Re: [lisp] Confirm/Deny changes on draft 6830bis

[Alberto Rodriguez-Natal <[e-mail]>]

Thanks for the summary Albert. I can confirm a few points but I prefer
to abstain on those I don't have a strong opinion. Answers and
comments inline.

Thanks,
Alberto

On Fri, Jan 12, 2018 at 5:20 PM, Albert Cabellos
<[e-mail]> wrote:

> A.- Remove definitions of PA and PI

Confirm.

I'm fine with LISP not being so focused on the original Internet use-case.

> B.- Change definitions of EID and RLOC as ‘identifier of the overlay’ and
> ‘identifier of the underlay’ respectively.

Abstain.

I'd prefer to avoid using "identifier" in anything regarding the
underlay but I could be fine with the proposed text.

> C.- In section 5.3, change the description of the encap/decap operation
> concerning how to deal with ECN and DSCP bits to (new text needs to be
> validated by experts):

Confirm.

The new text looks good to me.

> D.- Simplify section ‘Router Locator Selection’ stating that the data-plane
> MUST follow what´s stored in the map-cache (priorities and weights), the
> remaining text should go to an OAM document.

Abstain.

I see the value of an independent OAM document but I don't see it as a
requirement to advance 6830bis.

> E.- Rewrite Section “Routing Locator Reachability” considering the following
> changes:
>
> *    Keep bullet point 1 (examine LSB), 6 (receiving a data-packet) and
> Echo-Nonce
> *    Move to 6833bis bullet point 2 (ICMP Network/Host Unreachable),3 (hints
> from BGP),4 (ICMP Port Unreachable),5 (receive a Map-Reply as a response)
> and RLOC probing

Abstain.

I'd personally prefer to move those mechanisms that rely on
control-plane messages (i.e. 5 - receive a MapReply) to the
control-plane document but I'd not oppose to keep them on the
data-plane doc.

> F.- Move Solicit-Map-Request to 6833bis

Confirm.

This is one of the major results I hope to see out of this discussion.

> G.- Move sections 16 (Mobility Considerations), 17 (xTR Placement
> Considerations), 18 (Traceroute Consideration) to a new OAM document

Abstain.

Same comment as above regarding a separate OAM document.