the address. These are reserved for the subnet and broadcast addresses,
respectively. Thus, every time a network is subnetted, useable host addresses
are lost.

The 2n-2 Rule and Subnetted Networks


To avoid confusion, it was historically unacceptable to use the first and last
new networks created when subnetting, as it is possible for a classful
network to have the same subnet and broadcast address as its subnetted
networks. This required the 2n – 2 formula to also be used when calculating
the number of new networks created while subnetting.
However, this is no longer a restriction for modern equipment and routing
protocols. Specifically, on Cisco IOS devices, the following command is
now enabled by default:
Router(config)# ip subnet-zero

The ip subnet-zero commands allows for the use of networks with all 0 or all
1 bits in the stolen network portion of the address. Thus, the formula for
calculating the number of new networks created is simply 2n.
Remember though, the formula for calculating usable hosts is always 2n – 2.
(Reference: http://www.cisco.com/en/US/tech/tk648/tk361/technologies_tech_note09186a0080093f18.shtml)
Determining the Range of Subnetted Networks
Determining the range of the newly created networks can be accomplished
using several methods. The long method involves some binary magic.
Consider the example 192.168.254.0 network again, which was subnetted
using a 255.255.255.240 mask:
192.168.254.0: 11000000.10101000.11111110.00000000
255.255.255.240: 11111111.11111111.11111111.11110000
Subnetting stole four bits in the fourth octet, creating a total of 16 new
networks. Looking at only the fourth octet, the first newly created network is
0000. The second new network is 0001. Calculating all possible
permutations of the four stolen bits:
Binary Decimal Binary Decimal Binary Decimal
.0000 xxxx .0 .0110 xxxx .96 .1100 xxxx .192
.0001 xxxx .16 .0111 xxxx .112 .1101 xxxx .208
.0010 xxxx .32 .1000 xxxx .128 .1110 xxxx .224
.0011 xxxx .48 .1001 xxxx .144 .1111 xxxx .240
.0100 xxxx .64 .1010 xxxx .160
.0101 xxxx .80 .1011 xxxx .176
Note that this equates to exactly 16 new networks. The decimal value
represents the first (or the subnet) address of each newly created network. To
determine the range for the hosts of the first new network:
Binary Decimal Binary Decimal Binary Decimal
.0000 0000 .0 .0000 0110 .6 .0000 1100 .12
.0000 0001 .1 .0000 0111 .7 .0000 1101 .13
.0000 0010 .2 .0000 1000 .8 .0000 1110 .14
.0000 0011 .3 .0000 1001 .9 .0000 1111 .15
.0000 0100 .4 .0000 1010 .10
.0000 0101 .5 .0000 1011 .11
The binary value has been split to emphasize the separation of the stolen
network bits from the host bits. The first address has all 0 bits in the host
portion (0000), and is the subnet address for this network. The last address
has all 1 bits in the host portion, and thus is the broadcast address for this
network. Note that there are exactly 14 usable addresses to assign to hosts.
Determining the Range of Subnetted Networks (continued)
Calculating the ranges of subnetted networks can quickly become tedious
when using the long binary method. The shortcut method involves taking the
subnet mask (255.255.255.240 from the previous example), and subtracting
the subnetted octet (240) from 256.
256 – 240 = 16
Assuming ip subnet-zero is enabled, the first network will begin at 0. Then,
simply continue adding 16 to identify the first address of each new network:

0 16 32 48 64 80 96 112 128 144 160 176 192 208 224 240


Knowing the first address of each new network makes it simple to determine
the last address of each network:
First address of network 0 16 32 48 64 80 96 112 128 144
Last address of network 15 31 47 63 79 95 111 127 143 159
Only the first 10 networks were calculated, for brevity. The first address of
each network becomes the subnet address for that network. The last address
of each network becomes the broadcast address for that network.
Once the first and last address of each network is known, determining the
usable range for hosts is straightforward:

Subnet address 0 16 32 48 64 80 96 112 128 144


1 17 33 49 65 81 97 113 129 145

Usable Range

14 30 46 62 78 94 110 126 142 158


Broadcast address 15 31 47 63 79 95 111 127 143 159

Hosts on the same network (such as 192.168.254.2 and 192.168.254.14) can


communicate freely.
Hosts on different networks (such as 192.168.254.61 and 192.168.254.66)
require a router to communicate.
Class A Subnetting Example
Consider the following subnetted Class A network: 10.0.0.0 255.255.248.0
Now consider the following questions:
 How many new networks were created?
 How many usable hosts are there per network?
 What is the full range of the first three networks?
By default, the 10.0.0.0 network has a subnet mask of 255.0.0.0. To
determine the number of bits stolen:
255.0.0.0: 11111111.00000000.00000000.00000000
255.255.248.0: 11111111.11111111.11111000.00000000
Clearly, 13 bits have been stolen to create the new subnet mask. To calculate
the total number of new networks:
2n = 213 = 8192 new networks created
There are clearly 11 bits remaining in the host portion of the mask:
2n – 2 = 211 – 2 = 2048 – 2 = 2046 usable hosts per network
Calculating the ranges is a bit tricky. Using the shortcut method, subtract the
third octet (248) of the subnet mask (255.255.248.0) from 256.
256 – 248 = 8
The first network will begin at 0, again. However, the ranges are spread
across multiple octets. The ranges of the first three networks look as follows:
Subnet address 10.0.0.0 10.0.8.0 10.0.16.0
10.0.0.1 10.0.8.1 10.0.16.1

Usable Range

10.0.7.254 10.0.15.254 10.0.23.254


Broadcast address 10.0.7.255 10.0.15.255 10.0.23.255
Private vs. Public IPv4 Addresses
The rapid growth of the Internet resulted in a shortage of available IPv4
addresses. In response, a specific subset of the IPv4 address space was
designated as private, to temporarily alleviate this problem.
A public address can be routed on the Internet. Thus, hosts that must be
Internet-accessible must be configured with (or reachable by) public
addresses. Allocation of public addresses is governed by the Internet
Assigned Numbers Authority (IANA).
A private address is intended for internal use within a home or
organization, and can be freely used by anyone. However, private addresses
can never be routed on the Internet. In fact, Internet routers are configured to
immediately drop traffic with private addresses.
Three private address ranges were defined in RFC 1918, one for each IPv4
class:
 Class A - 10.x.x.x /8
 Class B - 172.16.x.x /12
 Class C - 192.168.x.x /24
It is possible to translate between private and public addresses, using
Network Address Translation (NAT). NAT allows a host configured with
a private address to be stamped with a public address, thus allowing that host
to communicate across the Internet. It is also possible to translate multiple
privately-addressed hosts to a single public address, which conserves the
public address space.
NAT provides an additional benefit – hiding the specific addresses and
addressing structure of the internal (or private) network.
Note: NAT is not restricted to private-to-public address translation, though
that is the most common application. NAT can also perform public-to-public
address translation, as well as private-to-private address translation.
NAT is only a temporarily solution to the address shortage problem. IPv4
will eventually be replaced with IPv6, which supports a vast address space.
Both NAT and IPv6 are covered extensively in other guides.
Reserved IPv4 Addresses
In addition to the three private IPv4 ranges, several other addresses and
ranges are reserved for specific purposes:
 The 0.0.0.0 /0 network is used to identify all networks, and is referred
to as the default route. If a default route exists in a routing table, it
will be used only if there is not a more specific route to a particular
destination. Routing and default routes are covered extensively in
another guide.

 The 0.0.0.0 /8 range is used to identify hosts on the local network.


Addresses in this range can only be used as a source address. The
most commonly used address in this range is 0.0.0.0 /32, which a host
will use when dynamically attempting to learn its IP address via
Dynamic Host Configuration Protocol (DHCP). DHCP is covered
extensively in another guide.

 The entire 127.x.x.x /8 range is reserved for diagnostic purposes. The


most commonly used address in this range is 127.0.0.1, which
identifies the local host, and is referred to as the loopback or
localhost address.

 The 169.254.x.x /16 range is reserved for Automatic Private IP


Addressing (APIPA). A host assigns itself an address in this range, if
it cannot dynamically obtain an address from a DHCP server.

 The 224.x.x.x – 239.x.x.x ranges are reserved for multicast, and are
referred to as Class D addresses.

 The 240.x.x.x – 255.x.x.x ranges are reserved for future and


experimental use, and were formerly referred to as Class E addresses.

 The 255.255.255.255 address can be used as a broadcast address for


the local network.
The IPv4 Header
The IPv4 header is comprised of 12 required fields and 1 optional field.
The minimum length of the header is 160 bits (20 bytes).
Field Length Description
Version 4 bits Version of IP (in this case, IPv4)
Internet Header Length 4 bits Specifies the length of the IP header (minimum 160 bits)
DSCP 8 bits Classifies traffic for QoS
Total Length 16 bits Specifies the length of both the header and data payload
Identification 16 bits Uniquely identifies fragments of a packet
Flags 3 bits Flags for fragmentation
Fragment Offset 13 bits Identifies the fragment relative to the start of the packet
Time to Live 8 bits Decremented by each router traversed
Protocol 8 bits Specifies the next upper layer protocol
Header Checksum 16 bits Checksum for error checking
Source Address 32 bits Source IPv4 address
Destination Address 32 bits Destination IPv4 address
Options Variable Optional field for various parameters

The 4-bit Version field is set to a value of 4 for IPv4.


The 4-bit Internet Header Length field identifies the length of the IPv4
header, measured in 32-bit words. The minimum of length of an IPv4 header
is 160 bits, or 5 words (32 x 5 = 160).
The 8-bit Differentiated Service Code Point (DSCP) field is used to
classify traffic for Quality of Service (QoS) purposes. QoS is covered in
great detail in other guides. This field was previously referred to as the Type
of Service (ToS) field.
The 16-bit Total Length field identifies the total packet size, measured in
bytes, including both the IPv4 header and the data payload. The minimum
size of an IPv4 packet is 20 bytes – essentially a header with no payload.
The maximum packet size is 65,535 bytes.
The IPv4 Header (continued)
Field Length Description
Version 4 bits Version of IP (in this case, IPv4)
Internet Header Length 4 bits Specifies the length of the IP header (minimum 160 bits)
DSCP 8 bits Classifies traffic for QoS
Total Length 16 bits Specifies the length of both the header and data payload
Identification 16 bits Uniquely identifies fragments of a packet
Flags 3 bits Flags for fragmentation
Fragment Offset 13 bits Identifies the fragment relative to the start of the packet
Time to Live 8 bits Limits the lifetime of a packet
Protocol 8 bits Specifies the next upper layer protocol
Header Checksum 16 bits Checksum for error checking
Source Address 32 bits Source IPv4 address
Destination Address 32 bits Destination IPv4 address
Options Variable Optional field for various parameters

An IPv4 packet that is larger than the Maximum Transmission Unit


(MTU) size of a link must be fragmented. By default, the MTU for
Ethernet is 1500 bytes.
Three fields are used when a packet must be fragmented - the 16-bit
Identification field, the 3-bit Flags field, and the 13-bit Fragment Offset
field. Each fragment of the packet is marked with the same Identification
number. The Fragment Offset allows the destination host to reassemble the
fragments in the proper order.
The Flags field dictates two conditions:
 Don’t Fragment (DF) – indicates the packet cannot be fragmented. If
a packet exceeds a link’s MTU size and this flag is set, then the packet
is dropped. An ICMP error message will then be sent back to the
source host.
 More Fragments (MF) – all fragments have this bit set to one,
except for the last fragment, where the bit is set to zero. This allows
the destination host to know when it has received all fragments.
The IPv4 Header (continued)
Field Length Description
Version 4 bits Version of IP (in this case, IPv4)
Internet Header Length 4 bits Specifies the length of the IP header (minimum 160 bits)
DSCP 8 bits Classifies traffic for QoS
Total Length 16 bits Specifies the length of both the header and data payload
Identification 16 bits Uniquely identifies fragments of a packet
Flags 3 bits Flags for fragmentation
Fragment Offset 13 bits Identifies the fragment relative to the start of the packet
Time to Live 8 bits Limits the lifetime of a packet
Protocol 8 bits Specifies the next upper layer protocol
Header Checksum 16 bits Checksum for error checking
Source Address 32 bits Source IPv4 address
Destination Address 32 bits Destination IPv4 address
Options Variable Optional field for various parameters

The 8-bit Time to Live (TTL) field limits the lifetime of the packet,
preventing it from being endlessly forwarded. When a router forwards a
packet, it will decrement the TTL value by one. Once the TTL value reaches
zero, the packet is dropped.
The 8-bit Protocol field identifies the next upper-layer header, and is
covered in the next section.
The 16-bit Header Checksum field is used to error-check the IPv4 header.
The receiving host will discard the packet if it fails the checksum
calculation.
The 32-bit Source Address field identifies the sending host. The 32-bit
Destination Address field identifies the receiving host. The value of both of
these fields can be changed as the packet is forwarded, using NAT.
The variable-length Options field provides additional optional IPv4
parameters, outside the scope of this guide.
IPv4 Protocol Numbers
The 8-bit Protocol field specifies the next upper-layer header within the data
payload of the packet. These upper-layer protocols are identified using IP
Protocol Numbers.
The following is a list of common IP Protocol Numbers, as assigned by the
IANA:
Protocol Upper-Layer Protocol
Number

1 ICMP
2 IGMP
6 TCP
9 IGRP
17 UDP
46 RSVP
47 GRE
50 IPSEC ESP
51 IPSEC AH
88 EIGRP
89 OSPF

In IPv6, this field is referred to as the Next Header field.

(Reference: http://www.iana.org/assignments/protocol-numbers)
Resolving Logical Addresses to Hardware Addresses
A host cannot directly send data to another host’s logical address. A
destination logical address must be mapped to a hardware address, so that
the Data-Link layer can package a frame to transmit on the physical
medium.
The Address Resolution Protocol (ARP) provides this mechanism for IPv4
on Ethernet networks. ARP allows a host to determine the MAC address for
a particular destination IP address.

Consider the above diagram. The following demonstrates the steps required
for HostA to communicate with HostB:
 First, HostA will determine if the destination IP address of 10.1.1.6 is
itself. If that address is configured on a local interface, the packet
never leaves HostA. In this example, 10.1.1.6 is not locally
configured on HostA.
 Next, HostA will determine if the 10.1.1.6 address is on the same
network or subnet as itself. HostA consults its local routing table to
make this determination. In this example, the subnet mask is /16.
Thus, HostA’s IP address of 10.1.1.5 and the destination address of
10.1.1.6 are on the same network (10.1).
 Because HostA and HostB are on the same network, HostA will then
broadcast an ARP request, asking for the MAC address of the
10.1.1.6 address.
 HostB responds to the ARP request with an ARP reply, containing its
MAC address (AAAA.BBBB.CCCC).
 HostA can now construct a Layer-2 frame, with a destination of
HostB’s MAC address. HostA forwards this frame to the switch,
which then forwards the frame to HostB.
Resolving Logical Addresses to Hardware Addresses (continued)
Now consider a slightly modified scenario between HostA and HostB:

 Again, HostA will determine if the destination IP address of 10.2.1.5


is itself. In this example, 10.2.1.5 is not locally configured on HostA.
 Next, HostA will determine if the 10.2.1.5 address is on the same
network or subnet as itself. In this example, the subnet mask is /16.
Thus, HostA’s IP address of 10.1.1.5 and the destination address of
10.2.1.5 are not on the same network.
 Because HostA and HostB are not on the same network, HostA will
parse its local routing table for a route to this destination network of
10.2.x.x/16. Hosts are commonly configured with a default gateway
to reach all other destination networks.
 HostA determines that the 10.1.1.1 address on RouterA is its default
gateway. HostA will then broadcast an ARP request, asking for the
MAC address of the 10.1.1.1 address.
 RouterA responds to the ARP request with an ARP reply containing
its MAC address (4444.5555.6666). HostA can now construct a
Layer-2 frame, with a destination of RouterA’s MAC address.
 Once RouterA receives the frame, it will parse its own routing table
for a route to the destination network of 10.2.x.x/16. It determines that
this network is directly attached off of its Ethernet2 interface.
RouterA then broadcasts an ARP request for the 10.2.1.5 address.
 HostB responds to the ARP request with an ARP reply containing its
MAC address (AAAA.BBBB.CCCC). RouterA can now construct a
Layer-2 frame, with a destination of HostB’s MAC address.
Resolving Logical Addresses to Hardware Addresses (continued)
Consider the following example again:

Note that as a packet is routed, the source and destination IP address remain
unchanged. However, both the source and destination MAC address did
change.
This is because a MAC address contains no network hierarchy, and thus is
only significant on the local network. In the above scenario, HostA and
HostB could not communicate directly using Layer-2 addressing. At every
routed hop, the source and destination MAC address are adjusted to reflect
the source and destination hosts on the local network.
The source and destination IP address will only be changed if NAT is used.

The ARP Table


A host can build an ARP table that contains a list of IP to MAC address
translations. The ARP table is only locally significant to that host. There are
two methods to populate an ARP table:
 Statically
 Dynamically
A static ARP entry is created manually on a host, and will remain
permanently until purposely removed. More commonly, ARP tables are built
dynamically by caching ARP replies. Cached entries will eventually be
aged out of the ARP table. The aging time will vary depending on the
operating system, and can range from several seconds to several hours.
Troubleshooting IP using ICMP
The Internet Control Message Protocol (ICMP) is used for a multitude of
informational and error messaging purposes.
The following is a list of common ICMP types and codes:
Type Code Description
0 0 Echo Reply

3 - Destination Unreachable
0 Network Unreachable
1 Host Unreachable
2 Protocol Unreachable
3 Port Unreachable
4 Fragmentation Needed – Don’t Fragment Flag Set
6 Destination Network Unknown
7 Destination Host Unknown
9 Destination Network Administratively Prohibited
10 Destination Host Administratively Prohibited

5 Redirect
8 Echo
11 TTL Exceeded

The two most common troubleshooting tools that utilize ICMP are:
 Packet Internet Groper (ping)
 Traceroute
Ping is a core connectivity troubleshooting tool, which utilizes the Echo
Request and Echo Reply ICMP messages to determine if an IP address is
reachable and responding. Ping will additionally provide the round-trip
time between the source and destination, usually measured in milliseconds.
Traceroute determines the routing path a packet takes to reach its
destination. Traceroute will not only identify each router the packet has been
forwarded through, but will also measure the delay experienced at each
router hop.

You might also like