| Internet-Draft | PIM Light | February 2024 |
| Bidgoli, et al. | Expires 29 August 2024 | [Page] |
This document specifies a new Protocol Independent Multicast interface which does not need PIM Hello to accept PIM Join/Prunes.¶
This Internet-Draft is submitted in full conformance with the provisions of BCP 78 and BCP 79.¶
Internet-Drafts are working documents of the Internet Engineering Task Force (IETF). Note that other groups may also distribute working documents as Internet-Drafts. The list of current Internet-Drafts is at https://datatracker.ietf.org/drafts/current/.¶
Internet-Drafts are draft documents valid for a maximum of six months and may be updated, replaced, or obsoleted by other documents at any time. It is inappropriate to use Internet-Drafts as reference material or to cite them other than as "work in progress."¶
This Internet-Draft will expire on 29 August 2024.¶
Copyright (c) 2024 IETF Trust and the persons identified as the document authors. All rights reserved.¶
This document is subject to BCP 78 and the IETF Trust's Legal Provisions Relating to IETF Documents (https://trustee.ietf.org/license-info) in effect on the date of publication of this document. Please review these documents carefully, as they describe your rights and restrictions with respect to this document. Code Components extracted from this document must include Revised BSD License text as described in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the Revised BSD License.¶
It might be desirable to create a PIM interface between routers where only PIM Join/Prunes packets are signaled over it without having a full PIM neighbor discovery. As an example, this type of PIM interface can be useful in some scenarios where the multicast state needs to be signaled over a network or medium which is not capable of or has no need for creating full PIM neighborship between its Peer Routers. These type of PIM interfaces are called PIM Light Interfaces (PLI).¶
The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in BCP 14 [RFC2119] [RFC8174] when, and only when, they appear in all capitals, as shown here.¶
RFC [RFC7761] section 4.3.1 describes the PIM neighbor discovery via Hello messages. It also describes that PIM Join/Prune are not accepted from a router unless a Hello message has been seen from that router.¶
In some scenarios it is desirable to communicate and build multicast states between two L3 adjacent routers without establishing a PIM neighborship. There could be many reasons for this desired, but one example is the desired to signal multicast states upstream, between two or more PIM Domains via a network or medium that is not optimized for PIM or does not require PIM Neighbor establishment. An example is a BIER network connecting multiple PIM domains. In these BIER networks PIM Join/prune messages are tunneled via bier as per [draft-ietf-bier-pim-signaling].¶
A PIM Light Interface (PLI) ONLY accepts Join/Prune messages from an unknown PIM router and it accepts these messages without receiving a PIM Hello message form the router. Lack of Hello Messages on a PLI means there is no mechanism to learn about the neighboring PIM routers on each interface and their capabilities or run some of the basic algorithms like DR Election between the routers. As such the PIM light router doesn't create any General-Purpose state for neighboring PIM and it only process Join/Prune message from downstream routers in its multicast routing table.¶
Because of this, a PLI needs to be created in very especial cases and the application that is using these PLIs should ensure there is no multicast duplication of packets. As an example, multiple upstream routers sending the same multicast stream to a single downstream router.¶
As per IANA [iana_pim-parameters] PIM supports more than 12 message types, PIM Light only supports message type 3 (Join/Prune) from the ALL-PIM-ROUTERS message types listed in RFC7761, other unicast destination message types are supported by PIM light. All other message types are not supported for PIM Light and should not be process if received on a PLI.¶
The following should be considered on a PIM Light domain since hello messages are not processed.¶
Since PLI does not process the pim hello message, processing of the join attributes option in pim hello as per [RFC5384] is also not supported, leaving PIM Light unaware of its neighbor capability of processing the join attribute. A PIM Light Router that does not understand the type 1 Encoded-source Address and per [RFC7761] should not process a join message that contains it. Otherwise the PLI can process the Join Attribute accordingly.¶
Since DR Election is not supported on PIM Light because of lack of hello messages, the network design should ensure that DR Election is achieve on the PIM domain, assuming the PIM Light domain is connecting PIM domains.¶
As an example, in a BIER domain which is connecting 2 PIM networks, a PLI can be used between the BIER edge routers. The PLI will be only used for multicast states communication, by transmitting ONLY PIM Join/prunes over the BIER domain. In this case to ensure there is no multicast stream duplication the PIM routers attached on each side of the BIER domain might want to establish PIM Adjacency via [RFC7761] to ensure DR election on the edge of the BIER router, while PLI is used in the BIER domain, between BIER edge routers. When the Join or Prune message arrives from a PIM domain to the down stream BIER edge router, it can be send over the BIER tunnel to the upstream BIER edge router only via the selected designated router.¶
Where multiple PIM routers peer over a shared LAN or a Point-to-Multipoint medium, it is possible for more than one upstream router to have valid forwarding state for a packet, which can lead to packet duplication. When this is detected PIM Assert is used to select one transmitter. That said as per [RFC7761] PIM Assert should only be accepted if it comes from a known PIM neighbor. With PIM LIGHT the implementation SHOULD ensure there is no duplicate streams arriving from upstream PIM Light routers to a single downstream PIM LIGHT router. If this condition is not possible to be met because of network design, the implementation should ensure there is no duplication of stream. As an example in PIM LIGHT over a BIER domain implementation, for IBBR (Down stream PIM LIGHT router) in a BIER domain to find the EBBRs closes to the source (upstream PIM LIGHT routes), SPF can be use with a post processing as per [draft-ietf-bier-pim-signaling] Appendix A.1. With this post processing if 2 EBBRs are found by the downstream IBBR, then this down stream IBBR router can choose one of the EBBRs with a unique IP selection algorithm. As an example the EBBR with lowest IP address or largest IP address can be the EBBR that the downstream PIM LIGHT (IBBR) router sends the join/prune message to. When this EBBR goes offline the downstream router can send the join to the next EBBR based on the IP selection algorithm. This IP selection algorithm is outside of scope of this draft.¶
Since a PLI doesn't require PIM Hello Messages and PIM neighbor adjacency is not checked for join/prune messages, there needs to be a mechanism to enable PLI on interfaces for security purpose, while on some other interfaces this may be enabled automatically. An example of the latter is the logical interface for a BIER sub-domain [draft-ietf-bier-pim-signaling].¶
If a system explicitly needs a PLI to be configured, then this system should not accepts the Join/Prune messages on interfaces that the PLI is not configured on, and it should drop these messages on a non PLI interface. If the system automatically enables PLI on some special interfaces, as an example interfaces facing a BIER domain, then it should accept Join/Prune messages on these interfaces only.¶
Because the hello messages are not processed on the PLI, some failures may not be discovered in PLI domain and multicast routes will not be pruned toward the source on the PIM domain, leaving the upstream routers continuously sending multicast streams until the out going interface (OIF) expires.¶
Other protocols can be used to detect these failures in the PIM Light domain and they can be implementation specific. As an example, the interface that PIM Light is configured on can be protected via BFD or similar technology. If BFD to the far-end PLI goes down, and the Pim Light Router is upstream and is an OIF for a multicast route <S,G>, PIM should remove that PLI from its OIF list. In addition if upstream PLI is configured automatically, as an example in BIER case, when the downstream BFR is no longer reachable, the upstream PIM Light Router can prune the <S,G> advertised by that BFR, toward the source to stop the transmission of the multicast stream.¶
[RFC6559] defines a reliable transport mechanism for PIM transmission of Join/Prune messages. PIM over reliable transport (PORT) uses TCP port 8471 which is assigned by IANA. [RFC6559] mandates that if a router is configured to use PIM over TCP or SCTP on a given interface it must include the PIM-over-TCP-Capable or PIM-over-SCTP-Capable hello option in its Hello messages for that interface. These options also communicate the connection ID of TCP for the appropriate address family. PIM light lacking Hello messages, can be configured to use PORT under a PLI. That said the TCP connection ID of local router and peer router has to be configured manually under each side of the PLI. The PLI uses these local and peer connection ID to setup a TCP connection. As per [RFC6559] section 4 the routers use the connection IDs to figure out which side will do a passive transport open and which side of the PLI with do a active open. If TCP connection failed to open then PLI will revert back to Datagram mode.¶
Since PIM Light can be used for signaling Source-Specific and Sparse Mode Join/Prune messages, security considerations of [RFC7761] and [RFC4607] SHOULD be considered.¶
It should be noted a PIM Light can also use [RFC5796] as indicated in [RFC7761] for authentication.¶
Would like to thank Sandy <Zhang Zheng> for her suggestions and contribution to this draft.¶