Illustrated TCP/IP by Matthew G. Naugle Wiley Computer Publishing, John Wiley & Sons, Inc. ISBN: 0471196568   Pub Date: 11/01/98

Chapter 67
What’s Wrong with the Address?

With the vast explosion of connectivity to the Internet starting in 1994, the Internet was soon running out of IPv4 addresses. Class As in the range of 64–126 were not assigned; Class Bs were at the point of exhaustion; and Class Cs, although plentiful, only allowed for 254 host addresses per network number assignment. Class C subnetting is not exactly painless. Most sites were given multiple Class C addresses and this was quickly filling up the Internet routing tables, Some estimates were as high as 85,000 routes on the global routing tables (those tables held by national Internet Service Providers such as Sprint and MCI). Yet, the computing power of the router and availability of RAM to hold those tables in the router were not ready yet. The size of the Internet was doubling every 9 months, yet the computing power of the routers was doubling every 18 months. Instead of producing faster and more powerful routers (like we did with mainframes in the 1970s and 1980s), we became smart and invented a holdover solution using the existing equipment and current IPv4 addressing scheme.

Now we hear about the exhaustion of IP address space. Can this be true, with over 4 billion addresses? But wait. We have 32 bits of address space. Ignoring the rules of class addressing this, 2ⁿ 32 allows for 4,294,967,296 unicast addresses to be assigned (in some formation of networks and hosts). Seems like a lot of addresses, but remember, IP lived in a class environment, wasting much of the available address space. Subnetting along with protocols such as RIPv2 and OSPF allowed for variable-length subnet masks which allowed for more efficiency of the address bits, but there is still a shortage of IPaddresses.

The original problems were three types of classful addresses and address allocation without a plan. It used to be that anyone who wanted an address was given one arbitrarily, and addresses were allocated without knowledge of their location or fully understanding their network requirements leading to the proper assignment of an address. In 1992, a study was performed and the conclusion was that not only was the address space near depletion (Classes A and B), assigning the remaining 2 million Class C addresses would cause the Internet’s router array to melt down. The Internet backbone routers were already congested and slow with the current routing tables of less than 30,000 routes.

Some organizations and network providers had multiple contiguous networks assigned. Yet, as we learned in the previous section on addressing, each address is a network and holds one record slot in the routing database. The idea of supernetting was introduced in RFC 1338 as a means of summarizing multiple network numbers (one entry details multiple network IDs), further reducing the number of routes reported. This was a 1992 RFC intended as a three-year fix, which matured into CIDR.