The “Seven-Layer” Vocabulary of Modern Internetworking

In the movies constituting our Ethernet series, and especially in the Internet movie entitled: “Ethernet Delivers the Internet,” we’ve shown how a message can originate anywhere in the world, can be sliced up and formatted inside IP packets, which can then be encapsulated and removed from various types of framing to traverse the entire world and end up at the doorstep of our local network. In a typical home-office or small-office environment, our local DSL modem or cable modem delivers that information to our local router inside an Ethernet frame, on a tiny little subnet managed and owned by our Internet service provider that only knows about one IP address for all of our computers. Other movies published at explain how our small local router asserts the additional intelligence necessary so that our local network can have more IP addresses and send that packet to one computer within our premises.

Experience has shown, however, that just getting the message to the right computer on our network is not sufficiently precise to allow real and useful work. Most computers today are controlled by “multitasking” or “multiprocessing” operating systems that are running several programs, or tasks, at any given moment. Even a typical small computer in a home is often running twenty or thirty different programs all at once, and somehow, our message from the Internet has to find the right program to receive and process it. Furthermore, most modern operating systems are oriented around a multi-user philosophy, where any of several people could be using the computer at any given time (sometimes, two or three people at once; sometimes just one). But in any event, work has to be associated with and attributed to a particular person among a group of people, and because different people have grown up in different cultures and speak different languages, reading and writing with different alphabets, it’s also important that our Internet message is somehow able to express itself in an appropriate character set and language.

So, stepping back and looking at this situation from the big picture demands that we allow a specific message, using a specific character set in a specific language, to be delivered or attributed to a specific person, using a specific program or process, on a specific computer inside our network, which is a subset of the world-wide Internet. This is a lot of selection and focus, and layer upon layer of intelligence.

At one end of this spectrum, the data represents real information in a useful format that will make real sense in a specific context to a specific person or a specific application; whereas, at the other end of the spectrum, the data just represents bits in transit from somewhere to somewhere else. In between these two extremes, various levels of intelligence can be individually identified, and perhaps interchanged with counterparts at the other end of the network at the same level of intelligence.

In the physical world, using the analogy of packages being transported by a postal service, we are already accustomed to thinking about counterparts at each end of the delivery, each performing similar functions. For example, the delivery van at the source end of a shipment has a counterpart delivery van at the destination. And the shipping company’s office at the source end of a shipment has a counterpart with a shipping company’s office at the destination end of a shipment.

Back in the early days of computer networking (in the 1970’s) computer networks were created in a proprietary fashion by a lot of different companies that used competing specifications, and they could not interoperate with each other at any level. Back in those days, if you bought your computers from IBM, you also bought your network from IBM. And your IBM computers really could not speak very effectively with computers from DEC, or Data General, or Hewlett Packard, etc. Attempts to make these competing network types interoperate sensibly resulted in a lot of frustration. Eventually, it became clear that these proprietary network systems would have to be abandoned, and would eventually be replaced by new network architectures built along an interoperable design.

In the late 1970’s, a group of influential vendors cooperating with standards bodies and with the involvement of governments throughout the world, formed an organization or joined forces with the International Standards Organization (ISO), and created a vocabulary, and a plan, and an architecture design for a new kind of networking. The design was called the Open Systems Interconnect (OSI), and the design divided the most common functions of networking into seven modular interoperable interchangeable layers. The open systems interconnect paper, published by the International Standards Organization back in 1977, provided the first really useful vocabulary to describe the logic of networks in a way that could be interpreted by many different vendors to allow interoperability. The paper became well-known as the “OSI Seven-Layer Architecture,” or the “OSI Seven Layer Reference Model for Network Design.”

As it turns out, this seven-layer design was highly regarded. It was received with enthusiasm by many different vendors and governments and customers, and it looked for a time as though all networks would eventually follow this seven-layer model. Unfortunately however, the ISO was burdened by a lot of bureaucracy. Although the OSI design was good, the actual implementation became increasingly complex and cumbersome and unwieldy, and by and large, the effort was entirely abandoned about 1996.

During that period of debate and transition, the Internet was growing and thriving, and the protocols used by the Internet turned out along similar lines to those used by the OSI, but the Internet protocols are simpler and easier to implement, and therefore became much more popular. Various bits of the OSI architecture survive today among the pieces of networks that do communicate, but by and large, the most important contribution of the OSI to networking is the vocabulary of networking and internetworking. Most network designers today still use the seven-layer vocabulary published by the OSI as they describe the components of today’s networking.

Because the OSI architecture divided the most common networking functions into seven hierarchical levels, it is commonplace today for network designers to describe their particular component as a layer-one component, or a layer-two component, or a layer-three component; all the way up to a layer-seven component, at increasing levels of intelligence. And the operations of our local area network in the Internet can be described pretty well in terms of the seven layers of the OSI model. You will be interacting with components in your small office network, or your home network that can be described very effectively with vocabulary of the OSI seven layer model.

Several of the networking movies published at have referred to the concept of encapsulation, and by now, viewers should be clear that one message can contain another message inside. This concept can be carried forward to several layers of messages within messages. And it is commonplace for the various layers of the ISO model to be represented as a package inside a package, or a message inside a message at successively larger sizes at the various layers as they encapsulate one another.

To help you visualize this concept in the context of the Internet and a local area network, we might go so far as to illustrate several envelopes inside envelopes, inside one another with a small envelope on the right inserted into successively larger envelopes layer by layer, and finally into a very large envelope at the left. We could label each of those envelopes with the layers associated with a particular OSI level of intelligence, and also with the kind of functionality performed by the Internet at that corresponding layer. This allows us to map the real exercise of the Internet with the vocabulary of the OSI in a way that we can visualize with our eyes to make it memorable.

Accordingly, let me draw your attention to these images showing a series of envelopes. At the large end, I have wrapped the envelopes in an Ethernet cable, representing layer one (the “physical” layer). The physical layer is well-represented by an Ethernet cable. Inner layers are encapsulated inside outer layers in transit, and removed from those outer layers upon delivery. Ultimately, the destination application layer receives the same information that was originally sent at the source application layer. And incidentally, at each intermediate layer in between, the destination side receives the same information originally input at the source side: layer by layer, match by match, which makes the layers tend to be interchangeable, and this in turn makes our network and its applications far easier to write, and more interoperable.

This is the text from the movie entitled "7-LayerVocabulary of Modern Internetworking", published at (The text also includes a few small, static illustrations from the movie). You can learn more about the video version and purchase an inexpensive personal license to view it by clicking right here. You can learn more about our "Internet Series" of licensed videos by clicking here. Our "Internet Series" is part of a larger set of "Networking" videos, and you can learn about all of them by clicking here.