Shraddha, Multicast receivers signal their group membership request to just one router. Even if 5 and 6 in your diagram are transit Replication nodes, this would work only when their Replication SIDs are the same.
How about converting a MUST NOT restriction to a MAY clause, something like: "A Replication Node MAY use an Anycast SID or BGP PeerSet SID in segment list to send a replicated packet to one downstream Replication node in an Anycast set if and only if all nodes in the set have an identical Replication SID and reach the same set of receivers."? -Rishabh On Fri, Dec 9, 2022 at 9:50 AM Shraddha Hegde <shrad...@juniper.net> wrote: > Rishabh, > > > > Pls see inline.. > > > > > > Juniper Business Use Only > > *From:* Rishabh Parekh <risha...@gmail.com> > *Sent:* Friday, December 9, 2022 10:50 PM > *To:* Shraddha Hegde <shrad...@juniper.net> > *Cc:* Xiejingrong (Jingrong) <xiejingrong=40huawei....@dmarc.ietf.org>; > James Guichard <james.n.guich...@futurewei.com>; SPRING WG < > spring@ietf.org>; spring-cha...@ietf.org > *Subject:* Re: [spring] WGLC for draft-ietf-spring-sr-replication-segment > > > > *[External Email. Be cautious of content]* > > > > Shraddha, > > > > As I clarified to Jingron in an earlier email in this thread, this > restriction applies to Replication nodes, not to intermediate nodes on the > paths between upstream and downstream replication nodes. > > <SH> I did see your response earlier and I am not convinced with the > reasoning and hence reiterating > > > > Even if two replication nodes in an Anycast set or PeerSet have the same > Replication SID, a packet sent to them using Anycast SID or PeerSet will be > delivered to only one of these nodes. > > <SH> Yes the packet is delivered to only one node not to both of them. The > reason the controller included Anycast SID in segment list is for > redundancy . The actual multicast tree really needs only one of these two > nodes to replicate. > > > > > > Pls see example below > > > > The segment list consists of anycast-SID for 5 and 6 and a replication > SID. The actual multicast tree doesn’t require traffic to go to 5 and 6 but > it requires traffic to go to only one of them. Use of anycast SID is for > redundancy in case of failure. > > > > We will update the text in the next revision to specify the restriction > only applies to the downstream nodes. > > <SH> I really don’t understand what you mean by this. I am expecting you > to remove the MUST NOT restriction. > > If you don’t agree, would be good if you can provide example where the > restriction makes sense. > > > > -Rishabh > > > > On Fri, Dec 9, 2022 at 4:29 AM Shraddha Hegde <shrad...@juniper.net> > wrote: > > Authors, > > > > I have below comment on the draft. > > > > 1. > > An Anycast SID or BGP PeerSID MUST NOT appear in segment > > list preceding a Replication SID. > > > > The replication SIDs are similar to Binding-SIDs and Binding-SIds do not > have any > > such limitation. If there is an anycast SID before Replication SID, > > The two anycast nodes should have same replication SID for the > <Replication-ID, root-id> > > pair. Intelligent controllers may be able to achieve this. The "MUST NOT" > restriction > > seems unnecessary. > > > > 2. BGP PeerSID MUST NOT appear in segment list > > preceding a Replication SID > > > > I understand this is meant to be PeerSet SID. > > These kind of restrictions don't seem necessary. > > If a multicast tree need to be built across multiple domains > > then it may use BGP PeerSetSID for the inter-domain link as long as the > same replication SID is used > > on the nodes which correspond to the PeerSetSID. > > > > Rgds > > Shraddha > > > > > >
_______________________________________________ spring mailing list spring@ietf.org https://www.ietf.org/mailman/listinfo/spring
