Re: [v6ops] draft-ietf-v6ops-64share WGLC

Le 19/09/2013 23:26, Teemu Savolainen a écrit :
>  > >Secondly, I don't agree DHCP server support is required, as the CE
> can include the DNS addresses in RAs as specified in RFC6106.
>  > >
>  >
>  > How could a CE convey NTP server settings to hosts using RAs?
> Accurate host time is pretty useful, if not essential.
>
> It cannot, but then again it probably would not help to have DHCP server
> implementation either, as, AFAIK, cellular networks commonly today do
> not provide DHCP service to terminals (to these "CEs" that would use
> 64share instead of DHCPv6 PD)....

In addition to considering whether or not the cellular network delivers 
DHCP, one should consider whether or not the local USB dongle delivers 
DHCP to the Host.

This or someone other unknown entity in the middle, between the network 
and the host, delivers DHCP and make that link look like ARP-capable 
link.  The QMI drivers and clients in linux see new evolutions in that 
sense.

But that is more like future work; 64share has place here where QMI is 
not yet widely deployed.

> If CE has no NTP settings, it cannot
> share them either. CE often gets time from cellular network, though.. So
> maybe CE should implement NTP server and with DHCP share its own address
> as NTP address?

CE to run a DHCPv6 Server?

> ... And this is why it is best to focus 64share to just prefix sharing;-)

In a sense I agree to let 64share just do the prefix sharing.

But obviously the client on the LAN side needs a way to get a DNS server 
address otherwise its connectivity is only partial.

Alex

>
> Best regards,
>
> Teemu
>
>
>
> _______________________________________________
> v6ops mailing list
> v6ops@ietf.org
> https://www.ietf.org/mailman/listinfo/v6ops
>