Re: [Detnet] DetNet OAM drafts progressing

[Greg Mirsky <[e-mail]>]

Hi Balazs and Fabrice,
thank you for the discussion. I agree with the solutions proposed by
Fabrice. I've updated the draft and welcome your comments. Attached, please
find the diff and new version of the draft. Would these updates be
acceptable?

Regards,
Greg

On Wed, Jun 8, 2022 at 1:13 AM Fabrice Theoleyre <[e-mail]>
wrote:

> Dear Bala'zs,
>
> Please see online my comments.
>
>
> In today’s call we have discussed two topics regarding section 5
> “Maintenance”
> of the OAM framework document
> (https://www.ietf.org/archive/id/draft-ietf-detnet-oam-framework-05.txt):
>
> Sorry, I zapped the interim meeting. It was too late when I noticed it.
>
>
> A) Section 5.1.  Replication / Elimination
> “ … Each MEP
>    will decide to trigger the packet replication, elimination or the
>    ordering process when a set of metrics passes a threshold value.”
> We may need some clarification here. Changing of PREOF functions is
> a task for the DetNet controller, the MP (MEP or MIP) will only trigger
> the controller to take actions. An update might be needed with better
> phrasing.
>
>
> Actually, this scenario is rather for wireless links. When the link
> quality degrades (under a threshold), some additional features may be
> triggered locally. These features have been installed by the controller,
> but the node decides by itself to activate them.
> However, that’s rather for RAW. We can probably safely remove it from the
> detnet draft.
>
>
> B) Section 5.3.  Soft transition after reconfiguration
> The need for such a sub-section regarding OAM may worth for some
> discussions. PREOF itself is a tool for soft transition. It may the
> implementation of PREOF that allow the soft configuration changes,
> not the related OAM tools. So, the question for the WG: do we need
> OAM specific soft transition? If yes, for what (any example)?
>
> Any comments/views are appreciated.
>
>
>
> I had the feeling that the monitoring process has to be adapted after a
> soft reconfiguration.
> Else, states may be inconsistent.
>
> However, a metric may perfectly reflect the SLO of a complex path, or
> individual subpaths as well.
> Thus, I don’t find a relevant example where it would be required.
>
> Best regards,
> Fabrice
>
>