apply clue with extreme force
for fuck's sake!
1) there are roughly twice as many people on this planet than there are ipv4 addresses. before you ever say anything about ip addressing again, come up with a viable solution that will give everyone just one ipv4 address. of course there will be a billion or two people who might never get connected but there will be at least that number who will each need more than one address.
2) nat is not the answer. it breaks too many things. like sip or video/audio streaming. try getting two or more people playing the same game over the internet at the same time when they all go through the same nat device.
3) even if nat was the answer and was guaranteed to work perfectly for every application and internet protocol forever, including the ones that have still to be invented, see 1).
4) net 10/8 is big enough for 16 million devices. the biggest telcos and cable companies have more than that number of customers already. this is why comcast, a us cable company, is all ipv6 now. they need ~300 million ip addresses: roughly 10-12 per customer (household). they have around 25 million customers today. they just can't hope to meet this with ipv4 address. vodaphone must be getting close to 16 million customers in england. if they're using nat, their network managers must be shitting themselves.
5) things like smart metering simply cannot work with nat. see 4). there are around 25 million gas meters in this country, most of them served by british gas. these won't fit into 10/8. the situation isn't quite as bad with electricity or water meters. oh, and you'll be seriously fucked because the meter will have to be renumbered (ie a site visit) whenever you switch providers => moving to a new utility company's wan. a nat solution (if it worked) would have that delightful property.
6) the intelligent grid will require end-to-end connectivity. nat breaks that. energy-hungry devices will have to be able to contact the power company to get real-time info about the cheapest and dearest times to power up. good luck making that work across the country with nat. or expecting everyone to reconfigure this mythical nat box in their house or office every time they plugged in a new kettle or telly.
7) anyone sitting on excess ipv4 space is unlikely to hand it back. now that those addresses are a scarce and almost exhausted commodity, carbon-based life-forms with a functioning brain will want to sell their spare addresses if they can. besides even once that ipv4 market starts, there still won't be enough addresses to go round. see 1).
8) the best thing vendors could do with nat is eliminate it. and apply clue to any fuckwits who think nat is the answer.
9) how many devices will be connected to the internet next year? more? less? same as now? what about in 5 years or 10 years? nat isn't going to save us. it will make things worse because all that nat shit will have to be ripped out and replaced with ipv6 some day. might as well have one migration to do instead of two.
10) every land-line will need a unique ip address at the exchange when the telcos switch to their next generation nets. incumbents like bt are already doing this. mobile operators won't be far behind. but they'll be connecting tablets and fondleslabs that sometimes get used to make phone calls. once you have 10+ million customers, network 10/8 and nat is just not going to do it. see 4).
11) iana handed out /8s to the regional internet registries. so it isn't worth handing back anything smaller than that to iana. and anyway, smaller chunks of free ipv4 space will be up for sale on ebay soon if they're not there already.
12) proper uptake of ipv6 puts a stop to all this nat fuckwittedness forever. and kills the trade in v4 addresses. it'll provide more than enough headroom for what we already expect we want to do on the internet for the next decade or two. and still leave vast amounts of unused space for whatever happens on the internet after our great-great grandchildren are long dead,
Biting the hand that feeds IT
