Re: [Roll] About merging two RFC drafts: draft-ietf-roll-enrollment-priority and DODAG size

["Pascal Thubert (pthubert)" <[e-mail]>]

Hello Huimin:

1) I agree that the (min) Priority in the draft could inherit from the DODAG size and is still quite different, since it could also reflect other gating factors such as an administrative willingness, the bandwidth availability / utilization near the Root, the memory in the Root more routes, etc... So constraining that the DODAG size derives from the priority degrades the value of the priority dramatically
2) I like the 12+4 bits representation that seems to bring the best of both worlds. I agree that it addresses the issues raised at the call.

Bottom line:
- I support the merge
- Is there another important information from the Root that we may want to add before we close the draft?
- to Michael's point, happy to contribute to the draft.

You all take care,

Pascal

> -----Original Message-----
> From: Huimin She (hushe) <[e-mail]>
> Sent: jeudi 4 février 2021 10:41
> To: Michael Richardson <[e-mail]>; [e-mail]
> Cc: [e-mail]; Pascal Thubert (pthubert) <[e-mail]>;
> Li Zhao (liz3) <[e-mail]>
> Subject: About merging two RFC drafts: draft-ietf-roll-enrollment-priority and
> DODAG size
> 
> Hi all,
> 
> Thanks for the nice discussions. Here are the answers for the questions
> raised in the interim meeting.
> 
> Question 1: Do we need the DODAG size as a complement to enrollment
> priority?
> Answer: Yes.
> 1a) Enrollment priority is a metric providing the abstraction of battery power,
> network bandwidth and available memory.
>       It represents the willingness of a 6LR to act as a joining proxy.
>       While DODAG size provides more information for a new node to decide
> which DODAG to join.
> 1b) The DODAG size is not only useful for a new node to choose a DODAG to
> join, but also useful for load balance between multiple DODAGs since nodes
> are able to migrate between DODAGs after knowing the DODAG size.
> 
> Question 2: Is 2-byte enough for encoding the DODAG size?
> Answer: Yes.
> 2a) If we directly use the 2-byte, the max size is 65535 which might not be
> enough.
> 2b) An alternative encoding format is as follows:
> - 4 bits for the unit. Let's set x = 0, 1, 2, ..., 15, so the unit equals to 2^x , i.e,
> unit = 1, 2, 4, 8, ..., 2^15.
> - 12 bits for the value.
> So the max number is 2^15 * 2^12, which is big enough for representing the
> DODAG size.
> 
> Please let me know your thoughts. Thank you.
> 
> Best regards,
> Huimin
> 
> On 2021/2/4, 01:30, "Michael Richardson" <[e-mail]> wrote:
> 
> 
>     Huimin She (hushe) <[e-mail]> wrote:
>         > Thanks for the nice discussion in the IETF interim meeting.
> 
>         > Attached please find the slides we presented. The draft is available
>         > here: https://github.com/roll-wg/dodag-metric
> 
>         > It is a good idea if our two drafts can be merged. How shall we
>         > proceed?
> 
>     As ietf-roll-enrollment-priority is already adopted, if it is the will of the
>     WG to include your metric, then merging it in would make sense.
>     I suggest a pull request on github at
>       https://github.com/roll-wg/draft-ietf-roll-enrollment-priority.git
>     would make sense.
> 
>     (Add Huimin She as a co-author, and Pascal too if desired)
> 
>     Maybe do the pull request first, and let's bring that discussion to the WG.
> 
>     --
>     ]               Never tell me the odds!                 | ipv6 mesh networks [
>     ]   Michael Richardson, Sandelman Software Works        |    IoT architect   [
>     ]     [e-mail]  http://www.sandelman.ca/        |   ruby on rails    [
>