Hi Alex, A few comments inline
L. Sent from my iPad > On 16 Dec 2021, at 12:48, Alexandre Petrescu <alexandre.petre...@gmail.com> > wrote: > > > >> Le 16/12/2021 à 12:29, Luigi Iannone a écrit : >> Hi Alex, >> Thanks, I might have been unclear. >> By “where” I meant to make a difference between the core infrastructure and >> “edges“, where the latter can be considered limited domains that need to >> interconnect to the Internet (the core) but as well among them not >> necessarily going through a centralized service. >> Is it clearer? > > YEs, it is clearer: a core-edge balance of localisation of innovation in > the Internet, in addition to the addressing aspect. > I would say as part of the addressing aspect. The degree of addressing innovation that can be done changes with respect to where you are (aka core vs edge) > With respect to where is Internet innovation situated - in the edge or > in the core: I think most patents are developped at the edges, in the > limited domains. There is where it is easier to innovate. A patent for > a technology in the core network would have a hard time to be agreed. I > dont know whether there are patents on BGP in the DFZ, or on DNS at > root, but I think not. > > With respect to addressing: the ULA addressing is an exemplary facet of > how limited domains might use a different addressing than the Internet > at large. However, ULA is still IP addressing - the 128bit addresses. Another simple example are IoT limited domains, where addresses are usually much shorter then 128 bits. > > Should one innovate in the edges, rebuild the edges rather than touching > on the core, and concentrate all 'addressing' work in in designing of > translation mechanisms? Yes, but would that significantly slow down the > Internet? If at every packet forward one has to look up a translation > table then it might be expensive. More energy-consuming big boxes to > translate and to offer new opportunities of controlling that translation? These are interesting questions. You do not need necessarily to translate, you can use overlays (but may still need a lookup in some table). Progress in hardware and software can help to cope with the additional operations. As for the energy consumption, it really depends where you can save it. Translating IPv6 addresses to shorter ones is worthwhile in IoT, because then you can save energy by transmitting less bits. Where to find the balance between communication and computation energy consumption depends on the specific use case IMO. Some additional question may relate about privacy and leakages. If privacy is desired how to protect the identity of communication end-point inside limited domains? From a more general perspective how to avoid information leakage outside limited domains? Ciao L. > > Alex > >> Ciao >> L. >>>> On 16 Dec 2021, at 11:00, Alexandre Petrescu >>>> <alexandre.petre...@gmail.com> wrote: >>> >>>> Le 16/12/2021 à 10:09, Luigi Iannone a écrit : Dear all, We have had a >>>> very nice discussion in the previous thread about what >>>> kind of features we would want from the Internet. We wanted to >>>> come back on another interesting point that has been raised >>>> during the side meeting held during IETF 112, namely is where the >>>> innovation happening? >>> Do you mean on what layer in the stack is innovation happening? >>> Do you mean on which part of the network: the end node, the core? >>> Do you mean where on Earth is it happening? >>> Alex >>>> During the discussion in the side meeting, there was a short exchange >>>> between Dino, arguing about “decentralization”, and Michael stating that >>>> we are “rebuilding the edges”; the importance of the role of overlays was >>>> also briefly mentioned. This is not a simple question, and may lead to an >>>> architectural argument, in line with Dirk K.’s viewpoint that only such >>>> architecture discussion may lead to possible changes to addressing, but >>>> also something that emerged in the previous thread. However, let’s at >>>> least start from the addressing perspective. Rebuilding the edges and >>>> utilizing decentralization may point to some approach to addressing that >>>> is not governed by a common addressing scheme. For instance, could we >>>> instead see a diversity of limited domain specific addressing schemes with >>>> most effort in ‘addressing’ being placed into the context >>>> translation that will need to inevitably happen? Or shall we >>>> instead follow the current path that forces the same context (IP >>>> semantics) to all participating edges (which goes counter the >>>> ‘rebuilding the edges’ comment)? Hence the question we would like >>>> to discuss with you on: how/where innovation, realizing the >>>> features discussed in the previous thread, should happen? This >>>> can help in strengthening the conclusion of the Problem Statement >>>> document >>>> (https://datatracker.ietf.org/doc/draft-jia-intarea-scenarios-problems-addressing/ >>>> > <https://datatracker.ietf.org/doc/draft-jia-intarea-scenarios-problems-addressing/>), >>>> in order to provide input on which way to tackle the problem. Luigi (on >>>> behalf of the co-authors) _______________________________________________ >>>> Int-area mailing list Int-area@ietf.org >>>> https://www.ietf.org/mailman/listinfo/int-area >>> _______________________________________________ Int-area mailing list >>> Int-area@ietf.org https://www.ietf.org/mailman/listinfo/int-area _______________________________________________ Int-area mailing list Int-area@ietf.org https://www.ietf.org/mailman/listinfo/int-area
