SUMMARY of http://www.ietf.org/internet-drafts/draft-terrell-math-ipaddr-ipv4-00.txt
By distinguishing apparently identical IP addresses by using different subnet masks one can increase the number of IP addresses distinguishable by a 32-bit number to greater than 2^32. No doubt. But... the IP packet have not _netmask_ field, and TCP/IP socket have not too. If you add this, it's easier to add extra address bits.
On the other hand, I can send the draft too -:). If we add 'PORT RANGE' field to the 'PTR' DNS record, and some trick to the 'xx.xx.xx.xx' address notation, we can split one IP address to the 4 - 8 hosts by allocating the different port ranges for every one. And it do not need to rewrite TCP stack and routers at all, only a little part in DNS and service resolver, or in the 'connect' and 'bind' function (and can be realised by the NAT just now. -:). There is not too big problem to increase IPv4 address space twise (cook one bit from the port field, and that's all). Through I wonder why people are making so many noice aroung unexisting IPv6 and don't try to improve existing systems a little... IPv4 have a few opportunities to create milti-level address hierarchy: - source routing - port/address mapping - netmask and AS numbers (for the routing only). Through after SNMP, MLPS etc I wonder to nothing...
Except for the problem that you need 32 extra bits to carry a mask or 5 extra bits to carry the masklen.
IETF should have waited until 2000/04/01 before posting this.
---------------------------------------------------------- Mike Bird Tel: 209-742-5000 FAX: 209-966-3117 President POP: 209-742-5156 PGR: 209-742-9979 Iron Mtn Systems http://member.yosemite.net/
Aleksei Roudnev, Network Operations Center, Relcom, Moscow (+7 095) 194-19-95 (Network Operations Center Hot Line),(+7 095) 230-41-41, N 13729 (pager) (+7 095) 196-72-12 (Support), (+7 095) 194-33-28 (Fax)