On Sep 24, Thomas Bushnell BSG <tb@becket.net> wrote: > > How it is assigned is obviously (?) not important in this context. An IP > > address is usually considered "dynamic" if it is shared among many users > > (so NAT outside addresses may qualify as well). > So this is a different definition of "dynamic" than Adam's. I'm not > particularly interested in the semantic question of which is the > "right" definition; I'll use whichever. If we are discussing the merits of using DULs then I'd say that knowing how they work is a major prerequisite to having an useful discussion. > Short of asking the ISP, there is no way of knowing if an IP address > is shared. This is correct, but as I explained there are some good heuristics which work well enough to make a well-maintained DUL being worth using on the production mail servers of large ISPs and organizations. > I would note here that many many ISPs make a good-faith > effort to assign the same dynamic address to the same MAC address > every time, provided it is still available. This depends on the ISP > and their user population, and the ratio of users to dynamic addresses > of course, but it is still not uncommon. Actually in my experience this is uncommon, and when true most users will not take advantage of this to keep an address for extended periods. > Moreover, if my computer is always online, and my ISP promises to > renew leases rather than force a new address assignment when address > leases expire, then my IP address is, in fact, not shared in any > meaningful way. Yes. But I'm sure you can understand how making an exception for every user who says this is not really feasible. > > Very good ISPs report dynamic pools themselves to the DULs. > > Good ISPs mark dynamically-assigned addresses in the whois database or > > in the rDNS. > Why is this "very good"? Can you document for me which RFC recommends > or encourages this as a best practice? It is very good because customers of these ISPs will not risk having their statically assigned IP addresses being mistaken. I am not aware of RFCs on this subject, but people with experience in email deliverability issues will confirm this. > Curious. So MIT is a bad ISP? MIT is not an ISP. -- ciao, | Marco | [8184 erE1aVpplsnDA]
Attachment:
signature.asc
Description: Digital signature