IPv6 multihoming

[Daniel Roesen]

On Sat, Feb 05, 2011 at 02:52:37PM +0100, Bernd Walter wrote:
> So multihoming a /48 PA or allow another ISP to announce it during
> transition phase is already considered bad practice?

The point of PI is Provider INDEPENDENCE. Using PA more-specifics
doesn't deliver that. Not for the renumbering problem, but also about
redundant global reachability as there will always be a lot of networks
who will filter more-specifics in PA space. This will make you still
depend on the ISP from which you get the PA space as that is the only
route visible to those filtering networks.

This is why PA more-specific multihoming conceptionally broken. It
doesn't deliver independence nor real redundency.

> You differentiate PI from PA space by using RIR databases?

Yes. Technically there is no difference anyway. RIRs have set aside
blocks from which they issue PI prefixes. All else is considered PA.

Best regards,
Daniel

-- 
CLUE-RIPE -- Jabber: dr at cluenet.de -- dr at IRCnet -- PGP: 0xA85C8AA0