 |
by
Matthew Martin
Transport Control Protocol over Internet Protocol (TCP/IP) is the main protocol used at the transport (layer 3) and network (layer 4) layers of the OSI model (q.v.) on the Internet. We have already looked at some of the basic concepts behind TCP/IP and its origins. We will now look in more detail at how each of these protocols works.
Internet Protocol (IP) addressing is based around four byte addressing. Each byte is referred to as an octet, since it consists of eight bits. In hexadecimal the values for eight bits (a byte) can be represented in two digits (00 to FF).
IP uses classes of networks to determine the nature of the addressing. The addressing scheme dictates that up to the initial four bits of the first byte are used to determine the class of the network (see the diagram below). The remainder of the bits are used to actually address the actual host in a network. There are four classes of network A, B, C, & D.
Class |
1st
byte |
2nd
byte |
3rd
byte |
4th
byte |
||||||||||||||||||||||||||||
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
A |
0 |
netid |
hostid |
|||||||||||||||||||||||||||||
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
B |
1 |
0 |
netid |
hostid |
||||||||||||||||||||||||||||
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
C |
1 |
1 |
0 |
netid |
hostid |
|||||||||||||||||||||||||||
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
D |
1 |
1 |
1 |
0 |
multicast address |
|||||||||||||||||||||||||||
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
The table below shows the IP addresses in base ten and binary for each of the four classes of networks. Also shown are the number of networks and the number of hosts supported on each type of network. Class D network addresses are reserved for multicasting.
Class |
Address
Range |
Number of Networks |
Number of Hosts |
|||
| base-10 |
binary |
|||||
| start |
finish |
start |
finish |
|||
|
A |
1.0.0.0
|
126.0.0.0 |
00000001. 00000000. 00000000. 00000000 |
01111110. 00000000. 00000000. 00000000 |
126 |
16777214 |
|
B |
127.1.0.0 |
191.254.0.0 |
01111111. 00000001. 00000000. 00000000 |
10111111. 11111110. 00000000. 00000000 |
16384 |
65534 |
|
C |
192.0.0.0 |
192.233.255.254 |
11000000.
00000000. 00000000. 00000000 |
11000000. 11101001. 11111111. 11111110 |
2097151 |
254 |
|
D |
233.0.0.0 |
255.0.0.0 |
11101001. 00000000. 00000000. 00000000 |
11111111. 00000000. 00000000. 00000000 |
n/a |
n/a |
The IP addresses have corresponding domain name services (DNS) addresses. The DNS names are based on names that are much easier for people to remember than the four sets of three digits that make up an IP address (in base-10). However there are no rules linking the two systems, the DNS system converts from one to the other simply by looking-up the correspondence in a table.
Originally when there was just the ARPANET, all computers had a list of all hosts and all the corresponding IP addresses. Today this would be too large for any one system to hold. So today the system is distributed among many machines. Addresses that are not known to a local gateway are forwarded to another, until a machine is found that can identify it. The DNS naming system is divided into name spaces called zones. The zones allow for a directed searching for a DNS resolution.
In some textbooks and online resources
a four layer network model is used to describe the workings of TCP/IP. The
four layer model simplifies many of the layers used in the OSI model, making
it easier to understand than the OSI model. The four layer model incorporates
the physical and data-link layers in the first layer (called the network access
layer), the network layer in layer two (Internet layer), the transport layer
in layer 3 and the application, presentation and session layers in layer four
(application layer).
Here we will focus on TCP/IP using the OSI model as this is the more widely used model and we can use TCP/IP as an example of how to use the OSI model as a tool for understanding the functioning of network protocols. But using the diagram above you should be able to understand how TCP/IP works both from the perspective of the four layer TCP/IP model and the seven layer OSI model. The four layer model is widely known with regards to TCP/IP and is therefore a good thing to know.
First of all it is worthwhile noting where the protocols of TCP and IP fit into the OSI model. As mentioned above they predominantly fit into layers 3 and 4, the transport and network layers. However, the functions of TCP are note restricted to just the transport layer.
The physical (layer 1) & data-link (layer 2) layers of the OSI model are not part of TCP/IP, since these work at higher layers. However it is worthwhile considering them in order to understand how they fit into the scheme.
Data is packaged into packets called frames. We will not consider the structure of the frames here. It is worth noting that the network and transport layer also package the data. Consequently there are layers of packaging.
Every piece of network hardware has a unique hardware address, the Media-Access Control (MAC) address. This number is fixed for hardware (such as Ethernet cards) at the factory. MAC addresses consist of six hexadecimal numbers, each of two digits.
The address resolution protocol (ARP) keeps track of the relationships between IP address (e.g. 192.168.0.11) and the hardware addresses, represented by the MAC address (e.g. 00:D1:5E:23:73:A1).
Internet Protocol (IP) is the protocol in the TCP/IP system that deals with the network layer, layer 3.
IP delivers the data to the address contained in the header of each packet. If the IP address cannot be found locally, then the data is passed on to t a local router or gateway that is able to forward it out of the local network. This forwarding process continues, perhaps through many routers, until it either reaches the correct IP address or is timed-out.
IP packages data into packets based on a 32-bit word. These packets are called datagrams. The packets contain information about the protocol being used and the destination address, as well as other information and the data itself. The structure of an IP packet header is shown below.
| Version |
IHL |
Service
type |
Total Length of Datagram |
|||
|
Identification |
Flags |
Fragmentation
offset |
||||
|
Time
to live |
Protocol |
Header checksum |
||||
|
Source Address |
||||||
|
Destination Address |
||||||
|
Options |
Padding |
|||||
|
Data,
starts here… |
||||||
On arriving at the destination, the IP packets may have become fragmented. This can happen as different machines re-package the data for forwarding. The IP software at the destination re-assembles the IP packets before passing the data up to the TCP software, at the transport layer. The protocol number contained in the datagram indicates which protocol is being used at the transport layer. (It is often TCP but could easily be UDP or another protocol.)
At the transport layer TCP provides connections between the source (host) and the destination. TCP provides reliable connections, using hand-shaking to establish connections from point-to-point.
TCP uses a three-way hand-shake to establish a connection, known as the SYN/ACK sequence.
TCP packages the data into packets based on 32-bit words. The structure of a TCP packet header is shown below.
| Source Port |
Destination Port |
|||
|
Sequence Number |
||||
|
Acknowledgement Number |
||||
|
Offset |
Reserved |
Flags |
Window |
|
|
Checksum |
Urgent Pointer |
|||
|
Options |
Padding |
|||
|
Data,
starts here… |
||||
Looking at the diagram for the four layer model, it is possible to see how the headers encapsulate the data for each layer of protocol.
Once data has been received by the TCP layer at the destination and reassembled from the packets, it needs to passed up to the application layer. In order to achieve this the port number is used. Different port numbers are used for different applications. e.g. port 25 is used for SMTP and port 80 for HTTP.
The data is passed up through the layers until it reaches the application or applications that will use it. Data is targeted to the correct application using the port number. Most common protocols and services have defined port numbers that are commonly used. It is possible to reassign these and also assign port numbers for new applications. Some services and port numbers typically used are shown below.
Service |
Port Number |
|
Echo |
7 |
|
FTP |
21 |
|
Telnet |
23 |
|
SMTP |
25 |
|
HTTP |
80 |