This chapter discusses some of the key issues involved in the design of network addressing models, the allocation of network addresses and names, and some of the techniques used to support a device configuration in medium- to large-scale internetworks. Since we are largely concerned with sizable networks, we will focus primarily on the IP address model, although we will briefly touch upon other addressing schemes that you may encounter. Network addressing is closely associated with routing design; hence, this chapter makes cross-references, as appropriate, to Chapter 3. The following topics are covered here:
An overview of the IP addressing schemes in IP versions 4 and 6 and the function of unicast, multicast, and broadcast addresses. IP address and Autonomous System Number (ASN) registration, private addressing schemes, and Network Address Translation (NAT) are also discussed.
Dynamic address and configuration services such as RARP, DHCP, and BOOTP.
Name-to-address mapping and directory services—DNS, WINS, X.500, LDAP.
Commonly used design techniques for efficient addressing models, static and Variable Length Subnetting (VLSN), route summarization, Supernetting/Classless InterDomain Routing (CIDR).
Figure 2.1 illustrates some of the basic components of the overall IP architecture required for naming, addressing, and configuration services. As we can see in Figure 2.1, several layers of abstraction are introduced as we move up the stack. Address allocation and configuration are typically performed by DHCP or BOOTP. Address resolution (IP to MAC and vice versa) is performed by ARP and RARP. Unreachability is indicated via ICMP. Dynamic name services are provided via DNS. Directory services are provided by LDAP. Electronic mail is provided via SMTP. Address translation services at both the protocol and application layers are performed by NAT.
Figure 2.1: Conceptual model showing the key protocols and context for addressing, naming, and configuration operations in an IP internetwork environment.
One of the problems of many large internetworks is that either they have evolved organically, or very little planning has taken place with respect to network address allocation. While at first this may seem a secondary issue for the network designer, a poorly thought out addressing model can result in severely degraded network performance and often limit scalability, as we will see shortly. A savvy designer can optimize valuable network bandwidth by using hierarchical addressing techniques coupled with modern hierarchical routing protocols such as OSPF.
Another key scalability issue for large-scale internetworks is the ability to manage address allocation and device configuration. On a network of 3,000 workstations it is clearly undesirable to hand-configure each device locally, especially if you have to do this more than once. Furthermore, there is an increasing trend toward mobile networking, so a static model is inappropriate. A means of dynamic configuration is required, and we will examine some common methodologies. On larger internetworks it is also likely that a naming service will be offered to provide a form of yellow pages directory service. In fact, with the scaling up of internetworks and the emergence of tools such as firewalls and dynamic dial-up VPNs, there is a growing need to configure and manage policies for features such as authentication, service quality, configuration data, and bandwidth. This huge mass of data is now beginning to be centralized and coordinated via directory services and protocols such as the Lightweight Directory Access Protocol (LDAP—described in section 2.5). We start by examining the underlying addressing model offered by IP.