Figuring out how TCP/IP lines up with the OSI Reference Model
Discovering important TCP/IP applications
Many years ago, Transmission Control Protocol/Internet Protocol (TCP/IP) was known primarily as the protocol of the Internet. The biggest challenge of getting a local area network (LAN) connected to the Internet was figuring out how to mesh TCP/IP with the proprietary protocols that were the basis of the LANs — most notably Internetwork Packet Exchange/Sequenced Packet Exchange (IPX/SPX) and NetBIOS Extended User Interface (NetBEUI).
But then, some years ago, network administrators realized that they could save the trouble of combining TCP/IP with IPX/SPX and NetBEUI by eliminating IPX/SPX and NetBEUI from the equation altogether. As a result, TCP/IP is not just the protocol of the Internet now, but it’s also the protocol on which most LANs are based.
This chapter is a gentle introduction to the Internet in general and the TCP/IP suite of protocols in particular. After I get the introductions out of the way, you’ll be able to focus more in-depth on the detailed TCP/IP information given in the remaining chapters of Book 3.
What Is the Internet?
The Goliath of all computer networks, the Internet links hundreds of millions of computer users throughout the world. Strictly speaking, the Internet is a network of networks. It consists of hundreds of thousands of separate computer networks, all interlinked, so that a user on any of those networks can reach out and potentially touch a user on any of the other networks. This network of networks connects more than a billion computers to each other. (That’s right, billion with a b.)
One of the official documents (RFC 2026) of the Internet Engineering Task Force (IETF) defines the Internet as “a loosely organized international collaboration of autonomous, interconnected networks.” Broken down piece by piece, this definition encompasses several key aspects of what the Internet is:
Loosely organized: No single organization has authority over the Internet. As a result, the Internet is not highly organized. Online services, such as America Online or MSN, are owned and operated by individual companies that control exactly what content appears on the service and what software can be used with the service. No one exercises that kind of control over the Internet. As a result, you can find just about any kind of material imaginable on the Internet. No one guarantees the accuracy of information that you find on the Internet, so you have to be careful as you work your way through the labyrinth.
International: Nearly 200 countries are represented on the Internet, from Afghanistan to Zimbabwe.
Collaboration: The Internet exists only because many different organizations cooperate to provide the services and support needed to sustain it. For example, much of the software that drives the Internet is open source software that’s developed collaboratively by programmers throughout the world, who constantly work to improve the code.
Autonomous: The Internet community respects that organizations that join the Internet are free to make their own decisions about how they configure and operate their networks. Although legal issues sometimes boil up, for the most part, each player on the Internet operates independently.
Interconnected: The whole key to the Internet is the concept of interconnection, which uses standard protocols that enable networks to communicate with each other. Without the interconnection provided by the TCP/IP protocol, the Internet would not exist.
Networks: The Internet would be completely unmanageable if it consisted of half a billion individual users, all interconnected. That’s why the Internet is often described as a network of networks. Most individual users on the Internet don’t access the Internet directly. Instead, they access the Internet indirectly through another network, which may be a LAN in a business or academic environment, or a dialup or broadband network provided by an Internet service provider (ISP). In each case, however, the users of the local network access the Internet via a gateway IP router.
The Internet is composed of several distinct types of networks: Government agencies, such as the Library of Congress and the White House; military sites (did you ever see War Games or any of the Terminator movies?); educational institutions, such as universities and colleges (and their libraries); businesses, such as Microsoft and IBM; ISPs, which allow individuals to access the Internet; and commercial online services, such as America Online and MSN.
A Little Internet History
The Internet has a fascinating history, if such things interest you. There’s no particular reason why you should be interested in such things, of course, except that a superficial understanding of how the Internet got started may help you to understand and cope with the way this massive computer network exists today. So here goes.
The Internet traces its beginnings back to a small network called ARPANET, built by the Department of Defense in 1969 to link defense installations. ARPANET soon expanded to include not only defense installations but universities as well. In the 1970s, ARPANET was split into two networks: one for military use (renamed MILNET) and the original ARPANET (for nonmilitary use). The two networks were connected by a networking link called IP — the Internet protocol — so called because it allowed communication between two networks.
The good folks who designed IP had the foresight to realize that soon, more than two networks would want to be connected. In fact, they left room for tens of thousands of networks to join the game, which is a good thing because it wasn’t long before the Internet began to take off.
By the mid-1980s, ARPANET was beginning to reach the limits of what it could do. Enter the National Science Foundation (NSF), which set up a nationwide network designed to provide access to huge supercomputers, those monolithic computers used to discover new prime numbers and calculate the orbits of distant galaxies. The supercomputers were never put to much use, but the network that was put together to support the supercomputers — NSFNET — was used. In fact, NSFNET replaced ARPANET as the new backbone for the Internet.
Then, out of the blue, it seemed as if the whole world became interested in the Internet. Stories about it appeared in Time and Newsweek. Any company that had “dot com” in its name practically doubled in value every month. Al Gore claimed he invented the Internet. The Net began to grow so fast that even NSFNET couldn’t keep up, so private commercial networks got into the game. The size of the Internet nearly doubled every year for most of the 1990s. Then, in the first few years of the millennium, the growth rate slowed a bit. However, the Internet still seems to be growing at the phenomenal rate of about 30 to 50 percent per year, and who knows how long this dizzying rate of growth will continue.
TCP/IP Standards and RFCs
The TCP/IP protocol standards that define how the Internet works are managed by the IETF. However, the IETF doesn’t impose standards. Instead, it simply oversees the process by which ideas are developed into agreed-upon standards.
An Internet standard is published in the Request for Comments (RFC) document. When a document is accepted for publication, it is assigned an RFC number by the IETF. The RFC is then published. After it’s published, an RFC is never changed. If a standard is enhanced, the enhancement is covered in a separate RFC.
Thousands of RFCs are available from the IETF website (www.ietf.org). The oldest RFC is RFC 0001, published in April 1969. It describes how the host computers communicated with each other in the original ARPANET. The most recent proposed standard (as of January 2018) is RFC 8311, entitled “Relaxing Restrictions on Explicit Congestion Notification (ECN) Experimentation.”
Not all RFCs represent Internet standards. The following paragraphs summarize the various types of RFC documents:
Internet Standards Track: This type of RFC represents an Internet standard. Standards Track RFCs have one of three maturity levels, as described in Table 2-1. An RFC enters circulation with Proposed Standard status but may be elevated to Draft Standard status — and, ultimately, to Internet Standard status.
Experimental specifications: These are a result of research or development efforts. They’re not intended to be standards, but the information they contain may be of use to the Internet community.
Informational specifications: These simply provide general information for the Internet community.
Historic specifications: These RFCs have been superseded by a more recent RFC and are thus considered obsolete.
Best Current Practice (BCP): RFCs are documents that summarize the consensus of the Internet community’s opinion on the best way to perform an operation or procedure. BCPs are guidelines, not standards.
TABLE 2-1 Maturity Levels for Internet Standards Track RFCs
Maturity Level
Description
Proposed Standard
Generally stable, have resolved known design choices, are believed to be well understood, have received significant community review, and appear to enjoy enough community interest to be considered valuable.
Draft Standard
Well understood and known to be quite stable. At least two interoperable implementations must exist, developed independently from separate code bases. The specification is believed to be mature and useful.
Internet Standard
Have been fully accepted by the Internet community as highly mature and useful standards.
Table 2-2 summarizes the RFCs that apply to the key Internet standards described in this book.
Domain Names — Implementation and Specification (DNS)
1939
May 1996
Post Office Protocol Version 3 (POP3)
2131
March 1997
Dynamic Host Configuration Protocol (DHCP)
3376
November 1997
Internet Group Management Protocol (IGMP) (Updates RFC 2236 and 1112)
7230 through 7235
June 2014
Hypertext Transfer Protocol – HTTP/1.1
5321
October 2008
Simple Mail Transfer Protocol (SMTP)
My favorite RFC is 1149, an experimental specification for the “Transmission of IP datagrams on avian carriers.” The specification calls for IP datagrams to be written in hexadecimal on scrolls of paper and secured to “avian carriers” with duct tape. (Not surprisingly, it’s dated April 1, 1990. Similar RFCs are frequently submitted on April 1.)
The TCP/IP Protocol Framework
Like the seven-layer OSI Reference Model, TCP/IP protocols are based on a layered framework. TCP/IP has four layers, as shown in Figure 2-1. These layers are described in the following sections.
FIGURE 2-1 The four layers of the TCP/IP framework.
Network interface layer
The lowest level of the TCP/IP architecture is the network interface layer. It corresponds to the OSI physical and data link layers. You can use many different TCP/IP protocols at the network interface layer, including Ethernet and token ring for LANs and protocols such as X.25, Frame Relay, and ATM for wide area networks (WANs).
The network interface layer is assumed to be unreliable.
Network layer
The network layer is where data is addressed, packaged, and routed among networks. Several important Internet protocols operate at the network layer:
Internet Protocol (IP): A routable protocol that uses IP addresses to deliver packets to network devices. IP is an intentionally unreliable protocol, so it doesn’t guarantee delivery of information.
Address Resolution Protocol (ARP): Resolves IP addresses to hardware Media Access Control (MAC) addresses, which uniquely identify hardware devices.
Internet Control Message Protocol (ICMP): Sends and receives diagnostic messages. ICMP is the basis of the ubiquitous ping command.
Internet Group Management Protocol (IGMP): Used to multicast messages to multiple IP addresses at once.
Transport layer
The transport layer is where sessions are established and data packets are exchanged between hosts. Two core protocols are found at this layer:
Transmission Control Protocol (TCP): Provides reliable connection-oriented transmission between two hosts. TCP establishes a session between hosts, and then ensures delivery of packets between the hosts.
User Datagram Protocol (UDP): Provides connectionless, unreliable, one-to-one or one-to-many delivery.
Application layer
The application layer of the TCP/IP model corresponds to the session, presentation, and application layers of the OSI Reference Model. A few of the most popular application layer protocols are
HyperText Transfer Protocol (HTTP): The core protocol of the World Wide Web.
File Transfer Protocol (FTP): A protocol that enables a client to send and receive complete files from a server.
Telnet: The protocol that lets you connect to another computer on the Internet in a terminal emulation mode.
Simple Mail Transfer Protocol (SMTP): One of several key protocols that are used to provide email services.
Domain Name System (DNS): The protocol that allows you to refer to other host computers by using names rather than numbers.
Introducing the Internet
My favorite RFC is 1149, an experimental specification for the “Transmission of IP datagrams on avian carriers.” The specification calls for IP datagrams to be written in hexadecimal on scrolls of paper and secured to “avian carriers” with duct tape. (Not surprisingly, it’s dated April 1, 1990. Similar RFCs are frequently submitted on April 1.)