CSC 311 CHAPTER ELEVEN INTERNET PROTOCOLS AND APPLICATIONS In this chapter we discuss THE INTERNET there can be many internets, but there is only one Internet We will be talking about TCP/IP Internet Protocol IP Transmission Control Protocol TCP
INTERNET PROTOCOLS AND
TCP Transmission Control Protocol
IP Internet Protocol
TCP provides connection oriented services for layer 5 of the
protocol stack and relies on IP to route packets through
Two ends that implement TCP execute a handshake that establishes
a logical connection between them. Each side then executes flow
control protcols, acknowledge segments, and responds to those
that arrive damaged.
UDP (User Datagram Protocol) is an alternative layer 4 protocol.
Connectionless, no flow control, no guaranteed delivery.
Some of the protocols supported by TCP/IP
a packet delivery service between two sites.
It is commonly but not exclusively used with TCP
Suppose two applications, A and B, need a connection-oriented
service. TCP provides the reliable connection and IP handles routing
through the different networks.
to users, an internet address has the form:
This might appear in your email address as:
for browsers, the term www represents the default server at
the specified location.
periods are used to separate the terms.
In this context, domain does not have the same meaning as that used
in prior chapters.
A DOMAIN is a collection of sites of a particular type. They have
no geographic significance.
Table 11.1 contains a list of domain names, some of which are new and
may not be familiar to youl
can be omitted.
We can have more components in this text address:
In this case, top domain is country us in this example
it is divided into two subdomains:
wi for Wisconsin
the wi subdomain is further divided into “state” to indicate
the remaining components indicate the legislative branch and the
default web server.
These addresses are translated into actual internet addresses
Use dotted decimal notation.
Actual address in a 32 value divided into 4 eight bit fields
the address on the previous slide actually has an address of:
The maximum value in any of these fields is 255 which is the largest
unsigned integer that can be represented with 8 bits.
IP has several classes of internet addresses.
0nnnnnnn xxxxxxxx xxxxxxxx xxxxxxxx
Class B: 16,384 possible networks with up to 65,536 nodes
10nnnnnn nnnnnnnn xxxxxxxx xxxxxxxx
Class C: 2,097,152 possible networks, with up to 256 nodes
110nnnnn nnnnnnnn nnnnnnnn xxxxxxxx
Class D addresses used for multicasting
1110 followed by a 28 bit multicast address
Class E reserved
IPv4 is beginning to encounter problems. With 32 bits addresses,
we could have over 4.3 billion unique addresses, but the class
address system is inefficient, many addresses are wasted and we
are starting to run out.
Using the smaller class C addresses increases the number of network
addresses that routers must deal with, thereby complicating
the routing problem.
One solution to the problem is a new addressing scheme:
IPv6 uses 128 bit addresses.
Another approach is to use classless addressing.
Called the classless interdomain routing or CIDR, it is currently
supported by BGP-4 (border gateway protocol version 4)
It specifies a group of addresses that do not fall into any of the
predefined classes, yet each address in the group can still be
interpreted as a network number followed by a local identifier.
The number of bits defining the network number varies to
allow networks of varying size.
It is commonly used to allocate multiple class C networks.
suppose and organization the need of up to 1000 stations.
The CIDR approach would be to allocate four consecutive
class C networks.
Consider the following addresses:
126.96.36.199 11010011-11000011-00001000-xxxxxxxx 188.8.131.52 to 184.108.40.206
220.127.116.11 11010011-11000011-00001001-xxxxxxxx 18.104.22.168 to 22.214.171.124
126.96.36.199 11010011-11000011-00001010-xxxxxxxx 211,195.10.0 to 188.8.131.52
184.108.40.206 11010011-11000011-00001011-xxxxxxxx 220.127.116.11 to 18.104.22.168
In general, these Class C networks correspond to the contiguous set of
addresses from 22.214.171.124 to 126.96.36.199
If you examine the addresses carefully, you will note that the first 22 bits of all
four addresses are the same. So we can view any of these Class C networks as
a 22 bit network address followed by a 10 bit local identifier.
Furthermore, a router could extract the network number (in this case 188.8.131.52)
via a logical AND between the 22 bit subnet mask and an IP address
If we used 8 Class C addresses the first 21 bits would be the same an you could
have 2K nodes
For 16 Class C addresses the first 20 bits would be the same and you could have
The first three bits determine whether it is a Class A, B, or C address
but what can be done when the number of bits in the network number
varies, as it does with CIDR
The router must know the number of bits in the network ID.
Consequently, the usual representation of a network address,
w.x.y.z is replaced by w.x.y.z./m, where m represents the number
of bits in the network ID.
For example, a router can represent the four networks above using
the single entry: 184.108.40.206/22, the 22 indicates the network number
is 22 bits long.
and others that register domain names for a fee.
These domains and IP addresses are kept in a distributed database,
host computer calls on one of these databases to translate the text
domain name into an internet address.
Zones in a DNS hierarchy
An IP Packet
Version: version of IP that created the packet
Header Length: number of 32 bit words in the packet header
Type of Service: packet handling requests. More recently QofS issues addressed by complex
Packet Length: length of entire packet
Identification, Flags, Fragment Offset: used in fragmentation
Time to Live: max. time for packet to remain on the Internet.
Protocol: Specifies higher layer protocol using IP
Checksum: Used for error detection on the packet headers
Source and Destination IP address.
Options: Used to request special treatment.
The packet’s identification value is placed into each packet’ Identification field.
The flag field contains a more fragments bit mfb
Each fragment will have an offset field to indicate where it goes in the reassembled
It measure offsets in units of 8 bytes.
Relies heavily on RIP-2 and BGP protocols previously discussed.
To get to the actual device on a LAN, we need a physical address
rather than an IP address
Answer: The router keeps tables correlating IP addresses of devices
on its network to their physical addresses.
The router, if it does not have this address, can obtain it by broadcasting
the IP address on the network, the device which has that IP address
will respond with its physical address. Obviously, we do not want to
have to perform such broadcasts for every packet received, so the Router
keeps a record of these responses.
What happens if a network card in one of the machines needs to be
replaced? The entries in the router database are purged periodically, so
new network cards would be detected fairly quickly. Much like the process
we discuss with bridge routing tables.
put in other stuff next time
There are fewer options in the header and the address fields
have been expanded to 128 bits.
The next header field allows insertion of an additional header
between the standard header and the payload to provide information
At present, there are 6 types of extension headers.
Obviously, 128 bit addresses provide for substantially more addresses
With IPv4 we had about 4.3 billion possible addresses.
With IPv6 we have 2128 possible addresses, an almost
unimaginably large number.
In fact, if IPv6 possible addresses were spread out evenly over the
surface of the earth, there would be 1024 addresses for each
square meter of the earth’s surface, many more than the total number
of IPv4 addresses currently available. It is inconceivable that this
supply of addresses could ever be exhausted.
Whereas IPv4 uses dotted decimal notation, IPv6 uses Hex/colon