Re: IPv6 like OSI is far more complex than necessary
The comment
"If IPv6 had been designed by engineers (rather than by theoreticians) it would have been much less complex - just increase the size of the addressing field by 2 bytes and map all existing IPv4 public addresses to IPv6 with the 2 additional address bytes being zero "
assumed me as I recall arguing some 20 years ago on IETF lists for precisely something like this (albeit more generally as variable length addresses). The argument put up against this was that "the community" was used to using mini-computers (PDP-11s, VAXes, etc) as routers and the algorithms worked on 16 or 32-bit lookups and anything new should be an extension of this i.e. an engineering argument - Engineers can be very conservative at times - and we ended up with a 128 bit (4 x 32) IPv6 address. The argument was rediculous - and I recall a CISCO Engineer pointing out that their routers didn't work this way anyway - and the tragedy was that shortly afterwards CIDR and BGP with its routing based on variable length prefixes completely invalidated the original argument - but to no effect.
There may also have been an aspect of NIH as variable length addressing was an OSI paradigm and not an IP one.
However, the underlying problem has less to do with address sizes and more to do with the problem that a properly worked through transition plan has never been on the table for IPv6. A transition plan requires that IPv4 and IPv6 ASs must be able to co-exist and interwork for an extended period of time and possibly forever. It can be done, but almostly certainly involves NAT and the use of the DNS to manage IPv6 to IPv4 address mapping. However, for the "ivory tower types" referred to in the article, NAT is evil and must be erradicated and IPv6 must become universal. A NAT based transition plan would be unacceptable as would the effective result: that IPv6 could well end up as an inter AS protocol only. Hence they are not interested - even though that is all we really need.
The final problem is that there is no financial incentive for ISPs to be the first to adopt IPv6. To them it is a cost without a benefit in terms of offering cheaper services.
IPv4 allows for 2**32 addresses. That's a big number and it will always be possible to spread it out just that bit more thinly. On the other hand, there's an inverse relationship between the efficiency of address allocation and routing efficiency. The more efficient (i.e. more densely allocated) IPv4 address allocation becomes, the less efficient inter AS routing will become and dealing with the 512K by just adding more memory will be no more than the networking equivalent of kicking the can down the road. Unfortunately, that is probably what will happen.
Biting the hand that feeds IT
