/127 between routers?

[Scott Beuker]

> rfc4291 is quite clear that only /64s should be used on interfaces.
> rfc3627 acknowledges the operational reality that other lengths can
> and are being used on ptp interfaces (/127, /126, /120, /112.)
> 
> That leaves me in a state of confusion.

Me too, and vendors don't seem to have clear answers. It seems at this
point in time we're left to make these early decisions on instinct, and
hope it doesn't come back to bite us.

For me it breaks down to perceived risk versus perceived reward, and I'm
starting to wonder just how much I care about loopback addresses that
are 4-8 characters shorter. I certainly don't care much about saving
space on our backbone infrastructure in a world where your typical home
user will get a /56 and only has 3 hosts.

> I could assign /64s per ptp without difficulty. Should I also assign
> /64s per loopback?

This is the toughest one for me... extremely little traffic will be
forwarded to my ptp prefixes; relatively speaking, much more will be
destined to my loopbacks. So if it's the traffic to the prefix that
matters, should we then be a lot more worried about the loopback
prefixes?

> Should I really being using some form of EUI-64 for
> these addresses, to guarantee the interface ID is unique no matter
> what the network prefix? What happens when I replace hardware and
> EUI-64 changes ptp/loopback addresses?

For what it's worth, we'll use manually assigned addresses everywhere
for infrastructure. Right now, I'm even debating if we should manually
assign the link local addresses (i.e. FE80::1) where we can, to make
troubleshooting easier.

- Scott