The End of the Beginning

By: Scott Bradner

By now most of you have seen various news stories about the IP Next Generation (IPng) recommendation that Allison Mankin and I made at the Internet Engineering Task Force (IETF) meeting in Toronto. Those stories covered many facets of the features we recommended to replace IP (To be known as IP6 after its assigned version number). Even so, I'd like to make a personal observation or two.

Its been a tiring and occasionally stressful process. Many people have invested huge amounts of effort in finding the best answer to the set of questions needed to frame the search for the protocol which could carry the Internet well into the next millennium. By Internet, I do not mean only what we see as the Internet today, but also the general data connectivity needs of the future.

The developers of the protocols (over 2 dozen names appear on the documents as authors), the developers of the criteria document and the 15 members of the IPng directorate have dedicated significant amounts of their, and thus their corporation's, time to this process. With all of this time, effort and ego involved, things could have been far worse than they have turned out to be.

The quest has not been simple because there is not, and most likely can not be, a single view of what the future of network entails. The proposals that were offered for review represented different approaches to solving the current problems of addressing scale and routing complexity. They represented different views of the problem set and sought to optimize different aspects in their solutions. In addition, each of the proposals sought to take advantage of the need to fix the addressing in order to add additional functionality and thus better meet the requirements for the future.

None of the proposals, as evaluated in May, fully met the criteria that had been developed. Since then, in the normal IETF fashion, people from various working groups produced a number of revised proposal components which better meet the predicted requirements. It also turns out that the revised proposals were simpler than the original ones.

For some people, it might be important to claim that one proposal "won" and the others "lost". This would be unfortunate. The recommended proposal is best viewed as a synthesis of efforts and we must now all attempt to ensure that the final details are worked out in ways that maximize the functionality while minimizing the complexity.

A if-this-were-a-tabloid-it-would-be-a-conspiracy side note. There has been a Canadian component in the IPng process from before such a process existed. The first proposal that I'm aware of describing a replacement for IP was made by Ross Callon (then a Canadian) in 1987. The basic projections that lead to the current effort were done at the IETF meeting in Vancover BC. Ross and Steve Deering, also a Canadian, were prime authors of two of the three final proposals that the IPng area evaluated. Finally, the recommendation was announced in Toronto and the recommendation asks both Ross and Steve to co-chair the IPng working group. In one way or another, it's been a Canadian-influenced process from beginning to end.

If one were to look at the overall process of getting from the initial understanding that a new IP was needed to the point where the new protocol will be the norm on new data networks, we have just about reached the end of the beginning. That is something to celebrate while recognizing the work still ahead.

Copies of the Toronto presentation are on hsdndev.harvard.edu for anonymous ftp or gopher access from the directory pub/ipng/presentations.

Disclaimer: Harvard tolerates my doing this sort of thing but will disavow all knowledge if I'm caught at it.

sob@harvard.edu