Exchange point requests for IPv6 address space

[bmanning at vacation.karoshi.com]

> 
> At 05:02 05/10/2001 -0400, Richard Jimmerson wrote:
> 
> >I believe it was discussed during the RIPE meeting that
> >not more than a /127 was needed to connect two peers at 
> >an exchange point.  
> 
> Based on the following definition of an exchange point, a /127 seems
> marginal, not so?

	Thats the same as assigning a /31 to connect two
	peers at an exchange.  This architectural 
	approach is roughly akin to a dense mesh of point2point
	links. Most exchanges do not fit this addressing model.

> >An Internet Exchange Point was defined as follows:
> >
> >3 or more ASes and 3 or more separate entities attached to a LAN (the
> >same infrastructure) for the purpose of peering and more are welcome
> >to join.

	Not bad, but there are a number of assumptions in there.
	As stated, its too restrictive.  There are other, viable
	exchanges using different architectural designs.

> For a shared medium, which seems to be implied, each interface would be
> assigned a value in the right hand 64 bits based, typically, on MAC address
> or the like.  So a /64 assignment would work (I agree with Randy), but
> would preclude expansion to provide subnetting.

	A /64 would work, for some architectures. For others,
	this might be problematic.

> Most providers greatly prefer exchanges to operate at Layer 2 rather than
> Layer 3, so maybe that's OK.  Perhaps I spoke too soon...  I guess that
> subnetting can indeed be excluded.

	Restricting choices will lead to a fairly easy to implement policy.

--bill