Just a note on the Sky side of things, I ran into issues running CGNAT at home with Sky on-demand services. The initial authentication could come from one public IP address but then accessing the CDN from another and the connection would not be accepted.
Luckily I was managing the ISP network at the time so could resolve it quickly, but this problem also appeared for online game consoles and limiting the same group of users to a single public IP rather than using a NAT pool solved that. I don't know if Sky have now moved to IPv6 for this though with the new Q boxes - I have IPv6 at home but not Sky Q! On Mon, 21 Dec 2020 at 14:10, <[email protected]> wrote: > Hi Ben, Dave, > > Some hopefully useful information below, it’s an interesting question on > autodetection. > > > > The UK IPv6 Council has run sessions on transition technologies, reach out > if there is something specific: > > IPv6 Transition Workshop, Sep 2018 – UK IPv6 Council > <https://www.ipv6.org.uk/2018/10/26/ipv6-transition-workshop-sep-2018/> > > Do you follow the council’s output? > > > > This a public list from Lee Howard (not sure if continues to be updated) > showing known deployments of IPv6 transitions across the globe: > > > https://docs.google.com/spreadsheets/d/1ksOoWOaRdRyjZnjLSikHf4O5L1OUTNOO_7NK9vcVApc/edit#gid=0 > > > > The best know lw4o6 deployment I know of is DT, Terastream project in > Europe. I am not aware of any “operational issues” output from them > specifically. > > > > Closer to home the Sky team have been deploying MAP-T in Italia I believe. > Richard’s slides: > > slides (PDF) (ipv6.org.uk) > <https://www.ipv6.org.uk/wp-content/uploads/2018/11/Richard-Patterson-MAP-Overview-UK-IPv6-Council-2018.pdf> > > > > In terms of summarising transition tech I find Jordi Palet an extremely > valuable contributor on deployment issues and comparisons, c.f. > > ipv6-transition-v7 (apnic.net) > <https://upload.apnic.net/uploads/ipv6-transition-deployment-v7_1519533393.pdf> > > > > With lw4o6 or MAP-T/E then the protocols have been designed to be able to > avoid the main and major operational issues, by using an IPv4 as a Service > approach, either employing translation and encaps. > > It is the operators choice to then couple these with the “addressing > sharing”, in line with their own IPv4 address constraints. So in terms of > operational issues is the address-sharing known problems that you seek. > Have a look at Jordi’s paper above, especially the awesome table on page > 56, and also the list of CGN problem areas… > > > > BTW Microsoft xbox supports/prefers IPv6! PC clients are still a problem > for IPv4 address sharing. > > > > Regards, > > Nick > > > > > > *From:* uknof <[email protected]> *On Behalf Of *Dave Bell > *Sent:* 19 December 2020 16:12 > *To:* Ben McKeegan <[email protected]> > *Cc:* [email protected] > *Subject:* Re: [uknof] Operational issues with IP address sharing with > MAP-E, MAP-T or lw4over6 on residential network > > > > Hi Ben, > > > > This is also something we have been investigating recently. We have found > that CPE support for transition technologies is pretty limited outside of > WRT. > > > > I also have concerns about gaming support. I don’t believe any of the > current generation of consoles support IPv6 for their services. I did a > quick experiment a few weeks ago and found that none of the popular gaming > services on PC (steam, uplay, epic) support v6 only. > > > > VoIP too is another service which concerns me. I think the only way to be > certain about the impact of these technologies is a trial. We are looking > to run one in the next few months. > > > > I would be interested also to hear of any real world deployments and the > support impact that they have. > > > > Regards, > > Dave > > > > On Sat, 19 Dec 2020 at 15:44, Ben McKeegan <[email protected]> wrote: > > Dear UKNOF, > > I have been asked to evaluate the feasibility of deploying lw4over6 with > IPv4 address sharing as an entry level option over an existing and > growing dual-stacked altnet, primarily for new residential customers. > > My google foo is failing me on this one as I am struggling to find any > reports within the last few years as to how nicely various consumer > applications and devices play behind a shared IPv4 address, where each > subscriber is allocated a fixed sub-range of ports. For example, one > area I am aware of which can present a problem with CG-NATs is the use > of games consoles for online multiplayer gaming, however, I am unsure to > what extent this can now be mitigated by enabling IPv6 on the consoles > and/or moving all the NAT to the CPE to give the user full control of > port forwarding (within their restricted range) and to avoid the > double-NAT. > > Before I put in a budget request for a large selection of games > consoles, online gaming subscriptions and other consumer devices to go > in my test lab, which will no doubt raise a few questioning eyebrows in > the finance department :-), I was wondering if anybody on this list had > any operational experience of running lw4over6 or similar technologies > such as MAP-E/MAP-T on a residential network, and would be willing to > share their experiences of what is and isn't likely to work. What > proportion of a typical residential user base are likely to be adversely > affected by IP address sharing? Can anybody point me to any recent > studies on this sort of thing? > > I also wonder to what extent we might be able to automatically detect > and mitigate the common problematic use cases, by flagging such > subscriber accounts as candidates for upgrade to dedicated IPv4 addresses. > > Many thanks, > Ben. > > -- > Ben McKeegan > Netservers Limited > > > -- Paul Bone Network Consultant PMB Technology
