Re: [DNA] DNAv6 Review from Mobopts

[JinHyeock Choi <jinchoe@gmail.com>]

Dear Wes

> The reason cable MSO's use multi-link subnets is that there is no
> connectivity between homes (for security and bandwidth purposes), so the
> alternative is that every home gets its own prefix.  Cable MSO's have
> chosen instead to share one prefix for every 100,000 homes.  This is to
> conserve IPv6 address space, which may actually be more limited than you
> might think. (Try requesting some someday! I had a lot of difficulty
> getting a /55 when I needed one!)  The powers that be wisely want to
> control the IPv6 address situation so that we don't end up in the same
> mess as we have for IPv4.  Given the perceived shortage of IPv6
> addresses, DHCPv6 PD service for 100 million homes may be harder to
> obtain than you might think.

hmmm. From 3GPP and WiMAX examples, i.e. generously allocating a
prefix per a mobile node, I assumed that it won't be necessary to
share a prefix among homes. Let's see.

> > Moreover, I don't see much need for DNA when a mobile node moves
> > between different homes. DNA is to reduce movement detection time to
> > achiever seamless handoff but, when you visit a house, at least you
> > should ring a bell and wait for a reply. :-)
>
> Good point!  Which means it may be perfectly acceptable to say "the
> transition will not be seamless when you move between houses" or "you
> may have to reset your network card when you move between houses" - but
> that means that the failure mode should be that you get no connectivity
> and it requires manual intervention rather than it "seamlessly thinks it
> works but actually creates mysterious, hard to understand behavior"
> which I think is what the current spec does.  That's why I say that one
> perfectly acceptable behavior is just to revert to regular ND (or have
> the option of regular ND) and require manual intervention in case of
> these types of network changes.

The problem is that regular ND doesn't talk much about how to
determine if a host's existing address is still valid. As of my
knowledge, there are two ways in ND,
1) check whether the prefix of the address is advertised in the
currently attached link.
2) check whether an existing default router is still reachable.

1) gives a problem in multi-link subnets as you pointed out.
2) gives a problem because of link-local scope of router address.

thanks for your sharing ideas with us. I wish you would also comment
on the existing simple DNA solution to facilitate its advancement.

Best Regards

JinHyeock