[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: AT&T Public Policy Research -- hiring for cypherpunks



Alan Olsen wrote:

| >	Addresses will not be easily 'transferable.'  The IETF is
| >discussing a 'Best Current Practices' document that talks about
| >address portability.  Basically, it can't happen, because the routers
| >only have so much memory, and the routers at the core of the internet
| >can't keep in memory how to reach every one; there needs to be
| >aggregation.  The only feasible aggregation seems to be provider
| >based, ie, MCI, Alternet, and other large ISPs get blocks of
| >addresses.  They give them to smaller companies, like got.net, which
| >gives them to customers.  The result?  The core routers have a few
| >more years.
| 
| A good point. Having parts of subnet shifting around could be pretty painful
| from an admin point of view.

Its not an admin's point of view thats worrisome.  Whats worrisome is
that the routers at the core of the net only have so much memory, and
if the routing tables grow beyond that, we're all hosed, becuase the
core of the internet will start thrashing.  So, in essense, you taking
your network address with you when you switch providers ('address
portability' causes costs that must be borne by the entire global
internet.


| What is the timeline for implementation of IPv6?

	Good question.  I think the address allocation just went to
last call, which means that we should have a policy for getting IPv6
addresses pretty soon.  After that, you need to wait for your router
vendor to announce an IPv6 capable version.  I'd guess it will be six
to eighteen months before you can call Netcom and ask for an IPv6 PPP
connection.

Adam

-- 
"It is seldom that liberty of any kind is lost all at once."
					               -Hume