On 7/1/24 10:32, Phillip Hallam-Baker wrote:
The fundamental problem here is that if I am running any sort of large mail service, I am evaluated on deliverability by my customers, not by the protocols I use. If 5% of email servers are IPv4 only, I have to support them or my deliverability will fail and I will get complaints. IPv4 addresses are not a scarce commodity as far as operators of mail services goes and so IPv4 will always be with us as far as SMTP is concerned.Remember BITNET? I remember people insisting that BITNET would never disappear, people who were arguing that BITNET was still growing (if slowly) every month... right up until it disappeared almost overnight.
Because once everyone on BITNET had IP service, people realized it wasn't worth maintaining BITNET in addition to maintaining their IP networks. This was true even though BITNET offered some services that didn't have direct IP equivalents. (I believe many of those BITNET sites were actually transmitting that traffic over IP by then, so they had already saved the cost of the leased phone lines that they had once used. It wasn't the communications cost so much as the administrative overhead that led to BITNET's demise.)
So I don't believe that IPv4 will always be with us, in any sense. Once IPv6 access is ubiquitous, people will start asking why they need to continue administering IPv4, except perhaps to communicate locally with legacy devices.
I'm actually in favor of explicitly deprecating the public IPv4 Internet, setting the date well in advance at which carriers start filtering IPv4 BGP advertisements and stop forwarding IPv4 traffic except by explicit arrangement. I nominate Jan 1, 2033. We need to stop insisting that sites must continue to administer two IP networks forever.
On Mon, Jul 1, 2024 at 11:06 AM Keith Moore <moore@xxxxxxxxxxxxxxxxxxxx> wrote:
I don't see that happening for SMTP because the big cost of managing SMTP services is the anti-abuse system, in fact that is pretty much the only cost. And going from 32 bits to 128 bits (or 64 if you want to look at it that way) is simply too much leverage to hand over to the attackers.
And that is also the reason I don't see SMTP worth fighting for. The principle of running open services that are not owned by a proprietary service provider is worth fighting for, the ability to use 50 year old protocols written when we didn't know what we were doing and didn't have the scale or the cryptography to solve the problems really isn't.
Your entire BITNET argument works much better for my plan than yours. If everyone moves to Mesh/Everything SMTP will go the way of the fax machine the way FTP has been replaced by SSH.
I do have an IPv6 strategy but it's part of my naming and IoT strategy.
The problem with our current setup is that ISPs don't actually provide email services any more, people have moved to services like Gmail because SMTP lacks address portability and so I stopped using my ISP's email when I moved house and had to change ISP.
Imagine a situation in which I get to critical mass with my scheme or someone else gets there instead of me. Providing the communications presence service in my scheme doesn't involve any of the cost center items it currently presents. I expect any rival would have to be likewise to beat me.
So now imagine that there is a ratings service for Mesh Service Providers. Level 1 is providing the minimal services needed to support device binding, messaging, etc. Level 2 adds persistent data storage so the user has a Terabyte rather than a Gig and so on.
While I can support all the features I want to support for IoT on IPv4, I can do it a lot better on IPv6 because I can issue every user IPv6 private address space along with their names. That means that if Alice and Bob want to connect up their non-Mesh appliances at different houses, I can glue them together with a bit of VPN and it's like they are on the same virtual LAN. And I can automate that whole process.
So to offer the Level 3 Platinum service, an ISP is going to have to provide IPv6. They will also have to support access to IPv4 only services but that can be through carrier grade NAT.
