Re: 6MAN Adoption Call for: draft-baker-6man-multi-homed-host-03

[Behcet Sarikaya <[e-mail]>]

On Tue, Oct 6, 2015 at 9:39 PM, Brian E Carpenter
<[e-mail]> wrote:
> Hi Behcet,
>
> On 07/10/2015 10:14, Behcet Sarikaya wrote:
>> Hi Ole,
>>
>> Here are my comments on this short (about 6 pages) draft.
>> First, I don't know if I am qualified to comment on a draft
>> co-authored by two ex-ADs and ex-IETF chairs, maybe I am not :-)
>>
>> Section 1.
>> It says:
>> The combination of existing recommendations for home
>>    gateways [RFC6092] [RFC7084] can also result in such filtering.
>>
>>  how could a set of  requirements text result in such a normative
>> looking behavior?
>
> I guess we mean "Implementations that combine existing recommendations..."
>
>> It says:
>>
>> Therefore, the only safe solution is to
>>    implement the features defined in this document.
>>
>>
>> It is not explained why (is it safe and secure)?
>
> It doesn't mean "safe" in the sense of security. It means "safe" in
> the sense that it will work in all scenarios. Needs rephrasing.
>
>> End of Section 1:
>> This section is normally about scope and related work. However the
>> related work discussion seems to have completely overlooked 6man WG
>> discussions that happened over quite long time period. Also related
>> solution proposals such as draft-pfister-6man-sadr-ra, etc.
>
> IMHO we can trace discussions of this problem back at least ten years
> (e.g. draft-huitema-multi6-ingress-filtering from 2004). If the WG adopts
> the draft and wants us to add a literature review, we can certainly do
> that (IMHO as an appendix to help the simplicity of the main text).
> Of course it would include your work and Pierre's.
>


Disagree.

I think you are missing the main point. Yes this topic goes way back
to Huitema's draft in 2004, but I think that the idea that the host
should do something has been introduced only recently.

No, I did not mean the whole literature survey, I meant past work
related to 6man should be included.

Behcet
>> Section 2.
>>
>> A host receives prefixes in a Router Advertisement [RFC4861]
>>
>> what about RFC 4191? 4191 also gives prefixes and the host may receive them?
>
> Good catch. We need to clarify that source-based next hop selection
> has priority over router preferences.
>
>>
>> Section 3.2
>>
>> This selection rule would be applicable in a host
>>    following the recommendation in the previous paragraph.
>>
>>  This is a single paragraph section. Which paragraph is the previous paragraph?
>
> It should be "previous section". This is an editing mistake.
>
>>
>> Lastly the title:
>>
>> Host routing in a multi-prefix network
>>
>> This title gives the impression that the draft is about "host routes"
>> while it seems like this draft has nothing to do with host routes.
>
> Yes, perhaps "Routing packets from hosts in a multi-prefix network"
> would be clearer.
>
> Thanks for the review. We'll hold these points until the WG adoption
> call is resolved.
>
>      Brian