Re: [v6ops] Implementation Status of PREF64

[Philip Homburg <[e-mail]>]

>It's not. Lorenzo has explained why many times, including in the last week
>or so on this mailing list.
>
>Lorenzo has worked out like I have worked out that reducing external
>dependencies (such as a dependency on a central server) makes things more
>reliable, and also makes deploying new capabilities to hosts easier and
>quicker.

I'll try to remember that this is now the reason the Android doesn't
support DHCPv6. For me it doesn't pass the laugh test.

Basically Android devs are telling operators that they are stupid. People
who have been running big IPv4 networks for decades don't know what they
are doing.

It is trivial to run DHCP in a decentralised mode. Every CPE is effectively
doing that. But somehow, misguided operators never figured that out and
insist on a central database. A cause that really needs to be faught by a
mobile handset OS.

In a typical 802.1x installation, credentials are stored in a central server.
So even getting access requires the central server to be up and running.
Including all the paths between the wifi access point and that server.
There are some hints that it would be good if every wifi host would get a
unique /64. Likely in large environments, this prefix would come from a
central server and there is nothing Android can do about that.

But they can still show their commitment to the cause by not supporting
DHCPv6. 

It seems rather sad to me.

I'm a bit confused about the 'deploying new capabilities to hosts'. 
Historically, DHCP has been very flexible in picking up new features.
The reason for that is that only the DHCP server need to serve them.
All other components, like routers are unaware.

Recently, 6man has a discussion about a generic RA option, in particular
because it is so hard to get agreement on new RA options. Let alone 
get vendor support for options that are added.

So it seems the complete opposite: if you want new capabilies quickly, 
go for DHCP. If you want to wait forever for your router vendor to support
something new, go for RA.