HOUR 3
Getting Data from Here to There: How Networking Works

What You’ll Learn in This Hour:

• Network protocols and the OSI model for computer networks

• Network addresses (IP and MAC)

• Introduction to Ethernet, IP, and ATM

In the preceding hour, we learned why packet switching is important to data networking and the transport of data between computers. In this hour, we learn more about how networks transport data. The first part of this hour expands on the concept of protocols, with an explanation of the famous Open Systems Interconnection (OSI) model. Next, network addresses are explained, followed by introductions to Ethernet and ring networks. The Asynchronous Transfer Mode (ATM) protocol is highlighted, as well as a protocol operating in almost all computers today: the Internet Protocol (IP). In subsequent hours, we return to these networks and protocols with more detailed explanations.

Network Protocols

To briefly review points covered in Hour 1, “An Overview of Networking,” computers communicate with each other with network protocols—rules governing how machines exchange data. We learned that physical protocols are employed to describe the medium (a copper wire, for example), the connections (a USB port, for example), and the signal (the voltage level on a wire, for example). We also learned that logical protocols consist of the software controlling how and when data is sent and received to computers, via the supporting physical protocols. In summary, protocols embody the rules—executed in various combinations of hardware and software—for sending and receiving data across a network.

To grasp the full concept of network protocols and the method by which data moves through computer networks, you need to understand their functions in relation to each other and a computer network. To begin, let’s examine the most popular conceptual model for networking: the OSI model.

The OSI Model (and Why You Should Be Familiar with It)

During the 1980s, two international standards bodies (the International Telecommunications Union [ITU] and the International Organization for Standardization [ISO]) created a model by which data communications protocols could be designed, executed, and maintained. In conjunction with the model, the ITU and ISO also published numerous protocols that followed the rules of the OSI model. The model provides a tremendously useful paradigm of how functions can be distributed among the various parts of a network.

Because the Internet protocols (such as TCP/IP) came into existence at about this time, the OSI protocols never found a wide following, but the OSI model became the archetype for computer networks. Interestingly, the Internet protocols developed somewhat separately from the OSI model, yet they closely parallel its structure.

As seen in Figure 3.1, the model is organized into seven layers. The layers are worth memorizing for debugging network problems—ranging from design issues to snafus with connections. The model is also helpful when conversing about a network. For example, Tommy might say to me, “I’m working on a Web application in Layer 7 of the model.” With that information, I know immediately the nature of the application and the underling features (protocols at Layers 6 through 1) that can be used (or must be used) to support his application.

FIGURE 3.1 The OSI model shows how data is transported through a network.

Each layer communicates only with the layer directly above or below it within the same computer. These communications take place with software function or library calls. If you have a background in software, you know these calls are also known as application programming interfaces (APIs). If you are not a software programmer, don’t be concerned about them; just remember that one layer invokes another layer by invoking the software residing in that layer. The good news is you won’t have to deal with this level of detail for bringing up your own network. But, as mentioned, it’s a good idea to know the model’s general structure.

The sole purpose of the layers’ communications is for one end user machine to transmit data to another end user machine, as seen in the left and right protocol stacks in Figure 3.1. The transport of this data is in the form of packets. In the figure, the solidlined arrows depict how the data, along with each layer’s control information (headers), is physically passed between layers (vertically). The dashed-lined arrows depict how the data and headers are logically exchanged across peer layers (horizontally). Finally, the dashed/dotted lines show the data as it is transmitted across the communications links.

The principal purpose of the vertical process is to execute the horizontal process. Of course, the actual transmission across the communications link or links only occurs at Layer 1, again as depicted by the dashed /dotted lines.

Take a look at the middle of Figure 3.1. All layers need not be executed in every machine in the network. For the relaying of packets through a network between two end-user computers, only Layers 1–3 are needed. Thus, when Tommy sends an email to me, the traffic only passes through Layers 1–3 of the routers in the Internet. Even though these routers contain all seven layers, for ongoing relay support, they are not concerned with the activities of Layers 4–7, which are usually end-to-end activities. The router passes these headers (and Tommy’s email) transparently to a next node. Consequently, by executing fewer layers and their associated functions, the routers relay the packets more quickly.

With this background, we’ll examine Figure 3.1 in a bit more detail. In so doing, we’ll explain the major functions of each layer and provide examples of prominent protocols residing in each layer. As well, we’ll compare the model to postal service operations.

• Layer 7 (application) contains the applications most familiar to users, such as email, text messaging, and file transfer. Applications such as the File Transfer Protocol (FTP) and Telnet reside in Layer 7. In the postal model, the application layer corresponds to writing or reading a letter. Products such as Microsoft Word and Excel operate in this layer, as does the widely used Hypertext Transfer Protocol (HTTP).

• Layer 6 (presentation) deals with the way different systems represent data. For example, Layer 6 defines the syntax of the data, such as IBM’s convention of coding characters entered from the keyboard. This layer can also make code conversions, such as displaying UNIX-style data on a Windows screen, or translating a Photoshop-specific image to a JPEG image.

Layer 6 does not have an analogue in the postal model, but if it did, it would be similar to the rewriting of the letter so anyone could read it. A fitting analogy is to a translator; using the postal model again, assume your letter written in English is being sent to Mexico. A translator (equivalent to presentation layer software) translates the data in your envelope into Spanish. Similar to the letter in the example, data can be “rearranged” to fit the kind of computer and software on which it executes.

An array of Layer 6 products are on the market, many of which are stored inside your computer or on your hard disk. You never see them directly, but you invoke them when prompted to, as in, “Choose from this list which program you wish to use to open this file.”

• Layer 5 (session) handles the dialogues between systems. This layer handles bidirectional (two-way) or unidirectional (one-way) communications. In the postal metaphor, the session layer is similar to a letter writer instructing a recipient to respond immediately to the letter, to not respond at all, or to respond at any time. In a text messaging application, one user might be keying in text on a cell phone and sending the message at about the same time another user is performing the same operation. In this situation, Layer 5 allows the users to send and receive data at the same time. In a file transfer application, one user may not be allowed to send files while she is sending a file. Layer 5 often forbids you to enter and send email when it is busy with other tasks.

• Layer 4 (transport) can be compared to the registered mail system. It is concerned with ensuring mail arrives safely at its destination. If a packet fails to reach the end user, the sending Layer 4 resends the packet. In effect, Layer 4 recovers from any errors at Layers 1–3. For example, if a router in a network suffers a temporary failure and loses traffic, the transport layer comes to the rescue. The sending machine’s Layer 4 must hold a copy of each packet it sends and can only discard this packet when it receives an acknowledgment from the receiving machine. If it is notified to resend, it does so. If it does not receive such an acknowledgment, it assumes something is amiss and resends anyway. If the receiving machine happens to receive duplicate packets, it uses sequence numbers in the packet to discard the redundant data.

It’s all rather extraordinary, wouldn’t you say? All these activities (sending packets, checking for errors, acknowledging the data, perhaps resending one or more packets) take place so quickly that they usually remain transparent to end users.

For the postal service, an end-to-end integrity service takes several days. For computer networks, it takes several fractions of a second, even if the end user session spans the globe. Chances are, you have come across the Transmission Control Protocol (TCP). It operates at Layer 4 inside your computer—without your intervention—to provide this wonderful end-to-end service.

• Layer 3 (network) provides addressing and routing services. When we send someone a letter, we use a street address and a ZIP code to identify the location of the recipient. When a computer sends data, it also uses addresses. For this operation, Layer 3 places two addresses in the packet: its own address (source address) and the address of the recipient of the packet (destination address). Thereafter, as Figure 3.1 shows, only Layers 1–3 need be executed in an internet until the packet arrives at its final destination.

Layer 3 is similar to the mail-sorting clerks at the post office, who aren’t concerned about the mail reaching its final destination. Instead, their concern is sorting and relaying the envelope to the next node (post office) toward the destination. Of course, the mailperson at the end office does indeed deliver the mail to the recipient.

This layer contains the Internet Protocol, the IP in TCP/IP, and the Internetwork Packet Exchange (IPX), an IP-like protocol used in older NetWare products.

• Layer 2 (data link) defines a set of rules for transporting traffic on one link (link is a physical communications channel or line) from one node to another node. Layer 2 has no awareness of conditions beyond this one link. In our postal model, Layer 2 represents conventions controlling the delivery of a postal envelope, such as dropping off the letter at a mail box, without knowledge that the letter might have to go to another mailbox. This layer contains the rules for the behavior of several widely used protocols, such as Ethernet and ATM.

The layer is concerned with finding a way for Layer 3 components to communicate transparently with Layer 1 components. In so doing, it keeps Layer 3 independent and unaware of the details of Layer 1—an extraordinarily useful service. For example, IP operating at L_3 never cares if its packets are sent over, say, a wireless cellular phone channel or a wire-based copper cable. Layer 2 performs the mediation needed to create this veil. IP sits in front of this veil, not caring if its packets are to be placed onto satellite, copper, cable, radio, or optic links. It is a brilliant way for conceiving this part of the model.

Layer 2 may place packets inside frames, which are used by hardware devices to send and receive traffic below Layer 3. This operation is akin to placing one postal envelope (a conventional envelope) inside another postal envelope (an overnight delivery envelope).

• Layer 1 (physical) is similar to the trucks, trains, planes, and rails that move the mail. This layer is concerned with the physical aspects of the operation, such as electrical, electromagnetic, and optical signals; the network interface cards (NICs); and the wire and cable. The modem is an example of a Layer 1 device. Also, many of the operations of broadband services operate at Layer 1. As examples, the Digital Subscriber Line (DSL) service from the telephone companies and the cable broadband service from the cable TV providers mainly operate at Layer 1 of the OSI model.

How to Identify the Type of Packet in an Ethernet Frame

In today’s networks, different kinds of packets may be exchanged between computers on an Ethernet link. In the past, the previous sentence with the words “may be” would read “are.” Times have changed and, with rare exceptions, the data inside the Ethernet frame is an IP packet.

Nonetheless, the Ethernet header contains a field called EtherType, which is filled in by the sending node to identify the specific packet being carried in the Ethernet frame. As examples, IPv4 is EtherType 0800, and IPv6 is EtherType 86DD. This field is quite valuable for organizations that are migrating to IPv6 but still have IPv4 components to support. In effect, dual stacks of software are maintained, and the EtherType field is used to pass the IP packet to the process in the appropriate software protocol stack.

As mentioned, before IP became so pervasive, IBM, Novell, DEC, Apple, and other vendors deployed their own proprietary L_3 protocols, and some machines, such as routers, had to support all of them. The EtherType field was used to pass the packets to the correct process in a protocol stack.

The Internet Model

The Internet also uses a protocol model, which is similar to the OSI counterpart. However, it is a five-layer scheme and does not include the presentation or session layers. Does this mean that the services associated with these layers aren’t available for Internet operations? No, it means the services exist in vendors’ products, and with some exceptions, they aren’t defined in the Internet standards.

In later hours, we examine many aspects of the Internet’s use of this model, as well as the standards and protocols published by the Internet Engineering Task Force (IETF). These specifications are now embedded into practically all vendor’s products. In just two decades, they have transformed the computer network industry.

Addresses: Network or Layer 3 Addresses

Continuing our postal service analogy, we now focus on those all-important addresses. Computer networks have a source and destination address appended to each packet header, similar to the postal envelope shown in Figure 2.1 (Hour 2, “The Benefits of Networking”). These addresses are embedded in the L_3 header, as seen in Figure 3.2. The principal job of a router (aptly named) is to use the destination address to route the packet toward its final destination. Because this operation takes place in Layer 3 of the OSI model, these identifiers are called network addresses.

The most widely used network address is the IP address. It is 32 bits long, which conceptually allows for 2³² addresses (4,294,967,296). When this address was conceived, the first 8 bits identified a network, and the remaining 24 bits identified a host (such as a computer) attached to that network. This convention assumed an IP address would not have to identify more than 256 networks! Hindsight is twenty-twenty (LANs had not yet made their presence known), and we will learn shortly that the Internet standards bodies have modified this structure and published other protocols to allow the 32 bits to identify many more networks and hosts. For now, let’s examine the conventional IP address format.

IP addresses are written in dotted decimal notation, with one byte (eight bits) between each dot. A dotted decimal IP address appears as 192.168.100.25.

Because each number is described by one byte, and because each byte is 8 bits (of binary 1s and 0s), each number can have a value ranging from 0 to 255. Because there are 4 numbers with 8 bits each, the total address space is 32 bits long (4*8 = 32). So the preceding address, in binary, appears as 11000000.10101000.01100100.00011001.

In the past, IP addresses were allocated to organizations in address blocks. Address blocks come in different sizes, based on the class of address. This scheme is explained here for background information. Because of its limitations, it was replaced by Classless Inter-Domain Routing (CIDR), described next.

• Class A addresses use 24 of the 32 bits in the address space for host addresses. A Class A address appears as X.0.0.0, where X is the network address.

• Class B addresses use 16 bits each for the network and host. A Class B address appears as X.X.0.0.

• Class C addresses use 24 bits for the network address space. Here’s an example of a Class C address: X.X.X.0.

• Class D addresses are used for multicasting: sending messages to many systems. Some 911 systems use multicast because it helps ensure that the systems receive all messages. Online teaching applications often use multicast for delivery of the lecturer’s voice and video packets to a wide audience.

Private addresses can be used when traffic does not leave a private network. Thus, they can be reused in each private network. The Internet authorities have allocated these values for private addresses:

Alternatives to the Conventional Address

In the previous section, we learned the IP address is only 32 bits long, and that its structure of four eight-bit boundaries restricts how the 32 bits are used. In this section, we examine three conventions published by the Internet authorities to compensate for the restrictions of the original IP address length and format. They are CIDR, NAT, and IPv6.

CIDR

The Internet standards bodies recognized the “classful” address described earlier would not meet future needs. Consequently, in 1993, CIDR (pronounced “cider”) was introduced. Instead of using the rigid 8-bit boundaries, CIDR specifies arbitrary (variable-length) boundaries of the 32 bits in the IP address. It also specifies a method to group network addresses with the same sequence of bits in the network “space” as only one entry in a router’s routing table. This technique, called address aggregation, greatly reduces the size of routing tables and speeds up routing table lookups.

NAT

The Network Address Translation (NAT) Protocol allows multiple hosts on one network (say a private LAN in an office) to use the Internet with only one public IP address. A router sits between the private network and the public Internet. The router is assigned a public IP address to communicate with the Internet. However, behind the router, on the local LAN, all computers (hosts) use private addresses. What is more, these private addresses are never revealed to the public Internet, which results in a valuable security service to a user. It is the router’s job to maintain a table that tracks and converts private addresses on outbound packets to the public address. Conversely, for inbound packets, the router converts the public address to a corresponding private address. Other identifiers, beyond this discussion, are used to help the router keep the traffic properly tagged.

IPv6 Addressing

IPv6 (version 6) is intended to be the successor to the current IP (IPv4). IPv6’s header contains larger address spaces, thereby eliminating the need for CIDR and NAT. Each address is 128 bits in length, compared to 32 bits for IPv4. Thus, IPv6 supports 2¹²⁸ addresses. This huge number is often compared to the number of people living today, which is about 6.5 billion humans. IPv6 provides 5 X 10²⁸ addresses for each of these people! A 128-bit address supports thousands of billions more addresses than a 32-bit address.

It’s reasonable to assume the IPv6 source and destination address fields will suffice for the foreseeable future. That stated, because of the effectiveness of CIDR and NAT, IPv4 continues to be the dominant IP version deployed in both public and private networks. Some governments have established deadlines for equipment and software to be IPv6 capable. The U.S. Government’s deadline is 2008. Time will tell if IPv6 becomes the prevalent version for running IP.

MAC or Layer 2 Address: That Is to Say, Ethernet Addresses

Each device on a LAN is identified with a Media Access Control (MAC) address. As mentioned earlier, it is also called a Layer 2 address. Originally, this identifier was called the Ethernet address. When the Ethernet specification was folded into the IEEE 802 standards, its name was changed to the MAC address because it is associated with the IEEE 802 Layer 2, which is called the Media Access Control layer. Some people refer to this value as a hardware address as well, because manufacturers might place the MAC address on a logic board (such as an NIC) inside the computer.

As with the source and destination IP addresses at Layer 3, a LAN PDU (a frame) contains Layer 2 source and destination MAC addresses. They are coded in the L_2 header, shown in Figure 3.2. (This layer contains a header and a trailer.) The address is 48 bits long and must be used on all LANs if they are to operate correctly. In the past, the Xerox Ethernet Administration Office assigned these values to LAN equipment manufacturers. Now the IEEE takes care of this responsibility.

Using Addresses to Relay Traffic

When Ethernet was coming into existence, so too was the Internet. Both were created by their inventors without these men knowing how often their creations would interact with each other in the future. Thus, two sets of computer network addressing standards found widespread use in the industry. When it became apparent that Ethernet-addressed networks would have to interact with IP-addressed networks, the question became: How?

As mentioned, a MAC (aka Ethernet) address is usually configured by the vendor on the NIC in each computer. Let’s assume you have two PCs attached to your router via Ethernet cables. For this discussion, we amplify a previous figure to that shown in Figure 3.3. Notice that the Ethernet LAN operates with IP at L_3 and with Ethernet MAC at Layers 1 and 2.

FIGURE 3.3 LAN and WAN layers

Also, notice that PCs C and D use wireless protocols and interfaces (W1 and W1) at Layers 1 and 2. In some wireless networks, this Layer 2 is similar to Ethernet’s Layer 2.

PCs A and B and the router use MAC addresses to ensure traffic is sent to the correct node on this local network. Let’s assume PC A is sending data, via the router, to PC B. It creates the frame with a MAC source address of A and a MAC destination address of B. The router is configured to pass this data unit to PC B by examining the MAC destination address of PC B. The router does not pass the data up to its L_3.¹

¹ Some people identify a machine that routes based on MAC addresses (and not IP addresses) as a bridge. High-end routers can be configured to operate with either MAC or IP addresses.

Upon receiving this frame, PC B determines the destination MAC address in the L_2 header of the frame is the same as the address of PC B, so PC B accepts the traffic and passes the packet to IP at L_3. (Because of some other control fields in the frame header, PC B knows it must pass this packet up to its L_3, which it does.)

PC B’s IP then looks at the L_3 destination address in the IP header. Lo and behold, this address is the same address as the IP address of PC B. Thus, PC B knows it must perform a number of IP services. It then passes the data up to Layer 4. Eventually—actually almost instantaneously—the data is presented to an end user application in Layer 7, such as email.

Alternatively, let’s assume that PC A wants to send a packet to a remote computer, somewhere in the Internet. In this situation, the destination MAC address is not that of the PC B, but that of the router. By previous agreements, PC A and the router have agreed that any nonlocal traffic will be sent to this router, but the final IP destination address has been placed in the IP L_3 destination address field by PC A. Thus, the router examines the L_3 header and its destination IP address. It learns that this packet is intended for a nonlocal computer.

The router then makes several forwarding decisions, which result in the packet being sent to the wide area network (WAN)—that is, the Internet—toward its journey to the destination user.

In this situation, the router no longer deals with the MAC L_2 address, which is strictly a local address. It strips off Ethernet’s L_1 and L_2 and sends the L_3 IP packet to its external link to the Internet (the WAN cloud in Figure 3.3) via DSL, broadband cable, satellite, or conventional dial-up. (For the dial-up option, and not shown in Figure 3.3, a router is not installed at the local site. The telephone company at its end office assumes the responsibility of providing a router interface into the Internet.)

Ethernet MAC is not designed to operate on a WAN link. Therefore, at this WAN interface, the IP data unit is placed inside at ATM data unit (shown as “ATM2” in the figure). ATM is designed for use in different kinds of computer networks. It has found a wide niche in WANs because it specifies procedures for negotiating services such as higher throughput and faster response time. It is a common L_2 protocol on locally attached routers, as seen in Figure 3.3.

The notation of “BB 1” (broadband at L_1) means Layer 1 is DSL, cable, satellite, or nonbroadband dial-up. IP and ATM don’t care about the physical media at L_1. Of course, network designers must take these options into account when they place assorted L_2 protocols over various L_1 interfaces.

To conclude this introduction on how data moves through computer networks, hereafter, the IP traffic is relayed through the Internet, using the L_3 IP addresses to aid in this operation. ATM also plays a major role in this partnership, which we examine in Hour 6, “Extending LANs with Wide Area Networks (WANs).” At the other end of this session, the processes in this example are reversed, all with the goal of delivering the email (the data) to the end user.

Fortunately, you won’t have to be concerned with these details. For certain, when you bring up your own network, you will come across terms such as IP, MAC addresses, and ATM. Therefore, knowing their functions will help you to better operate and manage your own network. But with some exceptions, the service provider is responsible for installing and configuring all these protocols and addresses, all to your benefit.

Overview of the Principal Protocols

In Hours 8, 14, and 15, we return to the principal protocols employed to “move data from here to there.” For the remainder of this hour, we examine several keys aspects of these protocols, which will allow us to proceed into subsequent hours.

Ethernet

The Ethernet standards are designed for LANs. The original Ethernet protocol was developed at the Xerox Palo Alto Research Center (PARC) in the early to mid 1970s.² Originally, Ethernet ran over a shared coaxial cable, as seen in Figure 1.1(c) in Hour 1. The LAN nodes are allowed to send frames at any time, without prior arbitration process. Consequently, traffic on the shared channel might cause a “collision.” Each node is capable of monitoring its own transmission. Therefore, when a node notices interference, it initiates a procedure to make certain all other nodes are aware of the problem, and then it resends the frame.

² I had an opportunity to visit PARC in the early 1980s to evaluate Ethernet for possible use by the Federal Reserve Board. We did not choose this LAN at this time, but I was impressed with Ethernet’s architecture. I was even more impressed with PARC’s use of computer screen icons (trash cans, and so on) that Apple and Microsoft later put into their products.

Such a procedure entails wasted resources during periods of high activity, resulting in more collisions. As well, a break in the cable, or a malfunction of a cable termination point, brings down the network. Consequently, the industry has migrated to Ethernet Star topologies (Figure 1.1(a), Hour 1). A variety of “Switched Ethernet” standards and products are available, which we examine in Hour 11, “Selecting Network Hardware and Software.”

To conclude this introduction to Ethernet and to set the stage for subsequent hours, the following facts are germane. Ethernet:

• Does not require an association (a logical connection) to be set up before frames (data) are exchanged

• Does not provide an acknowledgment for the successful receipt of a frame

• Does not check for out-of-sequence or duplicate frames

• Performs an error check for corrupted data; if corrupted, the frame is discarded

• Does not provide for the negotiation of services

• Is not designed for WANs

The Internet Protocol (IP)

IP is the most widely used Layer 3 protocol in the computer network industry. Originally developed for use in WANs, today it operates over Layers 1 and 2 in practically every data (and even voice) communications device in existence. Even cell phones, if Internet capable, employ IP. As mentioned earlier, IP carries the ubiquitous IP addresses in its header, which is the main reason for its popularity. Be aware that specific IP implementations might not use the header fields to perform these services:

• Type of Service (TOS)—Requests various levels of delay and throughput of packet delivery.

• Time to Live (TTL)—Time the packet can remain active as it finds its way to the destination. Often implemented with a maximum permissible hop count (number of nodes that can be traversed). This option is used to ensure IP packets do not “thrash around” in the network indefinitely.

• Fragmentation—Allows an IP packet larger than the permitted L_2 frame size to be reduced to fit into the frame and then reassembled at the receiving IP node.

• Options—Provides for other optional services; some are defined in the IP standard.

To conclude this introduction to IP and to set the stage for subsequent hours, the following facts are germane. IP:

• Does not require an association (a logical connection) to be set up before packets are exchanged

• Does not provide an acknowledgment for the successful receipt of a packet

• Does not check for out-of-sequence or duplicate frames, unless an IP option is supported

• Performs an error check for corrupted data; if corrupted, the packet is discarded

• Provides for limited negotiation of services

• Is not aware of the nature of the underlying Layers 1 and 2

The Asynchronous Transfer Mode (ATM)

ATM is a relative newcomer to the world of network protocols. It has not seen much use in LANs, at least in comparison to Ethernet. However, it’s employed extensively in WANs. If you have a home network with a DSL link to the Internet, chances are good that your router is running ATM at Layer 2 for communicating with your Internet service provider (ISP).

For the transport of traffic through the Internet, the use of IP addresses has proven to be excessively cumbersome and time consuming. Here is where ATM comes into play. Before IP packets are transported into the Internet, their IP addresses are correlated with other identifiers called virtual circuit/path IDs. These values are not addresses, but simple labels that identify the traffic. A “routing” table (which is called a label switching table) of these values can be accessed quickly, without resorting to the inefficient IP operations. Indeed, IP packets are transported transparently through an ATM-based network. Thus, ATM networks are noted for their efficiency and low delay. Much of the ATM logic can be executed in hardware, which further reduces the time to process traffic. In addition, ATM was designed to allow the network provider and network users to negotiate and provide an array of services.

To conclude this introduction to ATM and to set the stage for subsequent hours, the following facts are germane. ATM:

• Requires an association (a logical connection, called a virtual path or virtual circuit) to be set up before cells are exchanged

• Does not provide an acknowledgment for the successful receipt of a cell

• Does not check for out-of-sequence or duplicate cells. Protocols at higher layers assume this function (for example, TCP, discussed in Hour 14, “Connecting to the Internet: Initial Operations”)

• Performs an error check for corrupted data; if corrupted, the cell is discarded

• Provides for a rich set of services and extensive traffic management tools

• Is not aware of the nature of the underlying Layer 1

IPX

The IPX protocol is derived from a Xerox Network Services (XNS) product and is part of Novell’s NetWare operating system. IPX uses an address similar to the conventional IP classful address, but it makes some adjustments for the sake of efficiency. IPX was widely used in the 1980s and 1990s, but it has seen diminishing use as the industry has migrated to IP. A survey on the Internet³ revealed that 44% of companies using Windows Server did not run IPX; 34% ran IPX but were migrating away from it. In addition, Windows Vista does not support IPX, nor does any Mac OS later than version 9.2.2. We bid farewell to IPX, but we also assure IPX customers that Novell is now supporting them with its migration to TCP/IP.

³ Go to http://articles.techrepublic.com.com/5100-10878_11-5026038.html.

And Farewell NetBIOS and NetBEUI

Let’s also put to rest two other systems. The Network Basic Input/Output System (NetBIOS) is—in computer network time—an ancient protocol. Developed in 1983 for IBM PC networks, it provides services similar to the session layer of the OSI model. If used, NetBIOS runs over TCP/IP principally for some naming services.

The Network BIOS Extended User Interface (NetBEUI) has been used in small LANs, the old LANManager; as well as Windows 3.x. It isn’t routable. Computers attached to a nonlocal network can’t use it. Thus, IP has replaced most NetBEUI installations.

Summary

We’ve covered a lot of material in this hour. The information is important and will be a great aid for grasping many concepts in the next 21 hours. The topics in this hour are vital to the network or networks you have in your office and home. With the concepts presented in this hour, you should have the basic tools to begin setting up your own network.

Q&A

Q. In a typical data network, which layers of the OSI model are executed?

A. Only Layers 1, 2, and 3, which translates into faster relaying of traffic through a network.

Q. IP operates at which layer of the OSI model?

A. Layer 3.

Q. To deal with the limited address space of IPv4, the industry has devised three solutions. What are they?

A. Classless Inter-Domain Routing (CIDR), the Network Address Translation (NAT) protocol, and IPv6 are the three solutions devised by the industry.

Q. Why have IP “classful” addresses fallen into disfavor?

A. The use of rigid network and host bit boundaries restricts the number of addresses that can be obtained from a 32-bit address space.