Re: [v6ops] Review: draft-ietf-v6ops-unique-ipv6-prefix-per-host

Our comment was that using a unique /64 per UE would considerably increase
the amount of wasted addresses. It could be a problem in some deployments,
but the document doesn't address this potential issue.

I agree that using /48 makes planning easier, but this is also the standard
size of an enterprise allocation, so a company with this standard
allocation might have problems in following this recommendation.

Edwin Cordeiro

On Mon, May 23, 2016 at 1:40 PM, Ross Chandler <ross@eircom.net> wrote:

>
> On 23 May 2016, at 16:24, Edwin Cordeiro <edwinsc@gmail.com> wrote:
>
> Major Issues:
> - The DHCPv6-PD is mentioned for the first time only in section 4.3.1 and
> later at the "Future Work" session. From my understanding DHCPv6-PD is
> capable of implementing the network as desired in this BCP. Considering the
> intended status is BCP, I think that DCHPv6-PD should be added as valid
> implementation alternative.
> - The draft doesn’t mention possible issues of giving one /64 for each UE,
> for example, a public WIFI with support for 512 users will need a /55
> instead of a single /64.
>
>
> Edwin,
>
> Some comments on those comments as we’re planning on deploying WLAN-GWs
> that give out /64 per UE.
>
> /55 is probably unnecessarily parimonious and it isn’t on a nibble
> boundary to make the address plan easy to understand.
> I’d suggest a minimum of /48 for a pool giving out /64s. A larger pool
> size will help avoid bringing the IPv4 pool exhaustion problem seen in BNGs
> into an average WLAN-GW.
>
> Regards
> Ross
>
>
>
>
> Minor Issues and Nits:
> The document has some Nits problems as detailed in:
> https://tools.ietf.org/idnits?url=https://tools.ietf.org/id/draft-jjmb-v6ops-unique-ipv6-prefix-per-host-00.txt
>
> Questions:
> - Should this document address the best way to logging the prefix assign
> to the UE/subscriber?
> - Assigning the same unique prefix if the UE/subscriber reconnect to the
> same AP, is something that should be considered or avoided?
>
> Best regards,
>
> Edwin Cordeiro
> _______________________________________________
> v6ops mailing list
> v6ops@ietf.org
> https://www.ietf.org/mailman/listinfo/v6ops
>
>
>