ietf
[Top] [All Lists]

Re: Sequentially assigned IP addresses--why not?

2000-08-10 22:20:02
As was pointed out recently, IPV6 will croak much sooner than it needs to
for the simple reason that we structure routing intelligence into the
address assignment.

don't believe everything you read on the IETF list.

Wouldn't it be better by far, to assign new addresses from 000...1, and map
to routing information however we may code it? 

who is going to maintain the mapping database?

(warning: if you say any of: network soutions, iCANN, IANA, the us 
government, or the ITU, I'll have to hurt you)

once we have the mapping database, to what are we going to map?

(let's see, we're going to need a compact representation of points
in the network topology such that most routers on the network can
decide how to route a packet to its address in the absence of much 
external information...I suggest a vector of bits which represent 
a tree structure, each level of branching using a variable number of 
bits (depending on the number of branches at this level) which indicate 
which branch to take.  for high-speed routing efficiency the bits 
need to be at a fixed offset in the packet, which generally requires
them to be fixed-length fields.  naturally we need fields that
are 'big enough' for some expansion.  with some care in management,
128 bits should do fine for awhile....)

Keith