Network Working Group D. Estrin Request for Comments: 2362 USC Obsoletes: 2117 D. Farinacci Category: Experimental CISCO A. Helmy USC D. Thaler UMICH S. Deering XEROX M. Handley UCL V. Jacobson LBL C. Liu USC P. Sharma USC L. Wei CISCO June 1998 Protocol Independent Multicast-Sparse Mode (PIM-SM): Protocol Specification Status of this Memo This memo defines an Experimental Protocol for the Internet community. It does not specify an Internet standard of any kind. Discussion and suggestions for improvement are requested. Distribution of this memo is unlimited. Copyright Notice Copyright (C) The Internet Society (1998). All Rights Reserved.
1 Introduction This document describes a protocol for efficiently routing to multicast groups that may span wide-area (and inter-domain) internets. We refer to the approach as Protocol Independent Multicast--Sparse Mode (PIM-SM) because it is not dependent on any particular unicast routing protocol, and because it is designed to support sparse groups as defined in . This document describes the protocol details. For the motivation behind the design and a description of the architecture, see . Section 2 summarizes PIM-SM operation. It describes the protocol from a network perspective, in particular, how the participating routers interact to create and maintain the multicast distribution tree. Section 3 describes PIM-SM operations from the perspective of a single router implementing the protocol; this section constitutes the main body of the protocol specification. It is organized according to PIM-SM message type; for each message type we describe its contents, its generation, and its processing. Sections 3.8 and 3.9 summarize the timers and flags referred to throughout this document. Section 4 provides packet format details. The most significant functional changes since the January '95 version involve the Rendezvous Point-related mechanisms, several resulting simplifications to the protocol, and removal of the PIM-DM protocol details to a separate document  (for clarity). 2 PIM-SM Protocol Overview In this section we provide an overview of the architectural components of PIM-SM. A router receives explicit Join/Prune messages from those neighboring routers that have downstream group members. The router then forwards data packets addressed to a multicast group, G, only onto those interfaces on which explicit joins have been received. Note that all routers mentioned in this document are assumed to be PIM-SM capable, unless otherwise specified. A Designated Router (DR) sends periodic Join/Prune messages toward a group-specific Rendezvous Point (RP) for each group for which it has active members. Each router along the path toward the RP builds a wildcard (any-source) state for the group and sends Join/Prune messages on toward the RP. We use the term route entry to refer to the state maintained in a router to represent the distribution tree. A route entry may include such fields as the source address, the group address, the incoming interface from which packets are accepted, the list of outgoing interfaces to which packets are sent,
timers, flag bits, etc. The wildcard route entry's incoming interface points toward the RP; the outgoing interfaces point to the neighboring downstream routers that have sent Join/Prune messages toward the RP. This state creates a shared, RP-centered, distribution tree that reaches all group members. When a data source first sends to a group, its DR unicasts Register messages to the RP with the source's data packets encapsulated within. If the data rate is high, the RP can send source-specific Join/Prune messages back towards the source and the source's data packets will follow the resulting forwarding state and travel unencapsulated to the RP. Whether they arrive encapsulated or natively, the RP forwards the source's decapsulated data packets down the RP-centered distribution tree toward group members. If the data rate warrants it, routers with local receivers can join a source-specific, shortest path, distribution tree, and prune this source's packets off of the shared RP-centered tree. For low data rate sources, neither the RP, nor last-hop routers need join a source-specific shortest path tree and data packets can be delivered via the shared, RP-tree. The following subsections describe SM operation in more detail, in particular, the control messages, and the actions they trigger. 2.1 Local hosts joining a group In order to join a multicast group, G, a host conveys its membership information through the Internet Group Management Protocol (IGMP), as specified in , (see figure 1). From this point on we refer to such a host as a receiver, R, (or member) of the group G. Note that all figures used in this section are for illustration and are not intended to be complete. For complete and detailed protocol action see Section 3. [Figures are present only in the postscript version] Fig. 1 Example: how a receiver joins, and sets up shared tree When a DR (e.g., router A in figure 1) gets a membership indication from IGMP for a new group, G, the DR looks up the associated RP. The DR creates a wildcard multicast route entry for the group, referred to here as a (*,G) entry; if there is no more specific match for a particular source, the packet will be forwarded according to this entry. The RP address is included in a special field in the route entry and is included in periodic upstream Join/Prune messages. The outgoing interface is set to that included in the IGMP membership indication for the new member. The incoming interface is set to the interface used to send unicast packets to the RP.
When there are no longer directly connected members for the group, IGMP notifies the DR. If the DR has neither local members nor downstream receivers, the (*,G) state is deleted. 2.2 Establishing the RP-rooted shared tree Triggered by the (*,G) state, the DR creates a Join/Prune message with the RP address in its join list and the the wildcard bit (WC- bit) and RP-tree bit (RPT-bit) set to 1. The WC-bit indicates that any source may match and be forwarded according to this entry if there is no longer match; the RPT-bit indicates that this join is being sent up the shared, RP-tree. The prune list is left empty. When the RPT-bit is set to 1 it indicates that the join is associated with the shared RP-tree and therefore the Join/Prune message is propagated along the RP-tree. When the WC-bit is set to 1 it indicates that the address is an RP and the downstream receivers expect to receive packets from all sources via this (shared tree) path. The term RPT- bit is used to refer to both the RPT-bit flags associated with route entries, and the RPT-bit included in each encoded address in a Join/Prune message. Each upstream router creates or updates its multicast route entry for (*,G) when it receives a Join/Prune with the RPT-bit and WC-bit set. The interface on which the Join/Prune message arrived is added to the list of outgoing interfaces (oifs) for (*,G). Based on this entry each upstream router between the receiver and the RP sends a Join/Prune message in which the join list includes the RP. The packet payload contains Multicast-Address=G, Join=RP,WC-bit,RPT-bit, Prune=NULL. 2.3 Hosts sending to a group When a host starts sending multicast data packets to a group, initially its DR must deliver each packet to the RP for distribution down the RP-tree (see figure 2). The sender's DR initially encapsulates each data packet in a Register message and unicasts it to the RP for that group. The RP decapsulates each Register message and forwards the enclosed data packet natively to downstream members on the shared RP-tree. [Figures are present only in the postscript version] Fig. 2 Example: a host sending to a group If the data rate of the source warrants the use of a source-specific shortest path tree (SPT), the RP may construct a new multicast route entry that is specific to the source, hereafter referred to as (S,G) state, and send periodic Join/Prune messages toward the source. Note that over time, the rules for when to switch can be modified without
global coordination. When and if the RP does switch to the SPT, the routers between the source and the RP build and maintain (S,G) state in response to these messages and send (S,G) messages upstream toward the source. The source's DR must stop encapsulating data packets in Registers when (and so long as) it receives Register-Stop messages from the RP. The RP triggers Register-Stop messages in response to Registers, if the RP has no downstream receivers for the group (or for that particular source), or if the RP has already joined the (S,G) tree and is receiving the data packets natively. Each source's DR maintains, per (S,G), a Register-Suppression-timer. The Register- Suppression-timer is started by the Register-Stop message; upon expiration, the source's DR resumes sending data packets to the RP, encapsulated in Register messages. 2.4 Switching from shared tree (RP-tree) to shortest path tree (SP-tree)} A router with directly-connected members first joins the shared RP- tree. The router can switch to a source's shortest path tree (SP- tree) after receiving packets from that source over the shared RP- tree. The recommended policy is to initiate the switch to the SP-tree after receiving a significant number of data packets during a specified time interval from a particular source. To realize this policy the router can monitor data packets from sources for which it has no source-specific multicast route entry and initiate such an entry when the data rate exceeds the configured threshold. As shown in figure 3, router `A' initiates a (S,G) state. [Figures are present only in the postscript version] Fig. 3 Example: Switching from shared tree to shortest path tree When a (S,G) entry is activated (and periodically so long as the state exists), a Join/Prune message is sent upstream towards the source, S, with S in the join list. The payload contains Multicast- Address=G, Join=S, Prune=NULL. When the (S,G) entry is created, the outgoing interface list is copied from (*,G), i.e., all local shared tree branches are replicated in the new shortest path tree. In this way when a data packet from S arrives and matches on this entry, all receivers will continue to receive the source's packets along this path. (In more complicated scenarios, other entries in the router have to be considered, as described in Section 3). Note that (S,G) state must be maintained in each last-hop router that is responsible for initiating and maintaining an SP-tree. Even when (*,G) and (S,G) overlap, both states are needed to trigger the source-specific Join/Prune messages. (S,G) state is kept alive by data packets arriving from that source. A timer, Entry-timer, is set for the (S,G)
entry and this timer is restarted whenever data packets for (S,G) are forwarded out at least one oif, or Registers are sent. When the Entry-timer expires, the state is deleted. The last-hop router is the router that delivers the packets to their ultimate end-system destination. This is the router that monitors if there is group membership and joins or prunes the appropriate distribution trees in response. In general the last-hop router is the Designated Router (DR) for the LAN. However, under various conditions described later, a parallel router connected to the same LAN may take over as the last-hop router in place of the DR. Only the RP and routers with local members can initiate switching to the SP-tree; intermediate routers do not. Consequently, last-hop routers create (S,G) state in response to data packets from the source, S; whereas intermediate routers only create (S,G) state in response to Join/Prune messages from downstream that have S in the Join list. The (S,G) entry is initialized with the SPT-bit cleared, indicating that the shortest path tree branch from S has not yet been setup completely, and the router can still accept packets from S that arrive on the (*,G) entry's indicated incoming interface (iif). Each PIM multicast entry has an associated incoming interface on which packets are expected to arrive. When a router with a (S,G) entry and a cleared SPT-bit starts to receive packets from the new source S on the iif for the (S,G) entry, and that iif differs from the (*,G) entry's iif, the router sets the SPT-bit, and sends a Join/Prune message towards the RP, indicating that the router no longer wants to receive packets from S via the shared RP-tree. The Join/Prune message sent towards the RP includes S in the prune list, with the RPT-bit set indicating that S's packets must not be forwarded down this branch of the shared tree. If the router receiving the Join/Prune message has (S,G) state (with or without the route entry's RPT-bit flag set), it deletes the arriving interface from the (S,G) oif list. If the router has only (*,G) state, it creates an entry with the RPT-bit flag set to 1. For brevity we refer to an (S,G) entry that has the RPT-bit flag set to 1 as an (S,G)RPT-bit entry. This notational distinction is useful to point out the different actions taken for (S,G) entries depending on the setting of the RPT-bit flag. Note that a router can have no more than one active (S,G) entry for any particular S and G, at any particular time; whether the RPT-bit flag is set or not. In other words, a router never has both an (S,G) and an (S,G)RPT-bit entry for the same S and G at the same time. The Join/Prune message payload contains Multicast-Address=G, Join=NULL, Prune=S,RPT-bit.
A new receiver may join an existing RP-tree on which source-specific prune state has been established (e.g., because downstream receivers have switched to SP-trees). In this case the prune state must be eradicated upstream of the new receiver to bring all sources' data packets down to the new receiver. Therefore, when a (*,G) Join arrives at a router that has any (Si,G)RPT-bit entries (i.e., entries that cause the router to send source-specific prunes toward the RP), these entries must be updated upstream of the router so as to bring all sources' packets down to the new member. To accomplish this, each router that receives a (*,G) Join/Prune message updates all existing (S,G)RPT-bit entries. The router may also trigger a (*,G) Join/Prune message upstream to cause the same updating of RPT-bit settings upstream and pull down all active sources' packets. If the arriving (*,G) join has some sources included in its prune list, then the corresponding (S,G)RPT-bit entries are left unchanged (i.e., the RPT-bit remains set and no oif is added). 2.5 Steady state maintenance of distribution tree (i.e., router state)} In the steady state each router sends periodic Join/Prune messages for each active PIM route entry; the Join/Prune messages are sent to the neighbor indicated in the corresponding entry. These messages are sent periodically to capture state, topology, and membership changes. A Join/Prune message is also sent on an event-triggered basis each time a new route entry is established for some new source (note that some damping function may be applied, e.g., a short delay to allow for merging of new Join information). Join/Prune messages do not elicit any form of explicit acknowledgment; routers recover from lost packets using the periodic refresh mechanism. 2.6 Obtaining RP information To obtain the RP information, all routers within a PIM domain collect Bootstrap messages. Bootstrap messages are sent hop-by-hop within the domain; the domain's bootstrap router (BSR) is responsible for originating the Bootstrap messages. Bootstrap messages are used to carry out a dynamic BSR election when needed and to distribute RP information in steady state. A domain in this context is a contiguous set of routers that all implement PIM and are configured to operate within a common boundary defined by PIM Multicast Border Routers (PMBRs). PMBRs connect each PIM domain to the rest of the internet. Routers use a set of available RPs (called the RP-Set) distributed in Bootstrap messages to get the proper Group to RP mapping. The following paragraphs summarize the mechanism; details of the mechanism may be found in Sections 3.6 and Appendix 6.2. A (small)
set of routers, within a domain, are configured as candidate BSRs and, through a simple election mechanism, a single BSR is selected for that domain. A set of routers within a domain are also configured as candidate RPs (C-RPs); typically these will be the same routers that are configured as C-BSRs. Candidate RPs periodically unicast Candidate-RP-Advertisement messages (C-RP-Advs) to the BSR of that domain. C-RP-Advs include the address of the advertising C-RP, as well as an optional group address and a mask length field, indicating the group prefix(es) for which the candidacy is advertised. The BSR then includes a set of these Candidate-RPs (the RP-Set), along with the corresponding group prefixes, in Bootstrap messages it periodically originates. Bootstrap messages are distributed hop-by- hop throughout the domain. Routers receive and store Bootstrap messages originated by the BSR. When a DR gets a membership indication from IGMP for (or a data packet from) a directly connected host, for a group for which it has no entry, the DR uses a hash function to map the group address to one of the C-RPs whose Group-prefix includes the group (see Section 3.7). The DR then sends a Join/Prune message towards (or unicasts Registers to) that RP. The Bootstrap message indicates liveness of the RPs included therein. If an RP is included in the message, then it is tagged as `up' at the routers; while RPs not included in the message are removed from the list of RPs over which the hash algorithm acts. Each router continues to use the contents of the most recently received Bootstrap message until it receives a new Bootstrap message. If a PIM domain partitions, each area separated from the old BSR will elect its own BSR, which will distribute an RP-Set containing RPs that are reachable within that partition. When the partition heals, another election will occur automatically and only one of the BSRs will continue to send out Bootstrap messages. As is expected at the time of a partition or healing, some disruption in packet delivery may occur. This time will be on the order of the region's round-trip time and the bootstrap router timeout value. 2.7 Interoperation with dense mode protocols such as DVMRP In order to interoperate with networks that run dense-mode, broadcast and prune, protocols, such as DVMRP, all packets generated within a PIM-SM region must be pulled out to that region's PIM Multicast Border Routers (PMBRs) and injected (i.e., broadcast) into the DVMRP network. A PMBR is a router that sits at the boundary of a PIM-SM domain and interoperates with other types of multicast routers such as those that run DVMRP. Generally a PMBR would speak both protocols and implement interoperability functions not required by regular PIM
routers. To support interoperability, a special entry type, referred to as (*,*,RP), must be supported by all PIM routers. For this reason we include details about (*,*,RP) entry handling in this general PIM specification. A data packet will match on a (*,*,RP) entry if there is no more specific entry (such as (S,G) or (*,G)) and the destination group address in the packet maps to the RP listed in the (*,*,RP) entry. In this sense, a (*,*,RP) entry represents an aggregation of all the groups that hash to that RP. PMBRs initialize (*,*,RP) state for each RP in the domain's RPset. The (*,*,RP) state causes the PMBRs to send (*,*,RP) Join/Prune messages toward each of the active RPs in the domain. As a result distribution trees are built that carry all data packets originated within the PIM domain (and sent to the RPs) down to the PMBRs. PMBRs are also responsible for delivering externally-generated packets to routers within the PIM domain. To do so, PMBRs initially encapsulate externally-originated packets (i.e., received on DVMRP interfaces) in Register messages and unicast them to the corresponding RP within the PIM domain. The Register message has a bit indicating that it was originated by a border router and the RP caches the originating PMBR's address in the route entry so that duplicate Registers from other PMBRs can be declined with a Register-Stop message. All PIM routers must be capable of supporting (*,*,RP) state and interpreting associated Join/Prune messages. We describe the handling of (*,*,RP) entries and messages throughout this document; however, detailed PIM Multicast Border Router (PMBR) functions will be specified in a separate interoperability document (see directory, http://catarina.usc.edu/pim/interop/). 2.8 Multicast data packet processing Data packets are processed in a manner similar to other multicast schemes. A router first performs a longest match on the source and group address in the data packet. A (S,G) entry is matched first if one exists; a (*,G) entry is matched otherwise. If neither state exists, then a (*,*,RP) entry match is attempted as follows: the router hashes on G to identify the RP for group G, and looks for a (*,*,RP) entry that has this RP address associated with it. If none of the above exists, then the packet is dropped. If a state is matched, the router compares the interface on which the packet arrived to the incoming interface field in the matched route entry. If the iif check fails the packet is dropped, otherwise the packet is forwarded to all interfaces listed in the outgoing interface list.
Some special actions are needed to deliver packets continuously while switching from the shared to shortest-path tree. In particular, when a (S,G) entry is matched, incoming packets are forwarded as follows: 1 If the SPT-bit is set, then: 1 if the incoming interface is the same as a matching (S,G) iif, the packet is forwarded to the oif-list of (S,G). 2 if the incoming interface is different than a matching (S,G) iif , the packet is discarded. 2 If the SPT-bit is cleared, then: 1 if the incoming interface is the same as a matching (S,G) iif, the packet is forwarded to the oif-list of (S,G). In addition, the SPT bit is set for that entry if the incoming interface differs from the incoming interface of the (*,G) or (*,*,RP) entry. 2 if the incoming interface is different than a matching (S,G) iif, the incoming interface is tested against a matching (*,G) or (*,*,RP) entry. If the iif is the same as one of those, the packet is forwarded to the oif-list of the matching entry. 3 Otherwise the iif does not match any entry for G and the packet is discarded. Data packets never trigger prunes. However, data packets may trigger actions that in turn trigger prunes. For example, when router B in figure 3 decides to switch to SP-tree at step 3, it creates a (S,G) entry with SPT-bit set to 0. When data packets from S arrive at interface 2 of B, B sets the SPT-bit to 1 since the iif for (*,G) is different than that for (S,G). This triggers the sending of prunes towards the RP.
2.9 Operation over Multi-access Networks This section describes a few additional protocol mechanisms needed to operate PIM over multi-access networks: Designated Router election, Assert messages to resolve parallel paths, and the Join/Prune- Suppression-Timer to suppress redundant Joins on multi-access networks. Designated router election: When there are multiple routers connected to a multi-access network, one of them must be chosen to operate as the designated router (DR) at any point in time. The DR is responsible for sending triggered Join/Prune and Register messages toward the RP. A simple designated router (DR) election mechanism is used for both SM and traditional IP multicast routing. Neighboring routers send Hello messages to each other. The sender with the largest network layer address assumes the role of DR. Each router connected to the multi-access LAN sends the Hellos periodically in order to adapt to changes in router status. Parallel paths to a source or the RP--Assert process: If a router receives a multicast datagram on a multi-access LAN from a source whose corresponding (S,G) outgoing interface list includes the interface to that LAN, the packet must be a duplicate. In this case a single forwarder must be elected. Using Assert messages addressed to `18.104.22.168' (ALL-PIM-ROUTERS group) on the LAN, upstream routers can resolve which one will act as the forwarder. Downstream routers listen to the Asserts so they know which one was elected, and therefore where to send subsequent Joins. Typically this is the same as the downstream router's RPF (Reverse Path Forwarding) neighbor; but there are circumstances where this might not be the case, e.g., when using multiple unicast routing protocols on that LAN. The RPF neighbor for a particular source (or RP) is the next-hop router to which packets are forwarded en route to that source (or RP); and therefore is considered a good path via which to accept packets from that source. The upstream router elected is the one that has the shortest distance to the source. Therefore, when a packet is received on an outgoing interface a router sends an Assert message on the multi-access LAN indicating what metric it uses to reach the source of the data packet. The router with the smallest numerical metric (with ties broken by highest address) will become the forwarder. All other
upstream routers will delete the interface from their outgoing interface list. The downstream routers also do the comparison in case the forwarder is different than the RPF neighbor. Associated with the metric is a metric preference value. This is provided to deal with the case where the upstream routers may run different unicast routing protocols. The numerically smaller metric preference is always preferred. The metric preference is treated as the high-order part of an assert metric comparison. Therefore, a metric value can be compared with another metric value provided both metric preferences are the same. A metric preference can be assigned per unicast routing protocol and needs to be consistent for all routers on the multi-access network. Asserts are also needed for (*,G) entries since an RP-Tree and an SP-Tree for the same group may both cross the same multi-access network. When an assert is sent for a (*,G) entry, the first bit in the metric preference (RPT-bit) is always set to 1 to indicate that this path corresponds to the RP tree, and that the match must be done on (*,G) if it exists. Furthermore, the RPT-bit is always cleared for metric preferences that refer to SP-tree entries; this causes an SP- tree path to always look better than an RP-tree path. When the SP- tree and RPtree cross the same LAN, this mechanism eliminates the duplicates that would otherwise be carried over the LAN. In case the packet, or the Assert message, matches on oif for (*,*,RP) entry, a (*,G) entry is created, and asserts take place as if the matching state were (*,G). The DR may lose the (*,G) Assert process to another router on the LAN if there are multiple paths to the RP through the LAN. From then on, the DR is no longer the last-hop router for local receivers and removes the LAN from its (*,G) oif list. The winning router becomes the last-hop router and is responsible for sending (*,G) join messages to the RP. Join/Prune suppression: Join/Prune suppression may be used on multi-access LANs to reduce duplicate control message overhead; it is not required for correct performance of the protocol. If a Join/Prune message arrives and matches on the incoming interface for an existing (S,G), (*,G), or (*,*,RP) route entry, and the Holdtime included in the Join/Prune message is greater than the recipient's own [Join/Prune-Holdtime] (with ties resolved in favor of the higher network layer address), a timer (the Join/Prune-Suppression-timer) in the recipient's route entry may be started to suppress further Join/Prune messages. After this timer expires, the recipient triggers a Join/Prune message, and
resumes sending periodic Join/Prunes, for this entry. The Join/Prune-Suppression-timer should be restarted each time a Join/Prune message is received with a higher Holdtime. 2.10 Unicast Routing Changes When unicast routing changes, an RPF check is done on all active (S,G), (*,G) and (*,*,RP) entries, and all affected expected incoming interfaces are updated. In particular, if the new incoming interface appears in the outgoing interface list, it is deleted from the outgoing interface list. The previous incoming interface may be added to the outgoing interface list by a subsequent Join/Prune from downstream. Join/Prune messages received on the current incoming interface are ignored. Join/Prune messages received on new interfaces or existing outgoing interfaces are not ignored. Other outgoing interfaces are left as is until they are explicitly pruned by downstream routers or are timed out due to lack of appropriate Join/Prune messages. If the router has a (S,G) entry with the SPT-bit set, and the updated iif(S,G) does not differ from iif(*,G) or iif(*,*,RP), then the router resets the SPT-bit. The router must send a Join/Prune message with S in the Join list out any new incoming interfaces to inform upstream routers that it expects multicast datagrams over the interface. It may also send a Join/Prune message with S in the Prune list out the old incoming interface, if the link is operational, to inform upstream routers that this part of the distribution tree is going away. 2.11 PIM-SM for Inter-Domain Multicast Future documents will address the use of PIM-SM as a backbone inter- domain multicast routing protocol. Design choices center primarily around the distribution and usage of RP information for wide area, inter-domain groups. 2.12 Security All PIM control messages may use IPsec  to address security concerns. Security mechanisms are likely to be enhanced in the near future.