By John Oden
While writing a blog post the other day, I found myself guilty of talking about network layers with the above-noted assumption that everyone would know what I meant. The purpose of this post is to correct the error of my ways and to help you understand why it is important to understand the concept of layers as it relates to networking.
There are two primary reasons for organizing the idea of data networks into layers. First, the process of an application running on one computer communicating with an application running on another computer across a network is a complex process. Imagine everything that must happen from the time you open your Web browser and type an address into the address bar until the desired Web page appears on your screen.
This process involves multiple pieces of software and the operating systems on both host computers. It also involves the entire network infrastructure that connects the computers, including routers, switches, firewalls, telecommunication service providers, and the cabling media that connects it all together. Trying to understand and analyze all of this at once would be incomprehensible without some sort of framework to organize our thinking.
Organizing networking into layers, each dedicated to a particular set of functions, provides a framework that helps us learn, analyze, and discuss these concepts much more clearly than if we were to try to comprehend it in its entirety. A subsequent benefit of this organization is that troubleshooting is much more efficient.
The second primary reason for organizing networks into layers has to do with vendor interoperability. At present, it would be impossible to source every component needed to create a computer network from a single vendor. When networking is organized into layers, each layer acquires services from the layer below and exposes a consistent service interface to the layer above. This consistency is provided by standards and protocols that are used among vendors. Because of the consistent implementation of layers, we can replace a device that operates at a certain layer with a similar device from a different vendor without affecting other layers.
Figure 1 The ISO OSI Reference Model
Recognizing the need for a reference framework for computer networking, the International Organization for Standards (ISO) developed the Open Systems Interconnection (OSI) model shown in Figure 1. As the figure shows, this model encompasses everything from networking at its lowest level, which is the physical media that carries the electrical or optical signals, up to the highest level closest to the user. The highest level encompasses the protocols used by user-facing applications such as Web browsers and email programs.
Although there is only a single set of seven layers in the OSI model, Figure 1 shows two parallel sets of layers to represent one computer communicating with another. Each computer has its own set of layers. Keep in mind that this is a very simplistic network consisting of only two computers directly connected together. In any practical network implementation, there will be many devices of different types connected together. Typically, one group of computers would connect to another through a variety of intermediary networking devices such as switches, routers, and firewalls. Every device in the network possesses its own set of layers. In addition, data is characterized as it moves through these layers with a specific Protocol Data Unit (PDU) for each layer. These layers are often referred to by number. When you hear someone talking about Layer 1, they are discussing Physical-layer issues. When you hear Layer 2, that refers to the Data Link layer. Layer 3 refers to the Network layer, and so on.
I won’t go into an exhaustive discussion of each layer — there are many freely available resources that will give you all the details that you may need. However, let’s take a brief discussion of each layer and put it into perspective so that the next time you hear someone say that Device X operates at Layer Y, or that this problem is a Layer Z issue, you’ll have a better idea of what they mean.
The Physical layer defines that which can be seen and measured in the physical world. Things such as cabling standards, the arrangement of connector pins, and voltage levels fall within this layer. The physical and electrical portion of the Institute of Electrical and Electronics Engineers (IEEE) 802.3 Ethernet Standard such as 100BASE-TX Fast Ethernet is a common example of what is characterized in the Physical layer. The PDU at the Physical layer is the binary digit (bit). The Physical layer is often referred to as Layer 1. If you hear someone talking about a Layer 1 problem, they usually are referring to something related to the cables and connectors that link one device to another.
The Data Link layer deals with device addresses, called Media Access Control (MAC) addresses, which are burned in to the device at the time of manufacture; it also deals with the movement of data from one device to another device within the same network. The Data Link layer is often referred to as Layer 2. The most common PDU at Layer 2 in a local area network (LAN) is the frame. LAN switches are said to operate at Layer 2 because they build tables of Ethernet MAC addresses and use the source and destination MAC addresses in each frame to decide which switch port should be used to send the data toward its destination. When you hear Layer 2, think in terms of Ethernet switches and Ethernet configuration issues such as link speed and duplex settings.
Where the Data Link layer is concerned with moving data within a network, the Network layer is concerned with moving data from one network to another and determining the best path to follow when there are multiple potential paths to get to a distant network. The Network layer is often referred to as Layer 3, and the PDU at Layer 3 is the packet. By overwhelming majority, the most common protocol that operates at Layer 3 is the Internet Protocol (IP). IP addresses are logically assigned, either manually by the network administrator or automatically via one of several dynamic address assignment schemes. An IP address is a 32-bit binary number that is most often represent in dotted-decimal notation where the 32 bits are arranged into a series of four octets with each octet represented in decimal format. A typical IP address looks like this: 172.16.180.12. Network routers operate at Layer 3. When you hear someone referring to a problem at Layer 3, they are usually concerned with router configuration or the assignment and configuration of IP addresses.
The Transport layer is concerned with providing a set of services used by applications at higher layers to set up sessions between applications running on one computer with applications running on another. If you’ve ever marveled at the fact that you can have several Web browsers open on one computer with each browser displaying Web pages from different sites or even from the same site, you can thank the Transport layer, also referred to as Layer 4. Applications can request two different types of connections from the Transport layer: the Transport layer can provide either connection-oriented communication or connectionless communication. In the case of connection-oriented communications, the Layer 4 PDU is the segment. When connectionless communication occurs, the Layer 4 PDU is the datagram. The most common connection-oriented protocol is the Transmission Control Protocol (TCP), and the most common connectionless protocol is the User Datagram Protocol (UDP). The primary deciding factor between using TCP and UDP is error recovery vs. speed. Where integrity of the transmitted data is paramount, TCP is used because TCP contains a number of mechanisms for ensuring that the data received is exactly the same as the data sent. UDP has no such mechanisms, but it has lower overhead. It is interesting to note that TCP/IP is often incorrectly referred to as a protocol when in fact it is two separate protocols that are often used together.
This post is running a bit long, so I’m going to wrap up and finish this in Part II. Even though we’ve only discussed the first four of the seven layers, these four layers are the ones you’ll most often encounter in networking discussions, particularly at the ICND1 and ICND2 level. Next time we’ll discuss the Session, Presentation, and Application layers and talk about the encapsulation process that occurs when one host communicates with another. We’ll also learn that the OSI model isn’t the only model used to describe networks in terms of layers. Fortunately, the other model I’ll be describing in the next post is similar enough and has enough overlap with the OSI model that once you’ve learned the OSI model, the other one is a piece of cake!
If you’d like to learn more about the OSI model, here is a good resource that provides a lot of additional detail:
http://docwiki.cisco.com/wiki/Open_System_Interconnection_Protocols
Another resource you’ll want to know about is our Cisco Courseware from Boson. In addition to more complete coverage of the OSI model than I’ve provided here, this courseware covers all of the ICND1 and ICND2 topics you’ll need to master as you prepare for your CCNA certification.
See also: The Seven Layers of Networking Part II and The Seven Layers of Networking Part III.