Regarding: > My concern is that the active participants have very limited experience in > the field that is > necessary to answer the question Ross asked.
To be fair, I think that in the IETF/RRG/related groups we have more experience in this field (deploying technology related to TCP/IP) than any other standards group in the world. Also, some of us *have* at least put some thought into the deployability and economics of deployment of possible solutions. Thinking about this more, "technically possible to design, implement, and deploy" I would agree with. Whether it is economically feasible or whether the tradeoffs make sense is something that we (for some definition of "we") can figure out over time. Ross ________________________________ From: Eliot Lear [mailto:[email protected]] Sent: 26 April 2010 01:09 To: Joel M. Halpern Cc: Ross Callon; [email protected]; Noel Chiappa Subject: Re: [rrg] Next pass Joel, I agree with much of what you wrote, even though you disagreed with what I wrote ;-) Absent properly aligned incentives lots of pretty technical work doesn't get out the door, and Silicon Valley is littered with examples, as is the IETF. My concern is that the active participants have very limited experience in the field that is necessary to answer the question Ross asked. To my knowledge in the last year there has been a single paper presented in this forum that has even approached one side of the question, and that was Luigi's work on a cost comparison of LISP mapping approaches. Work on this sort of analysis HAS occurred, just elsewhere. For instance, I think I've previously mentioned a paper done by Richard Clayton at WEIS 2009 on an economic analysis of SHIM6, and why it is unlikely to ever see wide deployment. I was thrilled that he did that, because I'm hoping others will catch on and do that for other works in this arena, but note that he didn't present the paper at the IRTF, nor did he even participate in the IETF SHIIM6 working group. The work was presented elsewhere. And so the question we turn to is whether we should wait for others to perform that work, and there I answer a definitive NO because we might be gating for work that might never happen. Rather, as we have been discussing, we caveat the recommendation by saying it is TECHNICALLY feasible, and leave it to others to decide if it is ECONOMICALLY feasible. Or is anyone here offering an economist a job? Eliot I would disagree somewhat. While we are not MBAs, nor economists, and we are not (I hope) going to do detailed business case analysis of deployment, nonetheless, deployment is arguably the key question. If there is not sufficient incentive to deploy the solution we suggest, then it will not get deployed. And an undeployed solution is not a solution. (Many years ago Noel ran into this with Nimrod, where the difficulty in articulating a benefit for anyone deploying as long as only a few folks had deployed created an almost insuperable barrier to adoption.) There are actually similar issues with several of the solutions. If the interoperability depends upon certain kinds of deployment, then there better be reasonable reasons for those deployments to take place. Put differently, we know that flag days no longer happen. Hence there has to be an incremental deployment story, with reasonable benefits / incentives for that incremental deployment to take place. (How big the incentives have to be obviously depends upon multiple factors, including how big the costs are.) Yours, Joel Eliot Lear wrote: On 4/26/10 6:23 AM, Ross Callon wrote: I feel that it is misleading to suggest that "It is technically feasible to separate location and identity" unless there is at least one proposal that does this that we agree is feasible. Misleading to what end? To lead people to believe that there exists at least one approach for locator/identity split that is technically feasible? Has anyone claimed such an approach is infeasible? It seems to me the debate within this group (at least amongst active participants) hasn't been that, but rather which approach of the several that have survived is MOST feasible. In fact personally I think that it *is* feasible, but it is not yet clear whether it is worth the cost (which I suppose is not the point of "feasibility") [...] Looking at incentives for deployment to me seems like a very reasonable approach, but I don't know if it should be mandatory in this group, given the type of expertise that requires. Either we're getting into the MBA-level discussion of whether it's good for business, which seems entirely inappropriate, or we're getting into a formal economic modeling approach, which seems better handled elsewhere in the academic realm, where real economists participate. 'The RRG did reach a rough consensus that it would be desirable to separate location and identity, should we agree on a way to do so.' The problem with that, from my perspective, is that you're putting the engineering cart (the details of how to do it) before the architecture horse (the high-level goal of separating location and identity). To put it another way, you're holding progress on the architectural front hostage to agreement on the engineering front. If we ignore technical feasibility, then there are a *lot* of things which are desirable goals. Yes, but this is the one WE are considering whether it is important for US articulate. Eliot _______________________________________________ rrg mailing list [email protected]<mailto:[email protected]> http://www.irtf.org/mailman/listinfo/rrg
_______________________________________________ rrg mailing list [email protected] http://www.irtf.org/mailman/listinfo/rrg
