The future is a historic artifact
This is a short tale of how a set of recommendations for the future of Internet routing could only be released once they were defined as a bit of history. It is a brief background story on RFC 5772 A Set of Possible Requirements for a Future Routing Architecture and RFC 5773 Analysis of Inter-Domain Routing Requirements and History. I had the pleasure and pain of being one of the co-authors/co-editors of these RFCs.
Back in 2001 a group of Internet routing researchers worked on a set of recommended requirements for a new routing Internet architecture. It was actually two independent groups who went to work at the same time and only became aware of each other’s efforts while the work was in progress. It was decided at that time, and again several times after, that the groups would not combine their efforts but that they would continue as two different efforts and compare the outcome. It is remarkable that though one was a blank sheet approach, while the other was using an evolutionary approach, the results were not dissimilar.
The thing is the RFCs were essentially ready to go in 2003 when the first of 10 Internet-drafts of the documents were released. Since then, the quest to release these documents as Informational RFCs has been arduous. Other then adding a section of security issues, most of the revisions have been nits or have concerned adding equivocation about whether these where requirements or just things to think about. And though we had said, from the the very first draft, that these were just something to think about and had admitted that it was probably impossible to build an architecture that responded to all of the requirements other then in the context of engineering trade-off, it was never quite good enough. The concern was, I believe, that someone might actually think these were requirements from the Internet Research Task Force (IRTF) and its Routing Research Group (RRG). I was the chair of the RRG from 2003 until 2007 and while I was able to get the RRG’s approval for publications, I was unable to get the RFC Editor or the Internet Research Steering Group (IRSG) to approve release of these drafts as RFCs. A lot of the discussion involved the nature of these recommendations - what did it mean for the IRTF RRG to put out requirements - even if they were explicitly described as things one should think about and not really really requirements?
It was only in 2009, the RRG now having new chairs, that the formula for releasing the RFCs was found by RRG co-chair Lixia Zhang. While it looked like we might never get clearance for the release of Informational RFCs, we could get permission for the release of Historic RFCs. That is, we would not be providing information for the Internet community, but rather would be describing work that had been done at some time in history. Being very crispy by then, both my primary co-author/co-editor Elwyn Davies and I agreed. There is a moral there somewhere, but I am not sure what it is - other than, of course, in the value of perseverance, talented chairs of good will and flexibility.
In any case, I had the opportunity to reread these RFCs a few weeks ago during the 48 hour review (this is a last chance review for RFC authors that often lasts longer then 48 hours- this one lasted about a week). On rereading I had several thoughts:
- I was so happy, ecstatic even, that these RFCs were about to be be released. Not only a lots of work and angst gone onto these documents, I have been telling my academic department that they were about to come out any month now for years.
- While somewhat dated due to significant technological advancements and scaling challenges, I thought the ‘requirements’ were still relevant to anyone who woud design a new Internet routing architecture. As this is an exercise that is still ongoing in the RRG, it might be fun to analyse whatever comes out of that effort against these ‘requirements’. Wonder if my co-authors are up for it.
- We left out most any discussion of the social impact of these requirements. 2001 was before I got involved in Internet governance. I do not think I even knew about there being such as concept as Internet governance and certainly could not have given even a working definition of that concept. And while I was already not a believer in the purity of science and engineering, it was not part of my, or any of our, concerns in the document. That is not to say there isn’t some discussion of social requirements on Internet routing, but there is really somewhere between very little and no discussion of the privacy and freedom of expression issues that might be consequences of various architectural decisions. The absence was glaring to me as I reread the RFCs from my current point of view. Maybe we need an annex.
After a decade trying to produce some requirements on routing, our work is finally released and I think it is pretty good. I think it is still worth reading but I also think it has shortcomings. All in all, I am glad we did it, glad we stuck with it and most of all glad they finally came out as RFCs.
