It appears that Viktor Dukhovni <ietf@xxxxxxxx> said: >That of course has nothing to do with SMTP (the protocol) as such. >There are a few plausible tangential reasons: > > - Some receiving systems are more strict in what they're willing > to accept via IPv6 rather than IPv4. As a consequence, one can > choose to try to jump through more hoops, or just be sure to send > via IPv4. [ The extra scrutiny may be because with so much > address space in IPv6, IP reputation is perhaps less effective, > without additional signals. ] That's a big part of it. The IPv4 address space is small enough that mail systems can keep a separate reputation for every IP address and use that as part of the filtering rules. With v6 that's obviously impossible, particularly since it's not hard to hop to a new address for every message. While the obvious reponse is to aggregate by /64, that doesn't work because a lot of providers put multiple customers in a single /64 so you are at the mercy of whoever else happens to be in the rack with you. You don't need to tell me that they should give each customer their own /64 and some do but it's far from universal. > - IPv4 works well enough that the incentives aren't there to invest > scarce human cyles into bringing up outbound services on IPv6. >The issues are non-technical at this point, some major receiving systems >need to make it more (rather than less) attractive to send over IPv6. I might quibble about more vs. less. Google has stricter rules about what they will accept on v6 than on v4, but if you follow their rules, which is not hard if you're technically competent, they're more likely to deliver the mail you send. But that technically competent bit is the block here. R's, John
