Today I am writing for my nephews and babies who are starting their journey to networking world. There will be multiple articles on basics. This one is on IP Addressing.
Yes, all my brother who just want to go to the flashback can also read this article.
IPv4 Addresses
IPv4 addresses are 32 bits long (mean 32 0s or 1s), divided
into two parts network address and the host address. The network portion
identifies a physical links and is same for all the hosts connected to that
physical network. The host portion identifies a particular device only.
An IPv4 Address can be represented in different formats –
Bits format –
00001010110101100101011110000011
Decimal format –
181819267
Both the representation above are hard to remember and
calculate. Therefore a doted decimal representation developed and used by
network engineers throughout the networking journey (don’t be surprised router
sees and works only with binary format only 😊)
With doted decimal format, above 32-bit address is divided
into 4 parts, each part is called the octets which is 8-bits long or 1-byte
long. With doted decimal format, lets perform the first step –
00001010.11010110.01010111.10000011
And now convert the same to decimal format –
00001010.11010110.01010111.10000011
10.214.87.131
Points to remember – a possible value in each octets can be
between 0 – 255 (with all bits set to 1 will result 255 and all bits set to 0
will result 0)
An important factor of IPv4 addressing is that it allows you
to perform more calculations compared to other network layer protocols such as
NetWare and AppleTalk there we have fixed network and host portion – this
certainly makes life simple. In IPv4 the calculation is required to conserve
the IP address for a network – I mean when we have small number of users in a
network, there is no need to have 1000s of devices in that network. With the
help of subnetting we’ll see how effectively we can design our IP addressing schema
for network without wasting IPs.
IPv4 Addressing was designed for all types of network –
small, medium and large. Wait a minute what is definition of small, medium or
large I mean how many hosts can be there in each class of network. Ok that is
governed by “First Octet Rule” which we are going to see next.
First Octet Rule
There a in general three types sizes of network as measured
by the number of hosts – big, small and medium;
Big networks have a large number of hosts and less number of
large networks
Small network are just opposite to big networks with less
number of hosts and more number of small networks
Medium network has medium number of networks and medium
number of hosts in each network
Now this high-level of networks require three types of
classes of network addresses. Big network addresses should be capable to host
many hosts and few network addresses are required. This situation is opposite
for small network, as there are many small networks, a large number of
small-network addresses are needed, because small network has small number of
hosts, each of the network address require few hosts to address.
Address Class
These three types of address format can be represented as
follows
Figure 1-
Class A, B and C Address Format
The big, medium and small networks are mapped to these
address classes.
Class A IPv4 Addresses are for big networks. In class A
address, first octet is the network portion and the next three octets are the
host portion. If we do a simple math only 256 numbers are available for network
part. However a huge number of hosts can exist in each network – 224
or 16,777216 hosts.
Class B IPv4 Addresses are for medium networks. In class B
address, first two octets are for the network portion and the next two octets
are for the host portion. In class B address there are 216 or 65,536
networks and hosts in each network.
Class C IPv4 Address are for small networks. In Class B
address, first three octets are for the network position and the next 1 octet
is for the host portion.
As all the IPv4 address are 32 bits long and can have any
value between 0 and 255 how can one recognize whether an address is class A, B,
or C IP address? There comes the 1st octet rule in picture.
According to this rule –
1.
For class A address, the first bit of 1st
Octet (left most) is always 0. With this rule the class A IP address can starts
with any value between 0 and 127; IP address starting with 0 is reserved for
default address and 127 is reserved for internal loopback address. This leaves
a Class A valid IP address first octet is between 1 and 126.
2.
For class B address, the first 2 bits are set as
“10”. With binary to decimal conversion formula it is clear the first octet for
class B address falls between the 128 and 191.
3.
For class C address the first two bits are set
to 1 and third bit is set to zero. The result is, the first octet range is
between 192 – 223.
Address Mask
Looking at the IP address we cannot findout what portion identifies
the network address and the host address. To solve this puzzle, address mask is
used. The address mask is a 32 bit text where one bit for each bit of the IPv4
address. It is also represented in dotted decimal format same as the IPv4
address. It is to remember address mask looks like the IP address but it is not
the one. There are standard subnet mask for each class of IPv4 Address –
Figure 2:
Standard Address Masks
For each bit of IPv4 address, the device under the hood
performs the Boolean AND function with the corresponding bit of the address
mask. The AND function – compare two bits and derive a result. The result will
be one if and only if, both bits are one. If either or both bits are zero, the
result will be zero.
Ahh let’s try to understand the same with example,
Figure 3:
AND Operator
Figure 4:
Deriving NW Address using AND Operator
As per the figure 4, a logical AND is performed on the IPv4
address and it’s address mask for every bit; the result is the 172.21.0.0 which
is the network address for the host 172.21.35.17 with address mask 255.255.0.0.
Why Subnetting?
There can be multiple locations in a network and each
location needs to be logically identified so that traffic from one side can
reach to other side. This on a very high level is called the routing function.
Each network should be unique – also each host on the network must have an
address that identifies it uniquely in the network and it’s network from other
networks.
Can’t we use standard subnet mask?
Probably can be used (in a lab environment) but in real-life
scenario, a single class A, B, or C address can be used only on a single data
link.
Figure 5:
NW with default Subnet Masks
The problem with above network is scalability, with default
subnet mask and addressing the size of the network will be limited and
unnecessary there will be wastage of the host IP addresses. Consider the Data
Link 2 in Figure 5 is a point-to-point link (effectively I need 2 IPs here) but
as the default subnet and addressing is used, there are 65,000 devices can reside
wasting 64,998 address.
The only way to make IP address practical in real network,
it is necessary to perform the subnet where a big or major network is divided
into sub-networks addresses.
Let’s try to understand the same by simple example, look at
the below network architecture where 172.21.0.0 is divided into subnetworks –
Figure 6
- Subnetworks - using single network address across multiple links
You must have noticed that the subnet mask is not the
standard class B subnet mask with 16-bit reserved for network address. Instead,
we have borrowed additional 8 bits from the host portion of the address – so
now we have 24-bits to represent the network address. This resulted that the
major network 172.21.0.0 has been divided into subnetworks and each datalink
has been addressed instead of wasting the IP addresses using the major network
address across the datalinks.
Now after subnetting there are 3 part of the IPv4 Address –
network, subnet, and the host part. The address mask is now called – subnet
mask. First two octets are the major network (172.21.x.x) and will remain
unchanged. However, the 3rd octet is subnet bits instead of host
bit. I range from 0 to 255.
Just a point to remember, not all routing protocols
(classful routing protocol) can understand the subnet address where all
subnet-bits are 0 or all subnet-bits are 1. For example – 172.21.0.0/24 and
172.21.255.0/24. It is because they are not able to differentiate 172.21.0.0 which
is major subnet address or the first subnetwork 172.21.0.0. In this case both
are same. Likewise, 172.21.255.0 – it is major network broadcast address and at
the same time last subnetwork ☹. IGRP and RIPv1 are classful routing protocols that
don’t allow to use all-ones and all-zero subnets in a network. On the other
hand, classless routing protocol such as EIRP, OSPF etc. understand the
differentiator – as the subnet mask information is communicated in the routing
updates.
Sometimes we also get confused with the verbal communication
of the subnet during our day to day troubleshooting or support tasks. So what
is the correct way to communicate a piece of information when subnetting is
used as we have done in above feature. It means “Class B address with 8-bits of
Subnetting” or “Class B address with 24-bits mask”. I hope it is clear now.
A subnet mask can be represented in the dotted decimal
format – 255.255.255.0; in bits count format 172.21.1.0/24; or in Hexadecimal
format 0xFFFFFF00.
Designing a IP Addressing for a network
As it is already mentioned that all-zero subnet-bits and the
all-ones subnet-bits are not understood by classful routing protocols,
likewise, all-zero host address and all-ones host address are reserved in IPv4
networking. Router use all-zero host address as – subnet address and all-ones host
address as – broadcast address i.e. 172.21.1.0/24 is the subnet address and
172.21.1.255/24 is the broadcast address in subnet 172.21.1.0/24.
When you have been assigned a task to build a network and
designed the IP schema you have to go through the series of steps to
successfully come out as an architect –
Step 1 – Determine how many subnets and how many hosts per
subnets required
Step 2 – Use 2n – 2 formula to determine the
number of subnet bits and the number of hosts bits that will satisfy the
requirement established in step 1. If multiple subnet masks meet the
requirement, choose one that will best scale to future need.
Step 3 – working in binary, determine all available bit
combinations in the subnet space; in each instance, set all the host bits to zero.
Convert the resulting subnet address to dotted decimal. These are the subnet
addresses.
Step 4 – for each subnet address, again working in binary,
all possible bit combinations for the host space without changing the subnet
bits. Convert the result to doted decimal; these are the host addresses available
for each subnet.
Let’s work in a real scenario and try to design the IP
Addressing for a customer. This customer need IP design for 500 remote
locations and each location has around 100 hosts.
Step 1 – using the 9 bits for subnet can meet the
requirement - 29 - 2 = 510 (locations considering the scalability into
mind); if we use 8 bits for subnets than the requirement cannot be met as it
will give only 254 subnets only.
Let’s work on the below network setup where we do need 5
subnets for each data-link in the topology –
Figure 7: Network with 5 Datalinks and mini. 25 IP host
Network 192.168.100.0 to be divided in such as way that provide
at least 5 subnet addresses. The maximum host addresses are 25 for two LAN
subnets. Therefore full IP Addressing design require 5 subnets with each
minimum 25 host addresses.
Applying 2n -2 formula, three subnet bits and
five host bits will satisfy the requirement : 23 – 2 = 6 and 25
– 2 = 30. A class C subnet mask with 3 bits of subnetting is represented as
255.255.255.224 in the dotted decimal format.
Figure 8: Network, Subnet Bits & Host bits - Subnet
Addresses
Figure 9: Calculating the Subnet Address counting up
Subnet Bits from 0 in binary
Last step to calculate the host addresses available to each
subnet. This step is performed by choosing the subnet mask and keeping the
network & Subnet bits unchanged, writing all bit combinations in the host
space by counting up from zero in binary. Let’s take an example for 192.168.100.32
subnet –
Figure 10: Calculating the host address
counting up host bits from 0 in binary
Note – the first address, in which all the host bits are
zero, is the subnet address – that’s the reason it cannot be assigned to host.
Another address is the last address where all the host bits are 1, this is the
broadcast address in the subnet that represent to all hosts in the subnet. This
is also not assigned to host. In any subnet the usable IPs are – 2n –
2; 25 - 2 = 30.
With is I take a halt, the only way to mastery the IP
Addressing is to practice. A homework for you to calculate valid host for –
·
Class C IP addresses with 4 Subnet bits
·
Class C IP address with 5 Subnet Bits
Find out subnet broadcasts and network addresses as well. Happy
Learning!
