TCP/IP First-Step [Electronic resources] نسخه متنی

Speaking of Political Committees…

Interoperable products requires companies to agree on the technical specifications for those products. In other words, a consensus must be reached on important design criteria, such as what the product will do, how it will do it, what it won't do, and so on. Developing such a consensus amongst competing companies isn't easy and, more often than not, requires a neutral third-party to mediate or judge during the consensus-building process.

The world, it seems, is full of neutral third-party organizations whose business is to standardize all kinds of things. Rather than bore you to tears with an exhaustive review of all the standards bodies, the discussion is limited to just three that set standards for data networking and communications technologies. You look at three organizations:

Electronics Industry Association and the Telecommunications Industry Association (EIA/TIA)

Institute of Electrical and Electronic Engineers (IEEE)

Internet Engineering Task Force (IETF)

Many other organizations are chartered to set and maintain technical standards, some of which focus on networking technologies. This chapter focuses on these three.

EIA/TIA

One common set of technical standards is attributed to the EIA/TIA. These important standards focus on the physical wires that build networks and carry data between machines connected to networks. The acronym EIA/TIA actually stands for two different standards organizations: the Electronics Industry Alliance (EIA) and Telecommunications Industry Association (TIA). Thus, standards from the EIA/TIA are endorsed by both organizations.

This is not uncommon. Quite frequently, two or more standards organizations see a technology as being their responsibility. Such overlap usually results in cooperation, rather than competition, for the development of that standard. Sometimes that cooperation means working together to actively develop a standard; other times it means one group does the work and the other endorses its output.

The EIA/TIA standards for network cabling are interesting in that they do not specify physical standards! Instead, they standardize a set of performance levels that cables must meet or exceed. Most technology standards spell out, in minute detail, everything about how a technology should be built, how it will work, and all its specific attributes and capabilities. However, this is just a piece of wire; there's not much to it! You shouldn't care about what it looks like, or what kind of metal it is, or even how thick that metal might be. All you need to know is that it meets your needs. Consequently, the EIA/TIA settled on standardizing performance levels as opposed to more complex physical standards that define the wires' physical properties.

The EIA/TIA performance levels, known more properly as categories of performance, are numbered 1 through 6. The higher the category number, the higher the performance capability. Performance capabilities are defined in terms of how fast can data be transmitted and for what distance. The most commonly used cable these days is known as Cat-5, which is an abbreviation of Category 5. Cables that comply with the EIA/TIA performance Category 5 support transmissions of up to 100 megabits per second for up to 100 meters. Cables must meet this minimum capability set before they can be advertised and sold as Cat-5 cables.

For more information on the EIA, please visit its website at http://www.eia.org. Additional information about the TIA and its standards activities can be found at http://www.tiaonline.org.

IEEE

The Institute of Electrical and Electronic Engineers (IEEE or I Triple E, as it is more commonly pronounced) is an independent, nonprofit organization that sponsors research and establishes the technical standards for a wide variety of electrical and electronic technologies. The IEEE's contributions to the data networking industry are numerous, but the single most significant was the research project it launched in February of 1980. That project, known formally as Project 802 in deference to the year and month it launched, defined the technical standards for local-area networks (LANs). That research initiative resulted in the standardization of numerous LAN technologies, including Ethernet and Token Ring (among other, less recognizable ones).

Ethernet was a particular standardization priority. This LAN technology was developed by scientists at Xerox Corporation's Palo Alto Research Center (PARC) as a way of sharing an expensive new laser printer. Over time, the value of the LAN they created became apparent, and steps were taken to market it commercially. These steps resulted in a couple of Ethernet variations.

Though each variation could legitimately claim to be Ethernet, they were not necessarily compatible. Consequently, the IEEE set out to create an open standard version of Ethernet, as well as an open standard version of Token Ring. The bonus of using a layered model to organize their efforts: Interoperability between Ethernet and Token Ring also became possible.

To create technical standards for all the LANs popular at that time, the IEEE launched several engineer teams, each asked to develop the standards for a specific task. Recognizing that certain functions were common to any LAN, the IEEE sought to avoid duplicating the wheel. The first thing they did was identify all those logical stepsand their sequencenecessary in any LAN. This became the foundation for the IEEE's 802 LAN reference model.

Some of these basic steps and functions included a system for end-point addressing and tools for monitoring and managing the LAN. With the foundation built, all that was left was chartering different teams to develop the rest of the LANs that needed standardization.

To help correlate activities between the various engineering teams, the IEEE developed a reference model. This model helped everyone understand the boundaries of certain functions and kept everything organized. Figure 3.4 shows the IEEE's reference model.

The Project 802 reference model and the steps it organized became the foundation for all the IEEE's LAN standards within Project 802. Each of the LANs standardized within that initiative, including Ethernet (known as IEEE 802.3) and Token Ring (IEEE 802.5), shared this common foundation.

It is important to note that although the IEEE developed protocol standards for sending data out over several types of physical wires, it did not develop standards for those wires. That distinction is subtle, but important. Ethernet and Token Ring needed to support transmission over wires, so they relied on the work of other standards bodies, such as the EIA and TIA.

For more information on the IEEE, you can visit its website: http://www.ieee.org.

IETF

The Internet Engineering Task Force (IETF) is responsible for developing and maintaining the Internet's technologies, as well as for guiding its growth and development. A short list of their achievements include the definition of TCP/IP, IP version 6, Simple Mail Transfer Protocol (SMTP)which is the foundation for literally exchanging e-mail between open systemsand many other equally significant technologies.

The IETF is composed of engineers who volunteer their time to the various development efforts constantly underway at the IETF. Typically, a working group is chartered to perform a specific task. Numerous working groups might be launched to tackle related tasks. In that manner, a large and complex project can be completed more quickly, because of parallel work on smaller pieces.

Literally everything the IETF does is in the open; every participant in each working group has an equal vote in the proceedings and recommendations. Consensus is the mechanism by which standards are developed and accepted. Plus, consensus and request for input aren't limited to select members of any given working group. You see, all work is documented in a publicly accessible document. These documents, known collectively as Requests For Comments (RFCs), are available without charge to anyone via the Internet.

The phrase Request For Comments is a bit misleading. Not all RFCs are requests for comments in the literal sense. Some are little more than wild ideas an individual created and then posted for the world to see. In the spirit of openness, anyone can write and publish an RFC for the Internet community's consideration. Most RFCs are created by a working group that the IETF sponsored to solve a specific problem. Such RFCs probably carry the weight of a technical standard someday, although it might take a while for them to be accepted.

Some RFCs are jokes. It has become tradition for an outrageous technical proposal to be published on April 1 each year. To the uninitiated, these might at first appear serious, but the date should be a clear indication that someone is trying to have fun on April Fool's Day. To learn more about the IETF, please visit its website at http://www.ietf.org. To see the many RFCs that spell out the rules for virtually every aspect of TCP/IP and the Internet, please visit the RFC page on the IETF website.

Each RFC is serially numbered in the order it was published. E-mail, for example, came to life through RFC 821 in 1980. That document has long since been made obsolete by much more complicated and feature-rich versions of e-mail. Yet, you can see it at http://www.ietf.org/rfc/rfc0821.txt. Substituting any four-digit number for the 0821 in the preceding URL lets you browse that RFC.

RFCs are, quite literally, the source documents in which all the Internet's technical details are published. The vast majority aren't fun to read, but they are available. That's just how open the IETF is when it comes to enabling interoperability via the Internet!

Layers of Layered Standards

Having just seen a small sample of the standards bodies that organize data networking and communications technologies, you might have noticed that they focus on different aspects of network-based communications. The EIA and TIA standards are limited to the wires carrying electronic signals. The IEEE is best known for its work in standardizing Ethernet and other LAN technologies. Lastly, the IETF focuses on the Internet and all the technologies that let you communicate between LANs. Communicating from one LAN to another LAN is known as internetworking, and TCP/IP has become the world's preferred internetworking protocol.

The work of these three different standards bodies do not overlap; each focuses on a distinct area. However, their work is highly complementary! All three sets of standards are necessary to all network functions, including Internet functions. Assume you use at least three sets of open standards when you are online and that those standards always work well. The standards' creators logically, then, can keep things coordinated between themselves.

You saw the IEEE reference model for Ethernet and took a look at the IETF reference model for TCP/IP. The two models bear little, if any, resemblance to each other. Yet the two work so well together that IP over Ethernet has become the de facto standard for LANs around the world, including large enterprise networks and home networks.

How do you make sure two technologies, developed separately, continue to interoperate well over time? The answer is simple: by using a neutral, third reference model! The reference model most frequently used in this capacity is the OSI reference model.

OSI Reference Model

Sometimes, the alphabet soup-like atmosphere of the information technology industry gets carried to a ridiculous extreme. This section's reference model and the standards body that developed it form a marvelous example of that extreme.

The International Organization for Standardization (ISO) was chartered by the United Nations and founded in 1946. Its mission is to set global standards for virtually everything. Everything, that is, except for anything electrical or electronic. ISO developed a generic model for the interconnection of open systems. That model is known as the Open Systems Interconnection (OSI) reference model. This model contains seven layers that encompass every aspect of communication between networked computers.

Although it doesn't describe a particular product or technology, the OSI reference model has become the de facto standard means of correlating other, technology-specific reference models and for ensuring interoperability between open computer systems. For this reason, it has become the single most frequently encountered reference model. The problem is that though it is overused, it is seldom adequately explained. Consequently, most people who encounter it for the first time usually walk away, scratching their heads and wondering what it is all about!

ISO's OSI (how's that for alphabet soup?) reference model is shown in Figure 3-5.

iso means equal or standard. That seems much more fitting for an international organization.

You might be wondering how a reference model for connecting computer systems isn't a violation of the ISO charter. Remember: Things electrical or electronic are not within their domain. (Yet connecting computer systems seems to require electricity!) The answer is that the model describes a standardized process, but not any particular electronic or electrical product. session. The session layer describes all the mechanisms and processes needed to manage a session.

4

Transport layer

The next set of necessary functions manages the data sent and received in a session. When sending data, this includes breaking received data into smaller pieces for transmission and uniquely numbering each piece so the recipient knows how reassemble them. When receiving data, this set of steps includes making sure the data arrives intact (not damaged) and then putting everything together in its original order before handing the data off to the session layer.

3

Network layer

A communications session doesn't necessarily always occur between two computers on the same network. Sometimes, those computers are literally half a world away from each other. In such cases, the network layer contains the mechanisms that map out the best route for that session.

2

Data link layer

The data link layer is where the rules, processes, and mechanisms for sending and receiving data over a LAN are defined.

1

Physical layer

This layer includes all the procedures and mechanism you need to both place data onto the network's wire for transmission and to receive data sent to you on that same wire.

Note that, like your model for driving a car, these function layers work in both directions. In this case, those two directions are sending and receiving data.

Using the OSI Reference Model

The best way to understand the OSI reference model is to remember that it is a framework that establishes the sequence of events that interconnect two open computer systems. No single product or technology fulfills the requirements of all seven layers! Instead, you tend to find that communications between open systems is more a patchwork between at least four different sets of open-standard technologies. Consequently, the OSI reference model serves marvelously as a neutral frame of reference for correlating two or more dissimilar technologies.

Take a look at Figure 3-6, which shows the OSI model correlating the functions of the EIA/TIA, IEEE, and IETF open standards.

In Figure 3-6, you can see that any good piece of software, such as e-mail, takes care of the functions laid out in Layers 5, 6, and 7. The IETF standards for TCP/IP typically take care of Layers 3 and 4. Layer 4the transport layerprovides the logic for accepting data from applications (a higher level of functionality) and preparing that data for transport across a network. Layer 3, the network layer, provides the mechanisms for finding and accessing a remote computer system across a network.

TCP goes beyond just being a Layer 4 technology: It encompasses many small utilities that span up through Layer 6. A visual representation of this can be misleadingtruthful, but misleading. TCP/IP focuses on Layers 3 and 4 of the OSI model. You use it to support other products that span Layers 57. Such products are known generically as application software, and they can include anything from e-mail and browsers to chat or instant messaging. All need TCP/IP for their Layer 3 and 4 mechanisms.

Figure 3-6 also shows how the IEEE's Ethernet standard occupies Layers 1 and 2 (physical and data link layers, respectively). This illustration does not make clear why the EIA/TIA standards are not considered part of the physical layer. The answer lies hidden in the fact that this is a logical model onlyit does not describe a physical product or technology. Thus, what is described in the physical layer isn't the actual wire. Instead, the physical layer is limited to the processes and mechanisms required to place data on that wire and to receive data on that wire. This includes the wire's required performance category, the physical connector at its ends (such as the modular jack at the end of your telephone cord), but does not include a physical description of the wire.

Put Layers 1 and 2 together with the physical wire and you have a complete system capable of reliably providing communications between open computer systems. This is possible only because the clearly defined steps create a logical and well-understood sequence.

What It Looks Like

One of the neat things about having a reference model so well defined is that hardware and software manufacturers can focus on what they are good at, entrusting the remaining functions to someone else. The net effect is that a software developer that makes e-mail software for PC users can focus on e-mail features and functions, without worrying much about the other steps necessary for two users to communicate using e-mail.

From that software developer's perspective, as well as from the user's perspective, the two e-mail software packages appear to communicate directly with each other. This concept is known as logical adjacency. Figure 3-7 shows how logical adjacency works. For the sake of example, assume your application software (Layer 7 on the model) is a PC e-mail software package such as Microsoft Outlook.

In this illustration, John is sending an e-mail to Jane. Each enjoys the perception that the e-mail is sent directly from the e-mail software on his PC to the e-mail software on her PC. You know that many other steps lie between sending and receiving the e-mail. However, that perception feels real. The two e-mail packages appear to be communicating directly with each other.

What It Really Does

What happens when John sends Jane an e-mail? A whole series of lower-level mechanisms in TCP/IP and Ethernet actually send the e-mail. Jane's PC performs all the same steps as John's PC, only in reverse. This is shown for you in Figure 3-8.

Each layer of functions occurs in the order necessary for everything to work. Anyone who has ever sent an e-mail over the Internet knows just how smoothly it goes! You quite literally take the underlying mechanisms for granted because they work so reliably and consistently.