From what I've seen, rolling out dual-stack will take about 40% of your traffic to native v6. YMMV of course. In addition to services that don't support v6, there are also devices (looking at you, Roku) that don't support it, or things like smart TVs that don't have it turned on by default, and most users aren't going to go poking that deep in the menus to enable it.
With respect to the port usage, I've seen some CGN solutions that pre-allocate a block of ports per inside IP, but allow overflow, so they will allocate additional blocks of ports as needed. That seems to be a good balance because you don't burn a ton of ports for lighter users, and the logging requirements are pretty minimal since a log only gets generated when an additional block is allocated. It does mean that one user's traffic could be popping out of two different public IPs. On 10/10/24, 4:10 PM, "NANOG on behalf of Aaron Gould" <nanog-bounces+andrew.peterson=calix....@nanog.org <mailto:calix....@nanog.org> on behalf of aar...@gvtc.com <mailto:aar...@gvtc.com>> wrote: [You don't often get email from aar...@gvtc.com <mailto:aar...@gvtc.com>. Learn why this is important at https://aka.ms/LearnAboutSenderIdentification <https://aka.ms/LearnAboutSenderIdentification> ] [External Email] also, isp-embedded cdn caching was required to provide ipv6, iirc for most of mine, and I provided ipv6 subnets even if it was optional. now i just need to enable ipv6 on the last mile broadband and I'll be in business! i can't wait to see the results. as I previously stated, I do not want to plan growth for my cgnat boundary...ipv6 is my (the) answer to relaxing the use of my cgnat boundary. i've tested 6vpe successfully over my pre-existing ipv4 mpls l3vpn's, and it's just another rt import/export to get ipv6 flowing naturally out to the internet. i've currently been testing ftth in my lab with calix cpe, and have successful ia_na (wan) and ia_pd (lan) prefix delegation working. the linux engineer(s) I work with are just stumped at the moment on getting the new KEA dhcp server to provide all the same ISC dhcp v4 option handling that we want to carry into v6. any advice is welcome -Aaron On 10/9/2024 11:04 AM, Howard, Lee via NANOG wrote: > It's pretty high, at least in the U.S. > > https://stats.labs.apnic.net/ipv6/US <https://stats.labs.apnic.net/ipv6/US> > > Support in consumer electronics (TVs, game consoles) is weak, but a lot of > home gateways are fine. Netflix and YouTube stream over IPv6, and I think > Amazon Prime Video also does, but of course only if you're streaming to an > IPv6-capable device. > > https://www.vyncke.org/ipv6status/detailed.php?country=us > <https://www.vyncke.org/ipv6status/detailed.php?country=us> > > Definitely some laggards, but if you haven't looked in a while, you might be > surprised. > > > Lee > > -----Original Message----- > From: NANOG <nanog-bounces+leehoward=hilcostreambank....@nanog.org > <mailto:hilcostreambank....@nanog.org>> On Behalf Of Lucien Hoydic via NANOG > Sent: Tuesday, October 8, 2024 5:04 PM > To: nanog@nanog.org <mailto:nanog@nanog.org> > Subject: Re: CGNAT growing pains > > This message is from an EXTERNAL SENDER - be CAUTIOUS, particularly with > links and attachments. > > > > Anyone know the penetration rate of IPV6 for home users (cable modem)? I know > that some of the CPE doesn't even properly support IPV6 such as the stuff > being handed out by RCN/Astound. > > We just got our IPV6 allocation from ARIN and everything here is now dual > stack. Was relatively painless. > > On Tuesday, October 8th, 2024 at 3:19 PM, Jon Lewis <jle...@lewis.org > <mailto:jle...@lewis.org>> wrote: > >> >> I'm not so sure about that. Our customers are all offered dual-stack >> (DHCPv6, DHCPv6-PD). Do any of the common streaming services support >> v6 yet? Last I checked, Hulu did not. >> >> On Tue, 8 Oct 2024, Michael Thomas wrote: >> >>> Hi Jon, >>> >>> So is this easier than what the mobile carriers are doing -- >>> 464xlat, isn't it? Probably a sizeable portion of the traffic would >>> be running native v6, right? Obviously it wouldn't run into these sorts of >>> problems. >>> >>> Mike >>> >>> On 10/8/24 12:19 PM, Jon Lewis wrote: >>> >>>> We started rolling out CGNAT about 6 months ago. It was smooth >>>> sailing for the first few months, but we eventually did run into a >>>> number of issues. >> >> ---------------------------------------------------------------------- >> Jon Lewis, MCP :) | I route >> Blue Stream Fiber, Sr. Neteng | therefore you are _________ >> http://www.lewis.org/~jlewis/pgp <http://www.lewis.org/~jlewis/pgp> for PGP >> public key_________ -- -Aaron
