Re: [pim] [spring] WGLC for draft-ietf-spring-sr-replication-segment

[Rishabh Parekh <[e-mail]>]

Sasha,
Inline @ [RP]

On Sun, Dec 11, 2022 at 3:43 AM Alexander Vainshtein <
[e-mail]> wrote:

> Hi all,
>
> More questions regarding the definition of the Replication State of a
> Replication Segment:
>
>
>
> The draft states that:
>
>
>
> If a Downstream Node is an egress (aka leaf) of the multi-point service,
> i.e. no further replication is needed, then that leaf node's Replication
> segment will not have any Replication State…
>
>
>
> IMHO this is not aligned with the definition of the elements of the
> Replication Segment (already quoted below) and Replication State. It would
> be more appropriate to express the behavior of a Leaf node by stating that
> its Replication state (defined as quoted below) is *an empty list of
> branches*.
>
>
>
> With this approach a Replication SID of a bud node of a service could be
> defined as  an SID with the Replication state including,  as one of the
> branches – but not the only branch in the list - a Replication SID with
> itself as the Node-ID and with an empty list of branches.
>

[RP] The text " leaf node's Replication segment will not have any
Replication State…" effectively means an empty list of branches with an
indicator of Leaf role in a Replication Segment. A bud node is a combination

> My question is:
>
>
>
> Can the same "Downstream Replication ID in a given "Downstream Node"  be
> included/ in the Replication state of multiple Replication segments,
> especially of Replication segments identified by different Node IDs?
>

[RP] I assume you meant to ask  "Can the same Downstream Replication *SID*
in a given ...." .

>
>
> I have not found any answer to this question in the text of the draft. At
> the same time I think that a positive answer to my question would
> contradict the definitions in Section 3.1 of the MVPN and EVPN with SR
> P2MP and Ingress Replication draft
> <https://datatracker.ietf.org/doc/html/draft-ietf-bess-mvpn-evpn-sr-p2mp-06#name-mpls-label>
> because these definitions assume that the Replication SID can be used as
> the "context label" for resolving the context label space space for looking
> up the upstream allocated label advertised in the PTA attribute of the
> suitable Mutlicast VPN route (in the case of aggregated P-tunnels).
>
>
>
> IMHO and FWIW an explicit and unambiguous answer to my question should be
> provided by the authors in order to advance the draft.
>

[RP] The answer is theoretically yes, but that would be like trying to
share a Replication Segment across different SR-P2MP trees (described in
PIM WG SR P2MP draft). This requires a complex set of conditions that have
to be met to make it work and hence we have kept it out of scope of this
draft, PIM SR P2MP draft and the BESS MVPN/EVPN draft.

Just to clarify, Section 3.1 of BESS MVPN/EVPN covers the case where a
*single* SR-P2MP tree is shared across two or more MPVNs and hence an
upstream (or centrally) assigned "context" is required in packet encoding
to resolve the MVPN in which payload is processed. When a SR-P2MP tree is
associated with only one MPVN, this upstream "context" not required in
packet; the Replication SID at Leaf/Bud is sufficient to provide the MVPN
context in which payload is processed.

I hope I have answered your questions,
-Rishabh