Re: [Gen-art] Review: draft-ietf-pim-join-attributes-for-lisp-05

[Dino Farinacci <[e-mail]>]

> Two. One at all LISP sites (not one per LISP site) and part of the overlay and possibly one in the underlay.
> [Lucy] Thank, Dino. This is helpful. So the attributes proposed in this draft is used in the first PIM instance, right? These attributes are only used by ETRs to send join/prune msg to ITR (root).

Right.

> site, one is for LISP underlay network, and one among LISP xTRs, is 
>> this right? PIM protocol instance running for LISP xTRs is a special 
>> version with additional LISP related attributes? When configuring
> 
> I don’t know what you mean by “additional LISP related attributes”. The LISP xTR runs PIM and sends unicast PIM Join/Prune messages to other xTRs.
> [Lucy] I mean the attribute is per LISP specific implementation.

Well, you could use VXLAN encapsulation as the data-plane (arguably, with 14 bytes of unnecessary overhead).

> PIM protocol on a device, do we need to differ them in configuration? Sorry to ask these, I have not yet read RFC6831 but understand what you want to achieve here.
> 
> That is implementation specific.
> [Lucy] PIM is a protocol running on top of IGP. Now we have PIM LISP implementation specific. Within a 

You are conflating the relationship between PIM and LISP. PIM runs side-by-side with BGP, OSPF, IS-IS, and LISP. That is the way to look at it. Do they depend on each other? The answer is no. PIM uses the RIB that is populated by BGP, OSPF, IS-IS. And LISP uses the mapping database to find RLOCs where then it uses the RIB/FIB to forward to RLOCs.

There is no case where PIM messages are encapsulated by LISP. I mean it can be but it is not spec’ed anywhere and is not needed unless some new design feature is desired by it (like PIM Join/Prune messages need confidentiality so they are encapsulated in lisp-crypto).

> But the best test to know if a multicast RLOC should be used is to RLOC-probe it and see all the ETRs that reply. Then you know which ETRs can received the encapsulated multicast packet via a multicast underlay. And the ones that don’t, you unicast replicate. This is the case where multicast partially deployed in the underlay.
> [Lucy] You are the expert on this subject. From the draft text, I am not able to picture these. 

Just taking a holistic view.

Dino