The OSI Reference Model

The Open System Interconnection reference model was created by the International Organization for Standardization (ISO) to provide a model that network vendors could follow when developing network products. The concept for the model was established to further support for interconnectivity between systems and to modularize components used in network communications.

The OSI model is composed of 7 interconnected layers, as detailed in the following table:

A good portion of the exam focuses on an thorough understanding of these layers and how various protocols and services may map to these layers. Spend some time considering a variety of applications and protocols and think about where on the model they may fall.

There are many benefits to having an industry accepted layered model. Some of these benefits, as listed in ICRC, include:

There are several mnemonics that have been devised to help remember this model. The most popular is a top to bottom approach:

All People Seem To Need Data Processing

Another popular choice, for learning the order from bottom to top is:

People Don't Need This Stuff Presented Anyway

But wait, there's more ...

Please Do Not Tthrow Sausage Pizza Away

Princess Di Never Tried Screwing Prince Andrew

Philosophers Deign Not To Solipstically Ponder Anachronisms

Philys Did Networking Till She Passed Away

From the top again ...

All People Sstanding Totally Nude Don't Perspire

Data Encapsulation

Each layer within the OSI model is primarily responsible for communicating with a peer layer on another machine. In other words, when two clients communicate, one layer, such as the Session Layer, on one client is primarily responsible with communicating with the exact same layer, in this case the Session Layer, on the other client.

This communication between peers is done in "Protocol Data Units" (PDU). The actual terms for the PDU changes from layer to layer. For instance, as the above table depicts, the Transport layers communicate via Segments.

Even though this communication logically takes place between peers, each layer is actually dependant upon the layers below it for the actual delivery. Each layer passes it's PDU to the layer beneath. The underlying layer then adds a header, creating it's own PDU. In this manner, the "data payload" for lower layers is the PDU of the layer immediately above.

This process of taking one PDU and enveloping it within another is called data encapsulation. This process is often likened to that of a mail delivery system. Consider for a moment what it takes to to deliver a piece of mail from one building to another. At the top of the "mail delivery model" you have the composition of the letter itself. The letter is then placed in an "envolope" that gets address to the recipient. The envelope then gets placed into a mail carrier's bag. The bag is then placed into a delivery truck and driven to the recipient building. At the destination building the bag is opened and the letter given to the appropriate mail carrier. The letter reaches the recipient who then opens the letter and finally reads the contents.

The CCNA test objectives reference 5 steps for data encapsulation. There are:

1. The application message is converted into data for communication on the internetwork.
2. The data from step 1 is given a segment header by the "transport" system.
3. The segment from step 2 is given a network header that includes source and destination logical addresses.
4. The packet or datagram from step 3 is given a frame header, specific to the physical medium that is being used.
5. The frame from step 3 is converted into a series of bits and interpreted as electrical (usually) signal generated on the physical medium.

The following diagram depicts the 5 step encapsulation process:

Step 1

Step 2

Step 3

Step 4

Step 5
Data is converted to bits and transferred...

011111111100011100101010010011100011100011100011010010