TCP/IP

• Model and Layers

• Bits and Number Bases
• IPv4 Addressing
• Subnetting
• Classless Interdomain Routing
• IPv6
 Model and Layers
• The OSI model consists of
seven layers:
• At the beginning of the  7: Application
course, we discussed two
 6: Presentation
primary conceptual
 5: Session
models of networking: OSI
 4: Transport
and TCP/IP.
 3: Network
 2: Data Link
 1: Physical
 Model and Layers
• The OSI model has a protocol stack, but we use it
primarily as a conceptual reference.
• In contrast, the TCP/IP protocol stack has been much
more widely adopted.
• It is divided into four layers -- Application,
Transport, Internet, and Network Interface (a.k.a.
Link) -- that more or less "map onto" the OSI layers.
 Model and Layers
• TCP/IP Layers:
o 4)
Application: Deals with the applications that process
network requests, along with their associated ports.
 A port is an address to which you send data to be received...
 by a particular application...
 for processing.
 You might think of it as a transport-layer address.
 Transport protocols like TCP and UDP use 65,536 different ports,
which can be grouped into three categories...
 Model and Layers
 Port types:
 "Well-known": 1-1023
 Registered: 1024-49151
 Private: 49152-65535
 The well-known ports are used by some of the more common
networking applications, such as...
 Port 22: Secure Shell (SSH)
 Port 25: Simple Mail Transfer Protocol (SMTP)
 Port 80: Hypertext Transfer Protocol (HTTP)
 Port 443: Secure HTTP

 See Table 6-4 for more such examples.

 Model and Layers
o 3) Transport: Responsible for type of connection between hosts and
acknowledgments of data sent/received.
 The two main transport-layer protocols are TCP and UDP, which are
connection-oriented and connectionless, respectively
 Transport Control Protocol (TCP) is connection-oriented, where it
initiates/confirms a connection, manages transfer, and closes said connection.
 This begins with a 3-packet sequence (each is a type of packet):
 SYN: Synchronizing
 From Host A to B, attempting connection
 Sequence number (x) for tracking packets (SEQ#)
 Length of zero because it contains no data (LEN)
 Model and Layers
 SYN ACK: Synchronizing Acknowledgement
 From Host B to A, acknowledging package from A
 Sequence number (y)
 Acknowledgement number (x+1), also called ACK#
 ACK: Acknowledgement
 From Host A to B
 Sequence number (x+1)
 Acknowledgement number (y+1)
 This is called a "handshake", after which point data packets are
transferred.
 Model and Layers
 Connection is terminated via a 4-packet sequence, where
each host:
 Sends a FIN packet...
 ...and receives an ACK packet.
 The User Datagram Protocol (UDP) is connectionless:
A packet is sent from a source to a destination.
 There is no acknowledgement from the other side.
 Transfercontinues until the source stops sending or the
destination stops accepting.
 Model and Layers
o 2) Internet: The addressing and routing of data packets
 Internet Protocol (IP) : Defines addressing scheme for sources and
destinations of data packets sent within or between networks
 Address Resolution Protocol (ARP) : Associating IP addresses with MAC
addresses
 Internet Control Message Protocol (ICMP) : Data flow control and diagnostics.

o 1) Network Interface, or Link

 LAN segments
 WAN connections
 Bits and Number Bases
• Numbers are expressed in bases, where...
o The base is the number of possible values a digit can have.
o The
range of values for a digit will be zero through the base
minus one.
• Examples:
o Decimal: 0-9
o Binary: 0-1
 Bits and Number Bases
• Conversion: You calculate the value of the number by
multiplying each digit by exponents of the base.
o Generally, you start where the right-most digit
o Binary-to-Decimal: 1 0 0 1 1

Digit 1 0 0 1 1
Exponent * 24 * 23 * 22 * 21 * 20
Product 16 0 0 2 1
SUM 16 16 16 18 19 .
 Bits and Number Value Quotient Remainder

Bases 719
359
359
179
1
1
o Decimal-to-Binary: 7 1 9 179 89 1
89 44 1
 Divide the number by two
44 22 0
 Place the remainder on the 22 11 0
end 11 5 1
 Repeat with the quotient, 5 2 1
placing the remainder before 2 1 0
the previous digit. 1 0 1
 Do this until you get a
quotient of zero. 1 0 1 1 0 0 1 1 1 1
 Bits and Number Bases
• Hexadecimal: Hex Binary Hex Binary
o Values are 0 through 15 0 0000 4 0100
o Digits are 0 - 9, with 10 - 15 1 0001 5 0101
represented by A through F 2 0010 6 0110
o A hex digit is equivalent to a 3 0011 ...
quartet (4 bits)

o Example: 7 1 9  10 1100 1111  2 c f

o This way, you can easily convert back and forth between the two
 Bits and Number Bases
• A number expressed in binary digits is a bit string, and
you can think of them as being ON (1) or OFF (0)
• For example:
1 0 1 1 0 0 1 1 1 1
• Selecting bits :
o Sometimes, you will want to "turn" some bits on or off
o This
will be the case in scenarios where individual bits or bit
sequences in the string have meaning, in their own right.
 Bits and Number Bases
• This can be accomplished by using a bit mask , along
with bitwise operations.
oA bit mask is simply a bit string, where the different bits or bit
sequences have special meaning
oA bitwise operation acts upon a bit pair to produce 0 or 1,
and we will look at two of them:
 OR is used to turn bits on
 AND is used to turn bits off
 Bits and Number Bases
• OR operation:
o Any bit or 1 is turned/left ON
o In contrast, any bit or 0 is simply left unchanged
Bit Mask Result
Turned ON
1 OR 1 1 (if zero, would have been)

0 OR 1 1 Turned ON
1 OR 0 1 Unchanged
0 OR 0 0 Unchanged

o Ifyou use a bit mask with OR , it will turn some bits on while
keeping the others as they were.
 Bits and Number Bases
• AND operation:
o Any bit and 0 is turned/left OFF
o In contrast, any bit and 1 is simply left unchanged
Bit Mask Result
1 AND 1 1 Unchanged
0 AND 1 0 Unchanged
1 AND 0 0 Turned OFF
Turned OFF
0 AND 0 0 (if zero, would have been)

o If
you use a bit mask with AND , it will turn some bits off
while keeping the others as they were.
 Bits and Number Bases
• Let's look at an example:
o Our original bit string:

1 0 1 1 0 0 1 1 1 1
o Bit string's decimal value: 7 1 9

oA bit mask: 1 1 1 1 1 0 0 0 0 0 (992)

 Mask applied with OR

1 0 1 1 0 0 1 1 1 1 719
OR
1 1 1 1 1 0 0 0 0 0 (992)

First 5
bits are
turned ON 1 1 1 1 1 0 1 1 1 1 (1007)
 Mask applied with AND

1 0 1 1 0 0 1 1 1 1 719
Last 5
AND bits are
1 1 1 1 1 0 0 0 0 0 turned
(992)
OFF

1 0 1 1 0 0 0 0 0 0 (704)
 IPv4 Addressing
• IP addressing allows hosts and other devices to have
routable addresses, both in LANs and within wider
networks -- most notably the Internet.
o MAC addresses provide for forwarding within a LAN, but IP
addresses let us extend beyond that.
o Thepredominant version of IP today is IPv4, though we will
cover IPv6 later on.
• IPv4 addresses are divided into five classes, indicated by
letters A-E
 IPv4 Addressing
Class A: 0.0.0.0 - 127.255.255.255
Class B: 128.0.0.0 - 191.255.255.255
Class C: 192.0.0.0 - 223.255.255.255
Class D: 224.0.0.0 - 239.255.255.255
Class E: 240.0.0.0 - 254.255.255.255
• We will primarily be using the first three classes, though
Class D is relevant to Chapter 9, "Routing Protocols".
• These classes are useful for demonstrative purposes,
though the classification system is now outdated...
 IPv4 Addressing
• An IPv4 address is expressed in 32 bits:
o In theory, this allows for 232, or 4294967296, possible IPs
o Each octet (8-bit chunk) will have a value in the range 0-255
you will see IP addresses expressed in decimal
o Normally,
form, where the octets' values are separated by periods.
o Example: www.google.com
 Decimal: 146.115.22.166
 Binary: 10010010011100110001011010100110
 IPv4 Addressing
• Generally, the bits of an IP address are divided into two
parts that, in combination, give the full network location
of a particular host.
o The network bits comprise the first part of the longer bit
string, and they convey the location of the network where the
host resides.
o Followingare the host bits, which indicate the location of
the host within the network.
each octet in an IPv4 address contains either
o Traditionally,
network bits or host bits, according to address class...
 IPv4 Addressing
+---------+---------------+---------------+---------------+
| Address | Network | Host | # Hosts |
| Class | Octets (Bits) | Octets (Bits) | per Network |
+---------+---------------+---------------+---------------+
| A | 1 (8) | 3 (24) | 1.68e7 (224) |
| B | 2 (16) | 2 (16) | 6.55e4 (216) |
| C | 3 (24) | 1 (8) | 2.56e2 (28) |
+---------+---------------+---------------+---------------+
See also Figure 6-13
• Depending on the number of host bits (vs network bits),
different classes of networks will have a different possible
number of hosts per network -- specifically, two raised to the
power of number of host bits.
 IPv4 Addressing
• Within each class, some IPs are designated as private:
Class A: 10.0.0.0 - 10.255.255.255
Class B: 172.16.0.0 - 172.31.255.255
Class C: 192.168.0.0 - 192.168.255.255
o These are for internal networks, or intranets, such as...
 The IT Lab's inner network
 A home network
o Private IP addresses are not routable over the Internet!
 IPv4 Addressing
• On a wider level, the Internet Assigned Number
Authority (IANA) is responsible for the allocation of
IP addresses.
o However,
it delegates this task to regional Internet
registries (RIRs), who allocate addresses according to
geographical location.
o In
North America, the American Registry for Internet
Numbers (ARIN) assigns IP addresses.
o Largeentities like ISPs and universities are allocated blocks of
IP addresses to further assign as they choose.
 Subnetting
• A network can be partitioned into smaller entities called
subnets.
o These create a hierarchical network structure.
o Subnetsare separated at layer 3, in the sense that you use IP
and routing to move between them
• Example: You have a network at address 192.145.17.0
+----------+----------+----------+----------+
| 192 | 145 | 17 | 0 |
+----------+----------+----------+----------+
| 11000000 | 10010001 | 00010001 | 00000000 |
+----------+----------+----------+----------+
 Subnetting
• This network, however, might be divided into 4 subnets:
192.145.17.0 (IP range: 192.145.17.0 - 192.145.17.63)
192.145.17.64 (IP range: 192.145.17.64 - 192.145.17.127)
192.145.17.128 (IP range: 192.145.17.128 - 192.145.17.191)
192.145.17.192 (IP range: 192.145.17.192 - 192.145.17.255)

o Each such subnet is logically independent of the others.

o Traffic
from one subnet to another would pass through a
router.
 Subnetting
• Subnets within a network are established by declaring a
space of subnet bits:
o These subnet bits are borrowed from the host bits
o Together with the network bits, they establish the identity of
the network and subnet
o Those bits also become the basis of a subnet mask
• The material that follows will be especially pertinent to
Lab 8 and Homework 8.
 Subnetting
• Consider the subnets -- in particular, their fourth-octet
binary values:
0: 00000000
64: 01000000
128: 10000000
192: 11000000
• Notice that the only bits that vary are the first two.
• This is because two bits were borrowed -- from the
host bits -- to serve as subnet bits.
 Subnetting
• This is where the math starts to come in...
o Let's identify two variables
 x (# of bits borrowed from host portion)
 y (# of total host bits, by address class)
o Based on this, we can calculate two possible values:
 Number of subnets: 2x

 Number of possible IPs per subnet: 2y-x

• For 2 subnet bits, we get 4 subnets, with 64 IPs each...

 Subnetting
• For each subnet – such as 192.145.17.64 – two of the
possible IPs are reserved for special uses:
o The subnet address: (192.145.17.64)
 All host bits are zeroes (64: 01000000)
 The IP identity of the subnet itself
o The broadcast address: (192.145.17.127)
 All host bits are ones (64: 01111111)
 Data sent to this address is broadcast to all hosts within the subnet
• Thus, # of possible hosts per subnet equals 2y-x - 2
 Subnetting
• To distinguish the net and subnet portion of an IP
address from the host portion, you will apply a subnet
mask
• A subnet mask is a 32-bit (four-octet) value that
resembles an IP address when expressed in decimal form.
o Thefirst N bits are all set to a value of one, where N is equal to
the number of network and subnet bits.
o Youapply a subnet mask to a network address by AND-ing the
two (see previous slides about bit masking).
 Subnetting
• In the example above, where we had the network
192.145.17.0...
o It is a Class C address, so there are 24 network bits
o Inaddition, we borrow two of the host bits so that we can
have four subnets
o Thus, in our subnet mask, the first 26 bits are set to one
 Binary: 11111111.11111111.11111111.11000000
 Decimal: 255.255.255.192
192.145.17.0 the network was not subnetted, at all, then
o If
we would have a mask of 255.255.255.0 (first 24 bits)
 Subnetting
• We will look at another example:
o What we know:
 IP address: 172.27.213.94
 Subnet mask: 255.255.240.0
o So, what is the subnet address?
o Tostart with, let's put our IP address and subnet mask into
binary form:
 Addr: 10101100.00011011.11010101.1100001
 Mask: 11111111.11111111.11110000.0000000
 Subnetting
• If we AND the bits...
10101100.00011011.11010101.1100001
11111111.11111111.11110000.0000000
...then we get this result:
10101100.00011011.11010000.0000000
• So, the subnet is 172.27.208.0
 Subnetting
• The lab will ask you to do such things as:
o Calculating the subnet of an IP address, by applying a subnet mask
o Determining a subnet mask, based upon IP address class and the
number of subnets to be established
o Given a particular subnet mask...
 How many subnets?
 What are the subnet address and broadcast address for each subnet?
 How many possible hosts per subnet?
• This, of course, leads us into the topic of CIDR...
 Classless Interdomain Routing
• So far, we have been looking at classful addressing,
in which a network is simply defined by the first one,
two, or three octets -- depending on the address class.
o Thatway, the network would have a range of possible IPs,
according to the number of host bits.
o Forexample, a Class A network has 24 host bits, allowing for
224 possible addresses within it
• The problem? Lots of unused IPs!
 Classless Interdomain Routing
• What if a Class A network did not need all 224 possible
addresses?
• This is part of why classful addressing is now obsolete.
• In its place, we now have the practice of supernetting,
which lets us combine smaller networks (or subnets) into
larger networks.
• For this, we use classless interdomain routing (CIDR)
notation to express the subnet mask in a much shorter
form: A backslash, followed by the number of bits.
 Classless Interdomain Routing
• CIDR notation example:
oA subnet mask of 255.255.255.0 would be expressed as /24
o Ifyou have a subnet at address 172.27.208.0 with a
netmask of 255.255.240.0
o ...the CIDR notation would be 172.27.208.0/20
• Just as you might partition a network into subnets by
borrowing host bits – networks can be combined by
borrowing network bits.
 Classless Interdomain Routing
• Example: Networks 172.21.0.0, 172.22.0.0, and
172.23.0.0
o Networks 172.21.0.0, 172.22.0.0, and 172.23.0.0,
being Class B, have a subnet mask of 255.255.0.0 or /16
those IPs share the first 14 bits
o However,
10101100.000101XX.XXXXXXXX.XXXXXXXX
o Therefore,
those could be treated as part of a larger network
-- a supernet -- of 172.20.0.0/14
 Classless Interdomain Routing
• Multiple classful networks, grouped together as a
supernet, are also called a CIDR block.
• While classful IP addressing would limit networks to
certain sizes by allocating address space 1 octet (8 bits)
at a time, CIDR allows for a much more flexible
allocation of IP ranges.
• This way, you do not have to allocate IP addresses to a
network beyond its needs.
 Classless Interdomain Routing

Source: https://upload.wikimedia.org/wikipedia/commons/2/26/CIDR_Address.svg
 Classless Interdomain Routing
• When grouping subnets into a CIDR block, they must
resolve to the same IP when the subnet mask is applied
to them. For example...
172.20.0.0/14
172.21.0.0/14
172.22.0.0/14
172.23.0.0/14
• ...would not be a problem because the /14 mask
resolves them to the same value of 172.20.0.0
 Classless Interdomain Routing
• However, these would not work as a CIDR block....
172.22.0.0/14
172.23.0.0/14
172.24.0.0/14
172.25.0.0/14
• ...would be problematic because the mask resolves
some to 172.20.0.0 and others to 172.24.0.0
 IPv6
• IPv4 addressing, using 32 bits, allows for roughly 4.3
billion unique IP addresses.
• As the Internet grows and more devices are connected
to it, this number is becoming insufficient.
• This is where IPv6 (also known as IPng) comes in.
• IPv6 addressing uses 128 bits -- which allows for 2128
possible addresses
 IPv6
• Whereas IPv4 addresses are usually written in dotted
decimal form (e.g., 192.168.0.1), IPv6 addresses are
expressed in hexadecimal digits -- separated by colons.
o Example: 9b32:e6da:d14f:6698:a9e5:7fae:1ba2:ed81
o Theexample would be considered a full IPv6 address
because none of the hex digits are zero.
• When some of the digits are zeroes, there may be ways
to "compress" the zeroes to shorten the address.
 IPv6
• Zero compression: Replace consecutive zeroes with
two colons
o From: 9b32:0000:0000:0000:a9e5:7fae:1ba2:ed81
o To: 9b32::a9e5:7fae:1ba2:ed81
• Leading zero compression: For individual quartets,
omit leading zeroes
o From: 9b32:000a:d14f:0698:00e5:7fae:0002:ed81
o To: 9b32:a:d14f:698:e5:7fae:2:ed81
 IPv6
• Both compression types:
o From: 9b32:0000:0000:0000:00e5:7fae:0002:ed81
o To: 9b32::e5:7fae:2:ed81
• To recover the original IPv6 address from its compressed
form...
o Start with the rightmost digit (of the latter)

o Place each into their appropriate slots, from right to left

o Fill in zeroes as needed.
 IPv6
• To convert an IPv4 address (172.27.213.94) to IPv6:
o Convert the 32-bit address to 2 quartets of hexadecimal
digits
ac1b d55e
o Separate the quartets by a colon
ac1b:d55e
o Place two colons at the start, to indicate the leading zeroes:

::ac1b:d55e
 IPv6
• IPv6 addresses belong to three categories:
o Unicast: Associated with a single network interface
controller on a networked device.
o Multicast:Indicates a group of devices, and data sent to
such an address will be sent to the entire group.
o Anycast: Comes from a list of addresses.
• Although IPv6 allows for a much better range of
addresses, IPv4 is near-universal and will be in play for a
long time to come.
 IPv6
• There are a number of technologies out there to
facilitate the transition to IPv6.
• One such technology is the 6to4 prefix, which allows
IPv6 devices to use IPv4 networks.
o Thisinvolves the use of special 6to4 devices that do the
routing required.
oA 32-bit IPv4 address will be included within the larger 128-
bit IPv6 address
 IPv6
• Until IPv6 becomes more common, there are other solutions
out there for the issue of limited IPv4 addresses.
o For example, a private IP address -- not being routable over the
Internet -- can be used by many different hosts -- so long as it is
unique within a private network.
 In the IT Lab: 10.0.0.0/24 addresses
 On home networks: 192.168.0.0
o When hosts on private networks need Internet connectivity, they may
use Network Address Translation (NAT) -- where the router
replaces the inner, private source IP with its own outer, public one.