By Ed Taylor
Understand This First!
This chapter provides background information necessary before exploring Transmission Control Protocol/Internet Protocol (TCP/IP). Questions such as: Why Do I Need a Network?, Why Do I Need a Protocol?, What Does a Protocol Do?, Which Protocol Do I Need?, How are Networks Built?, and What is a Backbone? are presented in this chapter.
1.1 Why Do I Need A Network?
Excellent question. If you ask ten people this question, you will probably get ten different answers. Some consensus exists, however. Ask this question to financial people or managers in corporations and they may respond by saying it will help them maximize technical resources within their company. Networking can do this if properly implemented. Ask a documentation or training department the same question and the response may be entirely different. A typical response from such a department might be something like, “It would enable all workers in the departments to exchange files, have electronic mail, and have remote logon access to hosts not located on their desks.”
There are reasons for having a network. Commonalities exist among most networks, but differences also exist. Three common functions in any given network are the following:
Remote Logon. This service permits a user on his/her host to log on ‘‘remotely’’ to a host in a different location.
File Transfer. This service permits network users to exchange files. It saves time and can eliminate duplication of resources. And, most of all, it is convenient for users.
Electronic Mail. This service allows all users on the network to exchange mail electronically. The idea and function is similar to the current mail system except it is paperless, cost effective, and efficient.
Networks can (and usually do) provide services and features beyond these mentioned previously. The needs of the user and the type network implemented determine what is available. A particular advantage networks offer is that some networks support different vendor equipment, thus providing interoperability between unlike equipment. This alone is reason enough to install a network wherein lies heterogeneous equipment.
1.2 Why Do I Need a Protocol?
To connect computers, printers, disk drives, terminal servers, communication servers, and other devices requires some form of a network. For a network to function, rules and regulations must be followed. In the technical community, these rules and regulations are called protocols. Transmission Control Protocol/Internet Protocol (TCP/IP) is a network protocol that has rules and regulations which permit different vendor equipment to interoperate. This is a powerful statement. Examples will be provided later, but for now it means if a network is based upon TCP/IP, a DEC computer can communicate with an IBM, Convex, Apple, SUN, Unisys, or just about any vendor’s equipment.
In network terminology, these rules and regulations are referred to as network protocols. Network protocols define various aspects of user operations. For example, a network protocol defines how a user performs a remote logon, file transfer, or electronic mail.
1.3 What Does a Network Protocol Do?
A network protocol defines all operations within a network. Protocols even define how entities outside the network must interact with the network. For example, some network protocols define how data gets from point A to point B. Other network protocols define how a computer or device communicates over a particular medium, like a telephone line or other type connection. Simply put, protocols define how things are to be done if a device is going to operate in a network.
1.4 Which Network Protocol Do I Need?
To do this a point of reference is needed. A good reference point is a model of what constituent parts should exist in networks. A standards making body called the International Standard Organization has what they call the Open Systems Interconnection (OSI) model.
This OSI model defines that which should exist in any network. The OSI model used here will be a reference point to explain basic aspects about network protocols. The OSI model consists of seven (7) layers. To better understand this, picture a cake with seven (7) layers and envision the cake cut in half; the seven layers would be identifiable. The OSI model is similar to the seven-layered cake cut in half. OSI network layers have names and perform specific functions. This model, including the layers and their names, is identified in figure 1.1.
Before we examine what each layer does, consider this. Envision a network consisting of computers, software, cables, and everything that goes into making a network. Most networks can be divided into layers.
Network layers can be explained in accordance with their function. Usually, there is not a one-to-one correspondence when attempting to explain different networks by layers. Many network protocols do not appear like the OSI model.
A layer synopsis from top to bottom includes:
Application This layer provides services software applications require. For example, it provides services necessary for a file transfer program to operate. It is called the application layer because it works with or is a provider of services to applications (in certain network protocols).
Presentation This layer determines data syntax. In short, whether data is ASCII or EBCDIC is determined here. This layer performs encoding values that represent data types being transferred.
Session This layer is considered the user~ s interface into the network; however, the user is not aware of it. This layer is where logical connections are made with applications. The session layer has addressable end points that relate to programs or a user.
Transport On the sending node in a network the transport layer takes data from the session layer and puts a header and trailer around the data itself. Some transport protocols ensure the data arrives correctly at the destination; this type protocol is connection oriented. Conversely, connectionless oriented protocols do not ensure this. On the receiving node the transport layer removes the header and trailer and passes the data to the session layer.
Network This layer routes data from one location to another (source to destination). The network protocol in use determines how this layer works. In the case of TCP/IP this is Internet Protocol (IP).
Data Link The main goal of the data link layer is to provide reliable data transfer across a physical link. This layer puts data into frames, transmits these frames sequentially, and ensures they have been received in order by the target host.
Physical This layer is an interface between the medium and the device. This layer transmits bits (ones and zeros). Specifically, it transmits voltage or light pulses.
Different network protocols can be evaluated with the OSI model serving as a baseline. OSI itself is a network protocol, but the focus here is TCP/IP. The OSI model will be used later in the book to explore further aspects of TCP/IP.
1.5 Data Flow Through a Network
In a network data flows from the sending node from top to bottom (with respect to layers in the network). This means that as data passes down the network protocol stack, headers and trailers are added to the data, at each layer. Likewise in a network the receiving node’s data flows from bottom to top (with respect to the network layers). These headers and trailers are removed by layer as the data moves up the protocol stack.
In some cases the sending node is a terminal user and the receiving entity is an application program. In other cases both sending and receiving entities can be application programs.
These headers and trailers wrapped around the data include information particular to the needs of a specific layer. For example, the network layer header includes routing information. Consider figure 1.2 depicting the OSI model and how headers are added to data as it passes down the 051 protocol stack.
PH AH DATA
SH PH AH DATA
TH SH PH AH DATA
1.6 How Are Networks Built?
It is important to know how networks are built because this provides a framework for discussions about a network. One way to explain how networks are built is to explore how network devices are physically connected to a common medium. The idea of a common medium is fundamental to many networks. Granted this common medium may span great distances and be comprised of “different” types of media; collectively the medium can be considered as a whole. Networks can consist of devices such as computers, printers, servers, etc. connected together. These devices connect to the medium directly or indirectly. The medium may consist of different types, but the common thread is these devices connect to the medium, thus a network. They may be spread out physically around the world. In the technical community, the physical layout of a network has a technical term associated with it called a topology. A number of topologies exist, but our focus here will include:
In addition to the topologies mentioned above, two terms frequently used in discussions about networks need to be understood. They are presented after explanation of the Bus topology.
A Bus based network can best be understood by analogy. Think of a Bus as a street. In this case the Bus is a cable that serves as an access point for all network devices. It is similar to a street because each house on any given street has access to the street. Consider figure 1.3 depicting a Bus with computers, printers, file servers, and communication servers.
Figure 1.3 shows a straight line as the common link between all participating network devices. Notice figure 1.3 displays the Bus as a straight line. In reality this is generally not the case. Because the Bus is a cable, it usually gets shaped to fit the physical environment where it is installed.
One example of a Bus topology is where network devices attach to the cable via a transceiver. The transceiver serves as a connection point for network devices. Transceivers do more than serve as a connection point, but this is not the focus here. These transceivers have a cable that connects them to the network device interface card. This cable is typically called a drop cable. Review figure 1.3.
A Bus could be considered a data highway. It is the medium where data is passed from source to destination. Devices attached to the Bus can access it and send or receive data. In a very real sense it is a data highway.
The two terms needing understanding in light of topologies are logical and physical. Figure 1.3 shows a logical example of a hypothetical Bus network. It is considered a “logical” example because in real life the majority of Bus networks are not straight lines. If figure 1.3 were a real
implementation, the line depicting the Bus would likely be very crooked. Think about it. If a Bus is the main cable network devices connect to, it stands to reason the cable itself is probably between walls, in the ceiling, possibly under floors, and twisted about in other ways. The last thing it will be is straight.
The term “physical” is generally used to reflect the actual implementation (how it may appear). The next topology is a good example of how the terms “logical” and “physical” are used. Of course using these terms to explain a topology is a generalization, and it would be incorrect to extrapolate conclusions beyond that which is stated.
The terms “logical” and “physical” are used far beyond the bounds of network topologies. They are used in a variety of explanations, however our focus here is upon topologies.
When a Ring topology is mentioned, token ring may come to mind. Token ring is a protocol (way of passing data) at lower layers within a network, specifically the data link layer. Most pictures show a token ring network like figure 1.4.
Figure 1.4 is a “logical” example of a network based on a ring topology. It does not depict how a token ring network appears physically. A token ring network is built around a device called a Media (some call it a Medium) Access Unit (MAU).
If one goes looking for a token ring network, figure 1.5 is an example of what will be found.
Media Access Unit
The MAU has a ring inside, hence it is considered a ring based network. Unfortunately, many diagrams and explanations depicting a ring network either assume this knowledge on behalf of the reader, or for whatever reason it is omitted.
This could be funny. Can you imagine someone new to token ring networks and asked to isolate a problem with a token ring network. If nobody told the individual there is no visible “ring” to be found (outside of disassembling the MAU) they could look for days! Believe it or not, I have witnessed this.
Other types of ring based networks exist, however they are based on the same fundamental premise. They have a ring (or two) used to pass data. The ring topology uses a cable in a ring fashion and serves as the data highway for data to get from source to destination.
A hub is like its name connotes. It is a central point of connection. Figure 1.6 shows how a hub topology appears.
A hub is the central point of the network from a physical connection. In many cases certain type networks that employ a hub topology are easily implemented and maintained. Hub popularity has increased over the past few years, and as a result certain hubs are becoming inexpensive.
1.7 What Is a Backbone?
The term backbone is used frequently in conversations about networking. It means different things depending upon the context and point trying to be made. For example, the term backbone could be used in light of the topology of a network. If such were the case, the meaning conveyed is the physical connections and primarily the common point of connection for devices attached to the network.
The term backbone is also used to refer to a network protocol. When this is the case, a larger concept is usually conveyed. An example of this could be someone using the term to refer to routing from network A to network B through an unlike network, say network C. An example of this is figure 1.7.
The term backbone can be most confusing, especially if the speaker or writer does not communicate precisely the intended meaning. The
term backbone is not a descriptive technical term!
Networks can be a valued addition to a corporation or even a small company. The type of network chosen will dictate features and functions available for users, programmers, and others accessing the network.
Networks are comprised of protocols. Network protocols themselves are rules defining how things will be done such as remote logons, file transfers, and electronic mail for example.
The protocol chosen for a network should be based upon the one that best meets users’ needs. Consideration for interoperability among different vendor equipment should be taken into consideration also. Other issues may need evaluation, and each site should be able to define their own needs.
The physical layout of a network can vary. The site can dictate this to some degree. But, many protocols dictate which physical arrangement must be used with network implementations.
The difference between the physical layout of a network and the logical implementation can be confusing, but it is nevertheless important. The term backbone should be understood in light of how vendors use it to explain their equipment. A variety of meanings are currently associated with the term, and as a result confusion abounds.
Getting acclimated to computer networks is half the battle. The remainder of the battle is the constant challenge to remain current, understanding the technology used and changes to existing equipment. The pace of change in network technology now is greater than when I began (or I suppose I could be slowing down).
Now more than ever companies are harnessing the power in networks to leverage company resources. On a recent trip I realized the 1980s witnessed an unprecedented expansion in technology and development of products. And as surely as this was the case, so the 1990s will be the decade of learning and dissemination of knowledge about this technology.
The above article was taken as excerpts from Ed Taylor's book titled Demystifying Tcp/Ip (ISBN 1-55622-400-1), chapter 1. Ed is a consultant, lecturer and author specializing in network architecture and integration.