Mastering the TCP and OSI Models
Why Torture Students with This?!?
One of the questions that students sometimes ask me in my CCNA classes is, “why am I being tortured with this [expletive deleted] OSI model theory?” Wellhere is the answer. You see, these models were very dearly needed early on in networking to help ensure that equipment from various manufacturers would work together, and that the technologies could be better designed and understood. OK, but again, why do we need to torture students with it? The answer lies in the fact that networking technologies can really be better understood when you master the models. And if you understand the technologies better, you can implement them more effectively, and perhaps most importantly, you can troubleshoot them more effectively.
During troubleshooting, it can be very beneficial to test for the problem while envisioning the model. Is it the Physical layer, no, OK, is it the Data Link layer, no, is it the Network layerYES! This becomes so much more efficient and effectively that a haphazard, scattered approach to finding the problem.
So trust me, you are not just mastering this information for the exam with no good “real world” purpose. It will benefit you in your networking career directly.
The TCP/IP Model
I am not quite sure why this model is typically covered in courses after the OSI model receives its coverage...the TCP/IP model came first! Yes, that’s right, it predates the noted OSI model.
The TCP/IP model was fleshed out in the 1970s (yes, you read that right) by DARPA, an agency of the United Stated Department of Defense. You might see this model described as the Internet model or even the DoD Model.
The TCP/IP model was very loosely described as having layers. The description was certainly not as rigid and numbered like the OSI model. The layers are as follows:
The Link layer of this early model is concerned with the local network connection where you have the host attached. The obsession with TCP/IP was to make sure that the network communication of all of the stuff higher up in the model would work just fine no matter what the hardware networking technologies used underneath this Link layer. This layer is very concerned with things like the network card and software device drivers that power it. The Link layer of the early model became the Physical and Data Link layers of the OSI model.
The Internet layer is really concerned with host addressing and routing. This is where IP addressing is defined, and routing protocols are used to carry packets from source to destination across various nodes (routers). The Internet layer becomes the Network layer in the OSI model.
The Transport layer is concerned with things like end-to-end message transfer, error control, segmentation, flow control, congestion control and port numbers. It stays with the name Transport layer in the OSI model.
The Application layer of the TCP/IP model refers to the highest level protocols used by network applications. Examples include File Transfer Protocol (FTP), Simple Mail Transfer Protocols (SMTP), and Hypertext Transfer Protocol (HTTP). The Application layer becomes the Session, Presentation, and Application layers in the OSI model.
The OSI Model
The Open Systems Interconnection model (OSI) was created after the TCP/IP model was outlined. It was a product of the International Organization for Standardization.
This model, in contrast to the loosely defined TCP/IP model, features a VERY strict organization into numbered layers. As a CCNA, it is imperative that you memorize the layers and their order as a first step. These layers are as follows:
2. Data Link
There are two common mnemonics for memorizing these layers and their order. If you want to start from the top down, you can rememberAll People Seem To Need Data Processing. If you want to move from the bottom up, you can rememberPlease Do Not Throw Sausage Pizza Away. We instructors are stuck memorizing not only the layers, but both of the mnemonics.
Realize that the numbering of the layers is strict, like the names and the order. So quite often, network engineers will refer to the layers simply by number. You will hear things like “ahh, that is a Layer 3 issue” or “that is clearly a problem at Layer 2.”
Just as we described for the Application layer of the TCP/IP model, the Application layer of the OSI model refers to the highest level protocols used by network applications. Examples include File Transfer Protocol (FTP), Simple Mail Transfer Protocols (SMTP), and Hypertext Transfer Protocol (HTTP).
The Presentation layer is responsible for data representation. For example, JPG is a massively popular way to enjoy digital photographs. Ensuring this proper formatting for data is the stuff of the Presentation layer. The layer is also responsible for things like encryption and decryption.
The Session layer controls the “dialogue” between two computers. It helps to establish, manage, and terminate connections between the local and remote network application.
The Transport layer controls reliability through things like flow control, segmentation/desegmentation, and error control. Transmission Control Protocol (TCP) or User Datagram Protocol (UDP) can be used here in order to send information reliably, or unreliably.
We know from the TCP/IP model (where it was called the Internet layer) that this layer is concerned with host addressing and routing. This is where IP addressing is defined, and routing protocols are used to carry packets from source to destination across various nodes (routers).
2. Data Link
For the Data Link layer, it is important to provide the functional and procedural means to transfer data and make sure there are no errors in the Physical layer below.
For the Physical layer, it is all about the electrical and physical specifications for network devices. Here we have the relationship defined between a network device and a particular transmission medium. Important aspects are things like pins, voltages, cable specification, network adapters, etc.
Some People Fear Birthdays
There is indeed something else about these models that we should commit to memorythis is the name for the Protocol Data Unit (PDU) that is used at each of the bottom four layers. A PDU is a fancy name for the data and the encapsulated information that is prepended to the data.
Devices on the network send chunks of information between each other, and this information is interpreted by the corresponding OSI layer on the network device. Information at the top three layers of the OSI Reference Model (Application, Presentation, and Session) is simply referred to as Data. Once the data hits the Transport layer and receives its special protocol encapsulation, it gets a very special name. The name is Segment.
The information then makes it to the Network layer. Here the Protocol Data Unit is termed a Packet. Next we have the Data Link layer. The PDU gets the extra special name of Frame. Finally, the information makes its way to the Physical layer, where it is termed Bits.
How are we going to remember these special PDU names? And why is it important to remember them? Well, first of all, in order to ensure we remember them, I like to use a mnemonic that was invented by my good friend DJ Dennis back at KnowledgeNet. It is as follows:
Transport LayerSegments (SOME)
Network LayerPackets (PEOPLE)
Data Link LayerFrames (FEAR)
Physical Layer Bits (BIRTHDAYS)
Why is it so important to remember these special names? Well, we could certainly be quizzed on them during the CCENT (ICND1) exam. That is probably reason enough! But another important reason is the fact that in the maintenance and troubleshooting of networks, we want to refer to these names in order to give a great visual depiction of what we are referring to. For example, we describe a Layer 2 switch as processing a Frame, while we refer to a Multilayer switch as dealing with the Packet.