Re: [v6ops] Implementation Status of PREF64

[Alexandre Petrescu <[e-mail]>]

Le 13/10/2021 à 22:00, STARK, BARBARA H a écrit :
>     >> The predictability of assigned/managed addresses is less a problem for your
>     >> arbitrary client/phone and more of an issue for the potential use case, for
>     >> example, of IoT/instrumentation devices where the endpoints are perceived
>     >> to need more active management than the typical phone.
>     > 
>     > And this is a typical role for an Android mobile device?
> 
>     A lot of IOT development is using android as a base OS, yes.
> 
>     Mobile device? Not necessarily, but lots of IoT with WiFi or
>     Ethernet only interfaces
>     do use android as a base OS.
> 
>     > And this is so important that a general purpose mobile device can not do
>     > IA_NA/IA_TA, because some special-purpose IoT hub would need something
>     > else?
> 
>     I think he’s arguing that android should do IA_NA/IA_TA not only
>     because of the
>     situations that desire it for general purpose mobile devices, but
>     also because
>     it’s even more important for IOT things that are being built on the
>     android platform.
> 
>     I could be wrong, Clark can certainly correct me if I misunderstood
>     his concern.
> 
>     Owen
> 
> I can speak for Clark, but IOT and other things that are not general 
> purpose mobile devices are what I'm thinking about.
> 
> <bhs> When I look at the Nook eReader that uses the Android OS, I notice 
> that it just references a bunch of Open Source Apache 2.0 licenses for 
> the Android OS components they use. To me, this says IoT devices are in 
> no way beholden to a Google app store or Google restrictions around what 
> goes into an Android smartphone. I see on 
> https://source.android.com/devices/architecture/modular-system/networking <https://source.android.com/devices/architecture/modular-system/networking> 
> that it says Android 10 included a DhcpClient component that “obtains IP 
> addresses from DHCP servers so they can be assigned to interfaces”. So 
> it sounds like IoT devices that use Android already have the ability (if 
> the manufacturer wants) to support IA_NA?

Supposedly that DHCP in Android is v4, not v6.

The use of that v4 feature has several implications:

- IoT on IPv4 is a widespread reality, even though IoT at IETF is a
   completely different thing.  It  might be that when computers move to
   IPv6 at that scale 'IoT' itself will no longer be a buzzword, long
   replaced by something else that uses IPv6.

- IoT Android routers with IPv4 necessarily use NAT, this NAT being a
   feature one would not like to see propagated in IPv6.  I think Lorenzo
   says that NAT66 might come to IPv6 IoT if DHCPv6 was used, so he
   advises to not use DHCPv6.  To that, I would say that NAT66 comes to
   IPv6 IoT already and it is because of the '/64' limit, or because of
   lack of the DHCPv6 feature in Android.

To bring IPv6 to Android to IoT routers one necessarily needs to get
away from the '/64' limit or to use DHCPv6 on Android.  Until then IoT
Android is IPv4, and IoT Android routers are NATs.

Alex

> 
> 
> _______________________________________________
> v6ops mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/v6ops
>