Address Lending for Network Stability

RFC 2008 was published in October 1996. It foreshadowed RFC 2050/Best Current Practice (BCP) #12, and it remains in effect as Best Current Practice #7. Although it tackles the rather verbose and cumbersome topic of "Implications of Various Address Allocation Policies for Internet Routing," it remains noteworthy for establishing the concept of address lending. More importantly, it clearly documented the benefits of address lending in contrast to address ownership. RFC 2050, in comparison, takes a more limited focus by urging ISPs to provide their customers with an initial address space that should be reclaimed after each customer leaves.

By this point, you should be quite familiar with RFC 1918 addresses, RFC 2050/BCP #12, and NAT. We've looked at them separately a few times. RFC 2008/BCP #7 established the tandem benefits of having end-user organizations "borrow" IP address blocks rather than "own" them. This document was developed as a result of the IETF's experiments with the 39.0.0.0/8 address space (known as the Net 39 Experiments). Those experiments proved unequivocally the benefits of geographic aggregation of address space. Unfortunately, for address space to be aggregatable by region, the entire philosophy of handing out address space would have to be rethought.

No longer did the IETF deem it desirable to just hand out blocks of addresses sequentially to anyone who asked, regardless of geography. The result of that approach was an unwieldy Internet routing table with no hope of aggregating the smaller network blocks. By borrowing a network address block from an ISP, many of the Internet's original addressing problems could be satisfied. Unlike some other RFCs, the benefits of RFC 2008 accrue to both the end-user organization and the Internet at large.

Benefits to End-User Organizations

End-user organizations benefit tremendously by embracing the concept of address lendingthey just don't like the idea. During the last two years that I have been employed by AppliedTheory Corporation, I have never ceased to be amazed by the high value and importance that customers place on having their own privately registered address space. It's as if a certain cachet is associated with having your own address spacea sort of credential that announces your seriousness as an Internet user. Discussions about RFCs 2008 and 2050, or even explanations about the ill effects that directly registered address space can have on Internet routing tables, usually produce the same deer-in-headlights stare. It's obvious that they either don't understand what I'm saying, fail to appreciate how it relates to them, or are dumbfounded that I can't appreciate why they need their own address space.

Despite this personal history, I remain convinced of the benefits of end-user organizations. Borrowing an address space all but requires you (as an end user of the Internet) to configure and run NAT. You would use the borrowed address space as your inside global addresses, and you can use any of the reserved RFC 1918 addresses as your inside local addresses. This combination of technologies, as you have seen in this book, affords greater security by hiding internal IP addresses from external networks. More importantly, it enables that end-user organization to change ISPs without the hassle of having to renumber all its internal addresses.

Finally, the combination of NAT and borrowed addresses has the potential to make the IPv4 address space last almost indefinitely, which is of great benefit to anyone not willing to invest in the migration to IPv6. We'll look at IPv6 much more closely in Chapter 15, "IPv6: The Future of IP Addressing." For now, trust me when I say that this is a tremendous benefit! The migration, despite the plethora of transitional devices available in the IPv6 protocol suite, is nontrivial. Thus, I submit that staving off the migration is a huge benefit.

Benefits of the Internet

Although I view the benefits to end-user organizations as very compelling, the concept of address lending offers even greater benefits to the Internet itself in the form of absolutely minimal routing tables. The smaller the Internet's routing table, the more efficiently every IP packet that flows across it can be routed. Thus, the Internet operates more quickly, and end users realize better performance.

Even more importantly, the remaining supply of IPv4 addresses can be deployed in a very logical and efficient manner. Although that's not a lengthy list of benefits, the impact of each one is incalculable. Perhaps that's why RFC 2008/BCP #7 and RFC 2050/BCP #12 remain in effect.