Hi. /31 on ethernet links

[Bjoern A. Zeeb]

On Sat, 31 Oct 2009, Randy Bush wrote:

Hi,

>> However, I was simply reacting to the claim that it was *supported* by
>> Cisco.
>
> have you noticed a difference in the bug rate between things that are
> 'supported by cisco' and those that just happen to be there?  :)
>
> but you're right.  i liked.  our p2ps are /30s, not /31s.  and we're
> moving from /126 to /127.

I am sorry, I couldn't resist; I hope you won't take everything at
face value... though I hope you'll seriously think of some things...

Oh what /30 /31 bikeshed and how old it is?

I prefer to speak of p_t_p for point-to-point  in contrast to p2p for
peer-2-peer btw.  I seem to remember that it used to be like that but
unfortunately neither the vendors nor the people who are writing
(IETF) specs make a difference anymore.

I do not understand, though I know some, people who are not using a
/64 on an Ethernet IPv6 link; may it be ptp or not.  I know there is
an old enough bikeshed out about that as well as some prosposed
standards.  /127 really sounds fighting a system to me.
It's not that you couldn't address each atom in hour house already I'd
wildy guess with a /48 but ... some people always have trouble freeing
their mind from things that were like that 20 years and further back.
Have you ever thought of limiting your scoped link-local space on
Ethernet?  So why do you need valid IPs on your interfaces at all?
Why do you need more than a single global unicast address? Save your
IPv6 addresses for the neighbour's fridges and toasters.

/bz

-- 
Bjoern A. Zeeb       Even on Oct. 31st there is no candy with this mail.