Mattias' MYOB Blog

Graduate Software Developer

Follow me on GitHub

TCP/IP Basics (Work in progress)

The TCP/IP model is a simplification of the OSI model. In contrast to the OSI model, the TCP/IP model only has 4 layers, these are:

1. Application
2. Host-to-Host (Transport)
3. Internet (Network)
4. Network Access (Composed of the data link and physical layer)

I’ll primarily only be covering the Host-to-Host Layer and Internet Layer, as I feel that these are the most important layers when it comes sending information from one computer to another in regards to software.

Host-to-Host Layer

The purpose of this layer is to both encapsulate and de-encapsulate the necessary information needed to send information from one process on a computer to another process on another computer. In many cases, the host-to-host layer will divide large amounts of information from the application layer (i.e. from applications) into smaller chunks of information, and encapsulate these chunks into something known as a packets. All that these packets contain are chunks of information to be sent, and a header. In addition, the host-to-host layer is also in charge of reconstructing the sent packets into the initial message that was sent, and passes that information to the application layer.

Message from application > Encapsulated into packet > Sent to internet layer Internet layer pushes packet up > De-encapsulate packet > Send message to application

Header

As we can see above, the header contains a plethora of information that is necessary for information to be transported from one computer to another successfully. Some of the more noteworthy are:

  1. Source/Destination Port: This represents on which port the application in which our information is coming from and being sent to are.

  2. Checksum: The purpose of the checksum is to check if all the information that has been sent has been successfully received and not corrupted in any way.

Packets: As we mentioned before, a packet merely consists of the above header, and the original message from the application.

Host-to-Host Layer Primary Protocols

These protocols are basically ways in which we send the information, and which options in the header we are effectively choosing to use or not use. For example, for UDP, we don’t use the flags ACK, SYN, FIN etc.

  1. TCP (Transmission Control Protocol): The main purpose of TCP is to ensure that information is transported reliably and without errors. The transport layer does this task by establishing connections between network devices, acknowledging the delivery of information, and resending information that haven’t been received or it is corrupted when it arrives. TCP is known as a connection-orientated protocol. This means that for every message we send to another computer, we must receive a response from the other computer, that they have successfully received the entire message.

  2. UDP (User Datagram Protocol): If the main purpose of TCP is offer reliability , then the main purpose of UDP, is to ensure that information is transported as quick as possible, even if there may be errors. UDP is known as a connectionless protocol, and will send information without regard for reliability. E.g this means that when we do send a message, we aren’t expecting a response back indicating that the other computer has successfully received the message we sent.

For more in-depth analysis on how the transport layer works:
Resource 1
Resource 2

In addition, a great tool to actually visualize this process of encapsulation is wireshark: Wireshark

Internet Layer

The internet layer follows a similar approach as the host-to-host layer, in which it constructs and deconstructs information coming from the layer above and below it. Similar to the host-to-host layer, it encapsulates the information received into another packet, that consists of the given information, and another header. Additionally, the reason why we typically use TCP for reliability is that the internet layer is inherently unreliable, and therefore we must place something reliable on top of it, to ensure reliability.

(NOTE: The internet layer does a whole heap more, especially in the contexts of routers, i’ll slowly try to cover that material as well.)

Header

As we can see above, there is a tonne of information found in the packet. Some of the more noteworthy ones are:

  1. Source/Destination Address: These are the IP addresses of your computer and the computer to which you want to send your message to.

  2. Fragmentation Offset: These are used as a way to reconstruct the Packets received from the layer below, as packets aren’t always received in the order they were sent.

  3. Header Checksum: Similar to the host-to-host layer, the checksum is used to check if all of the information has been received and not corrupted in any way.

  4. Time-to-live (TTL): Represents how long we expect the packet sent to live for (i.e. an actual time value - 1 minute). We do this for a couple of reasons, one
    is that the packet may be stuck somewhere and we need some way to remove it. Another reason is to stop duplication of information, there are times where we may have to resend information, and if there exist duplicate packets, that can cause a lot of issues. Thus we put a timer on the packet.

IP (Internet Protocol): IP represents the protocol that is used on the internet layer for the TCP/IP model. You’ve probably heard of IPv4 and IPv6. IP represents the protocol and v4 or v6 represent the version types. Theres a couple of differences between the 2, the primary being the number of IP addresses available to each, additionally IPv6 doesn’t use fragmentation as a method of reconstructing packets.

Summary

From the above image, we can identify the process of encapsulation, from the application layer all the way to the link layer. At each layer, we can see new headers being applied to each chunk of data. And then similarly, as information is being passed up through the layers, each header will be de-encapsulated at the relevant layer.