UNIX System Administration Handbook - Evi Nemeth [167]
Rather than advertising every subnet to the outside world, in our 128.138.243.100 example you would only need to advertise a single class B network. Once a packet arrived within the subnetted area, its destination address would be reinterpreted with local netmasks, the real target network “discovered,” and the packet routed to its exact destination.
The IP address crisis
The Internet community realized in about 1992 that there were three fundamental problems with the original address allocation scheme. First, we were going to run out of class B addresses—the most desirable ones for moderately large organizations—by mid-1995. At the same time, the routing tables of Internet backbone sites were growing so large that they would not fit in the memory of available routers. And finally, IP addresses were being allocated on a first-come, first-served basis with no locality of reference; that is, numerically adjacent addresses could be within the same organization or on different continents. Imagine the confusion that would result if phone numbers or zip codes were assigned in this haphazard fashion.
To solve the problem, two solutions were advanced in tandem: one for the immediate future and one for the long term. Classless Inter-Domain Routing (CIDR), the short-term solution, is a different way of managing the existing four-byte address space that uses the available addresses more efficiently and allows routing tables to be simplified by taking numerical adjacencies into account. We discuss CIDR in more detail in the next section.
The long-term solution, IPv6, is a revision of the IP protocol that expands the address space to 16 bytes and incorporates several other lessons learned from the use of IP over the last 25 years. It removes several features of IP that experience has shown to be of little value, making the protocol potentially faster and easier to implement. It also integrates security and authentication into the basic protocol and eliminates fragmentation at intermediate routers. 16-byte addressing gives 2128 possible addresses: that’s 665,570,793,348,866,943,898,599 addresses per square meter of the Earth’s surface. IPv6 addresses were extended to 16 bytes after calculations showed a remote chance that 8 address bytes would not be enough. It has been suggested that the committee making the decision had rusty math skills and thought that 16 bytes would make twice as many addresses as 8 bytes did.
As of 2000, IPv6 is still in the process of being standardized, but CIDR has been fully deployed. CIDR is supported and used by the Internet backbone and by the major manufacturers of routing equipment. NAT, a scheme for reusing IP addresses that’s covered on page 279, also played a large role in reducing the demand for IP addresses.
The complexity of IPv6, the efficiency of CIDR and NAT, and the inertia of an Internet that already works pretty well all combine to suggest that it may be a long time before we move to IPv6, if indeed we ever do. Such a move will likely be driven by countries such as Japan or China that cannot get the IPv4 address space they think they need or a new killer application that requires IPv6. A good candidate for such an application might be a new generation of cell phones and other wireless devices that embed a telephone number in an IPv6 address. Voice-over-IP systems would also benefit from a closer correspondence between phone numbers and IPv6 addresses.
Some additional details on IPv6 addressing are given on page 281.
CIDR: Classless Inter-Domain Routing
CIDR, defined in RFC1519, eliminates the class system that formerly determined the network portion of an IP address. Like subnetting, of which it is a direct extension, it relies on an explicit netmask to define the boundary between the network and host parts of an address. But unlike subnetting, it allows, for purposes of routing, the network portion to be made smaller than would be implied by an address’s implicit class. Using a shorter netmask