ipv6-ops Digest, Vol 17, Issue 13

[Brent Sweeny]

As I read the list from S&D, they got /48s for *each* of their v6
facilities.  They list separate /48s for each one.

Another interesting (and frustrating) artifact at PAIX--I presume the
switch setup-- is that pings from one peer to another on the new
addresses don't work until both of them start speaking to each other.
All of the other peers there I've talked with are seeing the same
behavior: we don't see each other in ping or v6 neighbor replies until
at least we're both talking to each other and maybe until bgp is
flowing.  When we've both enabled bgp, however, it comes right up!
strange.				Brent Sweeny

>Date: Tue, 29 Aug 2006 19:52:33 +0200
>From: Daniel Roesen <dr at cluenet.de>
>Subject: Re: v6 peering at PAIX/Palo Alto
>To: IPv6 Ops list <ipv6-ops at lists.cluenet.de>
>
>>> Seems to me that if I was S&D, I'd be carving one /64 out of that /48
>>> for each exchange point fabric I operated.
>
>Actually, the idea is to have a /48 per IXP site as they are normally
>non-interconnected and thus standalone entities. The IXP /48s are
>assigned per IXP, not per operator.
>
>Best regards,
>Daniel