Chapter 3
TCP/IP Network Design

CISCO INTERNETWORK DESIGN EXAM OBJECTIVES COVERED IN THIS CHAPTER:

ü Choose the appropriate IP addressing scheme based on technical requirements.

ü Identify IP addressing issues and how to work around them.

Due in large part to the explosive growth of the Internet, the IP protocol has easily surpassed IPX, AppleTalk, DECNet, and all other desktop protocols in modern network design. The IP protocol has proven itself as a multivendor, scalable standard that supports mainframe, desktop, and server applications.

The roots of IP are well developed in the Unix arena. However, many consider its release into the Windows environment, with incorporated services like WINS (Windows Internet Naming Service) and DHCP (Dynamic Host Configuration Protocol), to be its actual migration to the desktop. Others believe that the Internet alone was responsible for its popularity and that Microsoft and other vendors caught up to the emerging standard.

There is little doubt that modern designers and administrators will have to develop and support networks that use IP, regardless of which theory is correct.

This chapter presents many of the issues in IP design that confront network designers, including:

•  Address assignments

•  Subnet masks

•  Address summarization

In order to understand the design criteria for IP networks, let’s define some of the terminology. The terms shown below are important not only from a vocabulary perspective, but also from a conceptual one. Most of these concepts incorporate repetitive themes in IP.

Classful A classful routing protocol does not include subnet information in its updates. Therefore, routers will make decisions based on either the class of IP address or on the subnet mask applied to the receiving inter-face. In classful networks, the network mask for each major network should be the same throughout the network. Recall from previous explanations (presuming that readers have obtained CCNA-level experience, if not certification) that the subnet mask defines the bits in the IP address that are to be used for defining the subnet and host ranges. A binary 1 in the subnet mask defines the network portion of the address, while a binary 0 defines the host portion. Routing is based on the network portion of the address.

	If concepts such as subnet masks and IP addresses are unfamiliar, you may wish to obtain and study the Sybex CCNA Study Guide.

Classless Classless routing protocols include subnet mask information in their updates.

Major network The concept of a major network is analogous to the concept of a natural mask and relates to the class of the address, which will be defined later in this chapter. For example, the major network for address 10.12.12.40 would be 10.0.0.0.

Subnet mask A subnet is a logical division of addresses within a major network, defined by borrowing bits from the host portion of the address.

Variable-length subnet mask Variable-length subnet masks (VLSM) provide the designer with address flexibility. For example, the designer could allocate two hosts to a point-to-point link, while expanding the mask to permit 500 hosts on a user subnet. VLSM support is provided by classless routing protocols, including EIGRP and OSPF. RIP and IGRP require all subnets to be equally sized and contiguous. As a general rule, link-state protocols and hybrid protocols (such as EIGRP) support VLSM. RIP v2 also supports variable subnets.

Discontiguous subnets A discontiguous subnet is a major network that appears on two sides of another major network. Classful routing protocols cannot support this configuration, and the designer is well advised to avoid this situation whenever possible. Should another solution be necessary, the designer may employ secondary interfaces or tunnels to link the two parts of the disjoined networks, or convert to a classless routing protocol. It is important to note that each of these solutions comes with some costs, including greater overhead, more difficult troubleshooting, and more difficult administration.

	The automatic summarization feature found in EIGRP can create problems with discontiguous subnets. Therefore, many sources recommend disabling this feature. It is included for easier integration and migration with IGRP.

Secondaries A secondary address permits two or more IP subnets to appear on the same physical interface. Secondaries may be used to link discontiguous subnets, as noted previously, or to support other objectives. These objectives include migration to larger subnet masks without converting to a classless routing protocol (support for VLSM) or instances where local routing is appropriate. It is important to note that local routing is no longer considered an acceptable practice—the use of switches and trunking technologies is recommended. Trunking is a concept that permits logical isolation of multiple subnets on a physical media by marking each frame with a tag. Examples of trunking include Inter Switch Link (ISL) and 802.1q.

IP Addresses

Unlike most other protocols, IP demands careful planning by the designer before address allocation. In subsequent chapters, the address formats of IPX and AppleTalk will be presented in greater detail; however, both of these protocols permit the designer to assign only the network portion of the address. IP places the responsibility for assigning the host portion of the address on the administrator. Please note that the host assignment must also be unique for each network.

It is easy to forget that the IP addressing scheme was originally developed for a handful of networks and hosts. Early adopters would have been hard-pressed to predict the millions of devices in use today. As written, the initial IP addressing model incorporated the concept of class, or a way to define the scope of a network based on a parameter defined within the address itself. This strategy made sense in the early days of the Internet because the routing protocols were very limited and address conservation was unnecessary. However, in the present time, it has led to a crisis and shortage of available addresses—particularly in the largest address class.

	RFC 760, the original IP specification, did not refer to classes. RFC 791 incorporated the term classful addressing.