Re: [Roll] Seed Set use across MPL Domains (was: Re: I-D Action: draft-ietf-roll-trickle-mcast-03.txt)

[Abdussalam Baryun <[e-mail]>]

Hi Esko,

comments below

On 1/30/13, Dijk, Esko <[e-mail]> wrote:
> Thanks Jonathan and Richard for this major update to the draft. The option
> to allow multiple MPL Domains to co-exist is a very useful addition. On this
> topic some questions / remarks:
>
> 1. adding "Seed Identifier" to the section 2 Terminology could help to
> clarify that a single MPL Forwarder/Seed has a single Seed Identifier. This
> SeedID remains identical across multiple MPL Domains it participates in,
> right?

I think each domain has its different seedID increment, otherwise, we
are not multicast within domains, if you mix incrementing you have
considered all domains, but this is conflict with the condition that
if outside domain we use IP-in-IP encapsulation.
>
> 2. I think for the text to be correct (sections 4.3, 6.2. 6.3, 7.3, 10.3,
> 11.X e.g. - any text where Seed Set is used in fact) there needs to be a
> separate Seed Set per MPL Domain a Forwarder participates in. Is this the
> solution? Peter already remarked the problem for sections 4.1/11.3.
>

I think that is good but still needs more, not sure :(

> Note that the Buffered Message Set already contains information on MPL
> Domain, somewhat hidden in the 'DataMessage' object in the tuple (Section
> 7.4). (The IPv6 Destination Address of this MPL Data Message identifies the
> MPL Domain). So for this data structure, there's perhaps no need to add an
> explicit "Domain identifier" to the tuple. But it would clarify the text to
> add such element to the tuple.

Why not, I think it is important parameter,

AB