27 Jan 12 Design Series – OSI Model

Enterprise Networking Defined

An enterprise network is a large network comprised of hundreds or thousands of workstations all connected with WAN circuits across many states and/or countries. The components are comprised of network devices, circuits, servers, applications, security and network management that connect employees for the purpose of supporting business processes. It is business processes that make companies profitable. The leveraging of new technologies to reduce costs and increase revenues is a business strategy that many companies have employed recently. That is contrasted with the ” ISP ” market, which refers to companies whose business it is to sell infrastructure, web hosting and application services to small, medium and enterprise companies. Many enterprise companies now rely on Internet Service Providers (ISP) for some or all of the mentioned components that comprise an enterprise network.

Usually all circuits are provisioned with the ISP. Some devices are ISP owned and the servers and applications can be all ISP owned if you have a hosting agreement with them that provides those services. Management and security of your network can be outsourced to an ISP as well, which is becoming popular with many Fortune 500 American Financial Corporations. Examples of that are AT&T and IBM, which offer outsourcing agreements with many different options.

OSI Model

The OSI model was developed years ago as a reference for network protocol and application designers to build their products with an open standard. That would promote a standard for developing network devices, interfaces, applications and services that would work together.

It is a model rather than a specification since it defines layers and services for those layers and how each layer works with the layer above and below it. The model is a representation of how a network operates much the same as a phone call. A phone call from your home to a particular destination requires a physical connection that electrically connects those two locations. When you take your phone off hook to make a call that is similar to Layer 1 physical and Layer 2 data link services. When you dial a number that is a Layer 3 network service or phone number routing. When someone answers the phone call that is similar to Layer 4 transport and Layer 5 session services with talking and listening. The specific language that you are speaking is Layer 6 presentation and Layer 7 application services. Today’s data networks use similar design principles.

Your desktop at work is connected to an application server over a physical media such as unshielded twisted pair (UTP) cable. That Layer 1 service is the electrical signaling from your desktop to a campus switch and from a campus switch to a server. Your desktop network interface card connects your desktop to the network cable and uses a Layer 2 data link protocol such as Ethernet that establishes a connection with the campus switch and defines when that desktop has access to the network. Before sending data, the desktop must know the address of the server to which it is sending data. That is a Layer 3 network service. The Layer 4 transport service establishes a logical connection between server and desktop for coordinating traffic flow. As well Layer 4 does error correction and re-transmission of packets if there are problems with data delivery.With the data connection established it is the Layer 5 session service that maintains a logical connection between application processes at the desktop and server. The formatting of the data for a particular application is a Layer 6 presentation service. Many applications will utilize their own formatting which is something you learn when you open files from different word processing vendors. The application interfaces with the OSI model at Layer 7, which is the application Layer. This level defines an API that applications use to develop their specific applications that will work with that layer.

The development of open industry standards at various Layers of the OSI model has contributed to the proliferation of multi-vendor network integration. Some examples of application Layer services are FTP, Telnet, SNMP and many applications that are used over a typical network. Microsoft has done a lot of work to standardize their Windows platform API so that third party application developers can write programs for that operating system. When you exam Windows 2003, for instance, you will notice that it provides services at Layers 3 – 7 of the OSI model. The argument could be made that Microsoft is involved with Layer 1 and 2 services as well since there are some software drivers for network interfaces that are available from them.

When describing client network software from Microsoft that runs at each workstation, it is describing all of the programs such as protocol stacks, network interface card drivers, and programs that integrate these elements. The network software is typically focused on Layer 1 to Layer 4 while application developers will focus on Layers 5 – 7 of the OSI model. It should be noted that the OSI model is a model which suggests that it provides guidelines for network developers. Some network software such as SNA doesn’t define a network Layer 3. The result is that it isn’t routable and a service such as encapsulation or tunneling must be used at the router for sending data across a network that is using a network layer and associated addressing. An example is integrating an SNA network, which doesn’t have a network Layer with a Windows network, which is using IP network services. 

The book Networking Technical Concepts is available at amazon.com