Handbook of Local Area Networks, 1998 Edition:LAN Interconnectivity Basics Click Here! Search the site:   ITLibrary ITKnowledge EXPERT SEARCH Programming Languages Databases Security Web Services Network Services Middleware Components Operating Systems User Interfaces Groupware & Collaboration Content Management Productivity Applications Hardware Fun & Games EarthWeb sites Crossnodes Datamation Developer.com DICE EarthWeb.com EarthWeb Direct ERP Hub Gamelan GoCertify.com HTMLGoodies Intranet Journal IT Knowledge IT Library JavaGoodies JARS JavaScripts.com open source IT RoadCoders Y2K Info Previous Table of Contents Next 3-2Practices for IP Addressing and Routing YAKOV REKHTER Information processing addressing and routing architecture and protocols are constantly evolving to meet the requirements of the rapidly growing Internet. This chapter describes current practices for IP addressing and routing. BACKGROUND Initially, IP address structure and IP routing were designed around the notion of network number classes (i.e., class ABC networks). The address was partitioned into two parts, a network number and a host number. The structure assumed that each data link subnetwork is assigned a unique network number, and that an interface on each host or router directly attached to that subnetwork has a unique (within that network number) host number. The boundary between the network number and the host number within an address was determined by the class of the network number. The class was determined by the high order bits of the address. Class A had the high order bit of the address set to 0. The next seven bits formed the network number part, and the remaining 24 bits formed the host number part. Class B had the high order two bits set to 10 (binary). The next 14 bits formed the network number part, and the remaining 16 bits formed the host number part. Finally, Class C had the high order three bits set to 110 (binary). The next 21 bits formed the network number part, and the remaining eight bits formed the host number part. The initial IP address structure turned out to be fairly inefficient with respect to the IP address space utilization. For example, because a single Class C network provides at most 255 host addresses, assigning addresses to more than 255 hosts connected a common data link subnetwork would require a Class B network. Although a single Class B network could accommodate as many as 65,536 216 different hosts, in practice the number of hosts attached to a given data link subnetwork would usually be much smaller, resulting in a highly inefficient address space utilization. The situation would be even worse with the Class A networks, because a single Class A network could accommodate as many as 16,777,216 different hosts, but it would be highly unlikely to have a single data link subnetwork with the number of hosts that would even remotely approaches what a single Class A network could accommodate. At the same time, the smallest block of address space that could be assign to a given data link subnetwork was a single Class C network. Therefore, a single point-to-point subnetwork would need to be assigned the whole Class C network (255 addresses), even if only two addresses (for the two end- points of the subnetwork) were needed. In addition to being fairly inefficient with respect to the address utilization, the initial IP address structure resulted in a fairly poor scaling properties of the IP routing system. The address structure required routers to maintain routes to each individual network number. Because the initial IP address structure assumed that every data link subnetwork has its own network number, the size of the forwarding tables on routers would scale linearly with the number of the data link subnetworks within an Internet. The resulting scaling characteristics would made such a routing system unsuitable for any but small Internets. Subnets The concept of subnetworks (subnets) was introduced to improve both the address space use and the scaling properties of the IP routing system. Instead of partitioning an IP address into two parts with subnets, the address was partitioned into three parts—a network number, a subnet number, and a host number. The boundary between the network number and the rest of the address was determined by the class of the network number. The subnet number part became sandwiched between the network number and the host number parts. The host number part remained the same as in the original structure. The concept of subnets improved the address space use by making it possible to subdivide addresses associated with a network number into multiple blocks (subnets), and assign each data link subnetwork just a subnet number, rather than a network number. This way, instead of using a single network number to assign addresses to the hosts on only one data link subnetwork, a single network number could be used to assign addresses to the hosts on several data link subnetworks. The concept of subnets improved scaling characteristics of the routing system by enabling aggregation of the addressing information above the level of individual data link subnetworks. With subnets, addressing information for hosts on multiple subnets (where each subnet would be associated with a particular data link subnetwork) could be aggregated into a single network number. As a result, a router would be able to maintain a route that would cover addressing information for a collection of data link subnetworks (whose subnet numbers are assigned out of a common network number), instead of maintaining one route per each individual subnetwork. Originally, the concept of subnets assumed that a given network number would be subdivided into several equal size subnets. As a result, the number of addresses per subnet (i.e., the size of a subnet) was assumed to be the same for all the subnets within a given network number. This still turned out to be fairly inefficient with respect to the address space use, as it did not accommodate diversity with respect to the number of hosts attached to different data link subnetworks. The concept of variable-length subnets was introduced to further improve address space use. With variable-length subnets, instead of subdividing address space associated with a given network number into equal size subnets, the address space could be subdivided into multiple subnets of different sizes. This way, a data link subnetwork with few hosts would be given a smaller number of addresses (smaller subnet) than a subnetwork with a large number of hosts (larger subnet). Previous Table of Contents Next Use of this site is subject certain Terms & Conditions. Copyright (c) 1996-1999 EarthWeb, Inc.. All rights reserved. Reproduction in whole or in part in any form or medium without express written permission of EarthWeb is prohibited. Please read our privacy policy for details.