Markku Savela <[EMAIL PROTECTED]> wrote:
|> Fine by me. It's just that dealing with scopes seems to be the problem that
|> most people are complaining about rather than the existence of the addresses
|> themselves.
|
|I cannot understand those people complaining about scopes.
They are complaining because a routing complexity issue has been pushed onto
applications. I have trouble working up a lot of sympathy for the complaints,
though. If they had put this much effort into complaining about the address
architecture in the first place we could have solved the routing problem at
the routing layer.
| - you don't get addresses from your ISP, everyone/every site is
| automaticly entitled to unique global address range. When one makes
| a contract with ISP about route, the ISP will make it packets with
| these addresses flow correctly.
Yes, that's how things worked in the good old days.
| - of course, any random address generation won't scale.
That isn't clear. Central routing table size scales linearly with the
number of addresses. Linear isn't as bad as some people make it out to
be. The aggregation hack was a bad tradeoff because while it reduced the
table problem from linear to log it changed the address usage from linear
in users to exponential in provider chain height. It's very hard to win
against an exponential.
Fortunately, it is very easy to distribute the routing process and reduce
the problem to a constant. So even if linear worries you there is an
alternative. I outlined a simple way to distribute the routing in a previous
message. It's not as neat as building it into the routing layer, but it
works without changing the infrastructure.
|But, the
| proposals for generating unique addresses from geographical
| position coordinates might be one way, and might scale.
|
|So, let's scrap the current address allocation and just define rules
|how to generate address from location (and maybe need a floor number,
|or height for buildings.. :-)
That would certainly be one approach. Anything that leaves sites with some
way to control address space will ultimately support a tunneling solution
independent of whether the ISPs allow the addresses to be routed.
Dan Lanciani
ddl@danlan.*com
--------------------------------------------------------------------
IETF IPng Working Group Mailing List
IPng Home Page: http://playground.sun.com/ipng
FTP archive: ftp://playground.sun.com/pub/ipng
Direct all administrative requests to [EMAIL PROTECTED]
--------------------------------------------------------------------
