Hi Noel, In "Re: [rrg] Ivip exemplifies Noel's support for 'Adventurous Practical Design'", you wrote:
>> Past and current LISP development seems to be predicated on the belief >> that this, or anything like it, is and will remain impossible. > > Yes, we do not believe that full-table distribution is feasible. You > disagree. It's clearly a matter of opinion, and you clearly don't agree with > our opinion. Its not just a difference of opinion. As far as I know, none of you have ever read how I propose to get real-time mapping to ITRs. I first wrote this up in an ID in July 2007 - and it was Dino who urged me most strongly to write the ID. Now I have an improved design, simpler and better in some ways and with an additional mechanism "Lost Payload Servers" which should make the system even more robust. The new design is here: http://tools.ietf.org/html/draft-whittle-ivip-fpr-00 You have an opinion about getting real-time mapping to all ITRs, but unless you have read and considered this ID, in my view, your opinion is not well-informed. > That's fine - just don't expect us to agree with yours, either. I am not expecting you to agree with me. If you disagree with me, I am expecting you to explain why on the RRG list. Do you think this is an unreasonable expectation? You want a bunch of people to choose LISP in favor of every other alternative. So I think you owe it to those people to explain what is deficient about these alternatives. If LISP was just an exercise for its own sake, I wouldn't expect LISP folks to engage in debate about their own or other proposals. However, since the team has submitted LISP to the RRG for selection as the architecture to recommend to the IESG, it follows that the team should be working to supporting the RRG debate. Some of the other proposals are not really scalable routing proposals. I don't expect you to debate or contribute to critiques about them. 5 or 6 of them are for CEE schemes - and these involve such barriers to widespread adoption that I don't see how the RRG or the IETF could consider them for the foreseeable future. So I don't expect the LISP team to debate them at length either. This leaves LISP, Ivip, TIDR and RANGER - all of which are CES architectures. All of them are intended to provide compelling benefits to early adoptors - and so are intended to meet all the constraints which arise from the need for widespread voluntary adoption: http://www.firstpr.com.au/ip/ivip/RRG-2009/constraints/ None of them involve new host responsibilities, so they are fine according to this set of arguments: http://www.firstpr.com.au/ip/ivip/RRG-2009/host-responsibilities/ CEE architectures are at odds with both of the above. You submitted LISP to the RRG because you believe it to be superior to all the alternatives - and the Ivip, TIDR and RANGER alternatives really matter. I think it is your responsibility to tell RRG folks why LISP is superior these three. General statements do not achieve this. You need to refer to the particular details of each alternative and show why LISP is superior. If you don't care about what the RRG recommends - then fine, don't submit LISP and don't contribute to critiques of other proposals. But you can't expect me, or probably other people, to respect your opinions or to have confidence in your approach to designing a solution if you don't explain clearly why you consider all other architectures to be inferior to LISP. I don't accept arguments about you being busy etc. We are all busy. You have had two and a half years to read and critique Ivip, APT and TRRP. RANGER and TIDR have been less prominent, and APT and TRRP are no longer being worked on. You have time to request and work in an IETF-WG for LISP. You proposed LISP-ALT to the RRG, but now I understand are abandoning it to work on a DNS-like mapping system - which is the approach Bill Herrin took with TRRP. You expect other people to choose LISP over Ivip, but you don't help with the critique of Ivip. I think you should also explain what you think is wrong with TIDR and RANGER, rather than leaving it to me or someone else. More generally, I think any proponent of a CEE architecture should argue why CEE architectures in general, or their particular architecture, is a practical solution - or the best solution of all - considering the barriers to widespread adoption which appear to affect all CEE schemes. I am not just complaining about LISP folks being largely absent from this debate. Ran Atkinson (ILNP) and Fred Templin (RANGER) haven't debated anything on the list for months. Like the members of the LISP team they have deep experience which should enable them to contribue well informed critiques of any proposal they read. Two LISP folks did contribute a few comments about some non-CES architectures. But considering LISP's prominence and claims to superiority - and considering there is a team of people, not just one, I think it is reasonable to expect more hands-on-deck reading and critiquing of other proposals. - Robin _______________________________________________ rrg mailing list rrg@irtf.org http://www.irtf.org/mailman/listinfo/rrg
