]> 5}="yes"?> China Mobile

China liuyisong@chinamobile.com

Futurewei

USA tte@cs.fau.de

Futurewei

USA michael.mcbride@futurewei.com

ZTE Corporation

China zhang.zheng@zte.com.cn

PIM LANs often have more than one upstream router. When PIM Sparse Mode (PIM-SM), including PIM Source Specific-Specific Multicast (PIM-SSM), is used, this can lead to duplicate IP multicast packets being forwarded by these PIM routers. PIM Assert messages are used to elect a single forwarder for each IP multicast traffic flow between these routers. This document defines a mechanism to send and receive information for multiple IP multicast flows in a single PackedAssert message. This optimization reduces the total number of PIM packets on the LAN and can therefore speed up the election of the single forwarder, reducing the number of duplicate IP multicast packets incurred.

Introduction In PIM-SM shared LAN networks, there is typically more than one upstream router. When duplicate data packets appear on the LAN, from different upstream routers, assert packets are sent from these routers to elect a single forwarder according to . The PIM assert messages are sent periodically to keep the assert state. The PIM assert message carries information about a single multicast source and group, along with the corresponding metric-preference and metric of the route towards the source or PIM Rendezvous Point (RP). This document defines a mechanism to encode the information of multiple PIM Assert messages into a single PackedAssert message. This allows to send and receive information for multiple IP multicast flows in a single PackedAssert message without changing the PIM Assert state machinery. It reduces the total number of PIM packets on the LAN and can therefore speed up the election of the single forwarder, reducing the number of duplicate IP multicast packets. This can particularly be helpful when there is traffic for a large number of multicast groups or SSM channels and PIM packet processing performance of the routers is slow.

Requirements Language 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 when, and only when, they appear in all capitals, as shown here.

Terminology The reader is expected to be familiar with the terminology of . The following lists the abbreviations repeated in this document. AT: Assert Timer RP: Rendezvous Point RPF: Reverse Path Forwarding SPT: Shortest Path Tree RPT: RP Tree DR: Designated Router

Problem Statement PIM Asserts occur in many deployments. See for explicit examples and explanations of why it is often not possible to avoid. PIM assert state depends mainly on the network topology. As long as there is a layer 2 network with more than 2 PIM routers, there may be multiple upstream routers, which can cause duplicate multicast traffic to be forwarded and assert process to occur. As the multicast services become widely deployed, the number of multicast entries increases, and a large number of assert messages may be sent in a very short period when multicast data packets trigger PIM assert processing in the shared LAN networks. The PIM routers need to process a large number of PIM assert small packets in a very short time. As a result, the device load is very large. The assert packet may not be processed in time or even discarded, thus extending the time of traffic duplication in the network. The PIM Assert mechanism can only be avoided by designing the network to be without transit subnets with multiple upstream routers. For example, an L2 ring between routers can sometimes be reconfigured to be a ring of point-to-point subnets connected by the routers. These L2/L3 topology changes are undesirable though, when they are only done to enable IP multicast with PIM because they increase the cost of introducing IP multicast with PIM. These designs are also not feasible when specific L2 technologies are needed. For example various L2 technologies for rings provide sub 50msec failover mechanisms, something not possible equally with an L3 subnet based ring. Likewise, IEEE Time Sensitive Networking mechanisms would require an L2 topology that can not simply be replaced by an L3 topology. L2 sub-topologies can also significantly reduce the cost of deployment.

Specification This document defines three elements in support of PIM assert packing: The PIM Assert Packing Hello Option. The encoding of PackedAssert messages. How to send and receive PackedAssert messages.

PIM Assert Packing Hello Option The PIM Assert Packing Hello Option () is used to announce support for the assert packing mechanisms specified in this document. PackedAssert messages () MUST NOT be used unless all PIM routers in the same subnet announce this option.

Assert Packing Message Formats The PIM Assert message, as defined in Section 4.9.6 of , describes the parameters of a (*,G) or (S,G) assert through the following information elements: Rendezvous Point Tree flag (R), Source Address, Group Address, Metric and Metric Preference. This document calls this information an assert record. Assert packing introduces two new PIM Assert message encodings through the allocation and use of two flags in the PIM Assert message header , the Packed (P) and the Aggregated (A) flags. If the (P)acked flag is 0, the message is a (non-packed) PIM Assert message as specified in . See . In this case, the (A) flag MUST be set to 0, and MUST be ignored on receipt. If the (P) flag is 1, then the message is called a PackedAssert message and the type and hence encoding format of the payload is determined by the (A) flag. If A=0, then the message body is a sequence of assert records. This is called a "Simple PackedAssert" message. See . If A=1, then the message body is a sequence of aggregated assert records. This is called an "Aggregated PackedAssert". See . Two aggregated assert record types are specified. The "Source Aggregated Assert Record", see , encodes one (common) Source Address, Metric and Metric Preference as well as a list of one or more Group Addresses. Source Aggregated Assert Records provide a more compact encoding than the Simple PackedAssert message format when multiple (S,G) flows share the same source S. A single Source Aggregated Assert Record with n Group Addresses represents the information of assert records for (S,G1)...(S,Gn). The "RP Aggregated Assert Record", see , encodes one common Metric and Metric Preference as well as a list of "Group Records", each of which encodes a Group Address and a list of zero or more Source Addresses with a count. This is called an "RP Aggregated Assert Record", because with standard RPF according to (), all the Group Addresses that use the same RP will have the same Metric and Metric Preference. RP Aggregation Records provide a more compact encoding than the Simple PackedAssert message format for (*,G) flows. The Source Address is optionally used by assert procedures to indicate the source(s) that triggered the assert, otherwise the Source Address is set to 0 in the assert record. Both Source Aggregated Assert Records and RP Aggregated Assert Records also include the R flag which maintains its semantic from but also distinguishes the encodings. Source Aggregated Assert Records have R=0, as (S,G) assert records do in . RP Aggregated Assert Records have R=1, as (*,G) assert records do in .

PackedAssert Mechanism PackedAsserts do not change the PIM assert state machine specification. Instead, sending and receiving of PackedAssert messages as specified in the following subsections are logically new packetization options for assert records in addition to the (not packed) Assert Message. There is no change to the assert record information elements transmitted or their semantic. They are just transmitted in fewer but larger packets and fewer total number of bytes used to encode the information elements. In result, PIM routers should be able to send/receive assert records faster and/or with less processing overhead.

Sending PackedAssert messages When using assert packing, the regular Assert message encoding with A=0 and P=0 is still allowed to be sent. Routers are free to choose which PackedAssert message type they send. It is out of scope of this specification for which conditions, such as data-triggered asserts or Assert Timer (AT) expiry based asserts, an implementation should generate PackedAsserts instead of Asserts. Instead, Implementations are expected to specify in documentation and/or management interfaces (such as a YANG model), which PackedAssert message types they can send and under which conditions they will. Implementations SHOULD NOT send only Asserts, but no PackedAsserts under all conditions, when all routers on the LAN do support Assert Packing. When any PIM routers on the LAN have not signaled support for Assert Packing, implementations MUST send only Asserts and MUST NOT send PackedAsserts under any condition. When a PIM router has an assert record ready to send according to , it calls send Assert(S,G) / send Assert(,G) (Section 4.2), Send Assert(S,G) / SendAssertCancel(S,G) (Section 4.6.1), Send Assert(,G) / Send AssertCancel(*,G) (Section 4.6.2) and send Assert(S,G) (Section 4.8.2). Each of these calls will send an Assert message. When sending of PackedAsserts is possible on the network, any of these calls is permitted to not send an Assert message, but only remember the assert record, and let PIM continue with further processing for other flows that may result in additional assert records - to finally packed PackedAssert messages from the remembered assert records and send them. The following text discusses several conditions to be taken into account for this further processing and how to create and schedule PackedAssert messages. Avoiding possible additional and relevant delay because of further processing is most critical for assert records that are triggered by reception of data or reception of asserts against which the router is in the "I am Assert Winner" state.  In these cases the router SHOULD send out an Assert or PackedAssert message containing this assert record as soon as possible to minimize the time in which duplicate IP multicast packets can occur. To avoid additional delay in this case, the router should employ appropriate assert packing and scheduling mechanisms, such as for example the following. Asserts/PackedAsserts in this case are scheduled for serialization with highest priority, such that they bypass any potentially earlier scheduled other packets. When there is no such Assert/PacketAssert message scheduled for or being serialized, the router immediately serializes an Assert or PackedAssert message without further assert packing. If there are one or more Assert/PackedAssert messages serialized and/or scheduled to be serialized, then the router can pack assert records into new PackedAssert messages until shortly before the last of those Assert/PackedAssert packets has finished serializing. Asserts triggered by expiry of the AT on an assert winner are not time-critical because they can be scheduled in advance and because the Assert_Override_Interval parameter of already creates a 3 second window in which such assert records can be sent, received, and processed before an assert losers state would expire and duplicate IP multicast packets could occur. An example mechanism to allow packing of AT expiry triggered assert records on assert winners is to round the AT to an appropriate granularity such as 100msec. This will cause AT for multiple (S,G) and/or (*,G) states to expire at the same time, thus allowing them to be easily packed without changes to the assert state machinery. AssertCancel messages have assert records with an infinite metric and can use assert packing as any other Assert. They are sent on Override Timer (OT) expiry and can be packed for example with the same considerations as AT expiry triggered assert records. Additional delay is not "relevant" when it still causes the overall amount of (possible) duplicate IP multicast packets to decrease in a condition with large number of (S,G) and/or (*,G), compared to the situation in which no delay is added by the implementation. This can simply be the case because the implementation can not afford to implement the (more advanced) mechanisms described above, and some simpler mechanism that does introduce some additional delay still causes more overall reduction in duplicate IP multicast packets than not sending PackedAsserts at all, but only Asserts. "Relevant" is a highly implementation dependent metric and can typically only be measured against routers of the same type as receivers, and performance results with other routers will likely differ. Therefore it is optional. When Asserts are sent, a single packet loss will result only in continued or new duplicates from a single IP multicast flow. Loss of (non AssertCancel) PackedAssert impacts duplicates for all flows packed into the PackedAssert and may result in the need for re-sending more than one Assert/PacketAssert, because of the possible inability to pack the assert records in this condition. Therefore, routers SHOULD support mechanisms allowing for PackedAsserts and Asserts to be sent with an appropriate DSCP, such as Expedited Forwarding (EF), to minimize their loss, especially when duplicate IP multicast packets could cause congestion and loss. Routers MAY support a configurable option for sending PackedAssert messages twice in short order (such as 50msec apart), to overcome possible loss, but only when the following two conditions are met. The total size of the two PackedAsserts is less than the total size of equivalent Assert messages, The condition of the assert records flows in the PackedAssert is such that the router can expect that their reception by PIM routers will not trigger Assert/PackedAsserts replies. This condition is true for example when sending an assert record while becoming or being Assert Winner (Action A1/A3 in ). It is sufficient that assert records are not packed up to MTU size, but to a size that allows the router to achieve the required operating scale of (S,G) and (*,G) flows with minimum duplicates. This packing size may be larger when the network is operating with the maximum number of supported multicast flows, and it can be a smaller packing size when operating with fewer multicast flows. Larger than sufficient packets may then not provide additional benefits, because they only reduce the performance requirements for packet sending and reception, and other performance limiting factors may take over once a sufficient size is reached. And larger packets can incur more duplicates on loss. Routers may support a sufficient amount of packing to minimize the negative impacts of loss of PackedAssert packets without loss of performance of minimizing IP multicast packet duplication.

Receiving PackedAssert messages Upon reception of a PackedAssert message, the PIM router logically converts its payload into a sequence of assert records that are then processed as if an equivalent sequence of Assert messages where received according to .

Packet Formats This section describes the format of new PIM extensions introduced by this document.

PIM Assert Packing Hello Option

The PIM Assert Packing Hello Option is a new option for PIM Hello Messages according to Section 4.9.2 of . OptionType TBD: PIM Packed Assert Capability Hello Option Including the PIM OptionType TBD indicates support for the ability to receive and process all the PackedAssert encodings defined in this document.

Assert Message Format

shows a PIM Assert message as specified in Section 4.9.6 of with the common header showing the "7 6 5 4 3 2" Flag Bits as defined in Section 4 of and the location of the P and A flags, see .  As specified in below, both flags in a (non-packed) PIM Assert message are required to be set to 0.

Simple PackedAssert Message Format

PIM Version, Type, Checksum: As specified in Section 4.9.6 of . "7 6 5 4 3 2": IANA registry handled bits according to Section 4 of . Zero: Set to zero on transmission. Serves to make non assert packing capable PIM routers fail in parsing the message instead of possible mis-parsing if this field was used. Reserved: Set to zero on transmission. Ignored upon receipt. P: packed flag. MUST be 1. A: aggregated flag. MUST be 0. M: The number of Assert Records in the message. Derived from the length of the packet carrying the message. The format of each Assert Record is the same as the PIM assert message body as specified in Section 4.9.6 of .

Aggregated PackedAssert Message Format

PIM Version, Type, Reserved, Checksum: As specified in Section 4.9.6 of . "7 6 5 4 3 2": IANA registry handled bits according to Section 4 of . P: packed flag. MUST be 1. A: aggregated flag. MUST be 1. Zero: Set to zero on transmission. Serves to make non assert packing capable PIM routers fail in parsing the message instead of possible mis-parsing if this field was used. M: The number of Aggregated Assert Records in the message. Derived from the length of the packet carrying the message.

Source Aggregated Assert Record

Reserved: Set to zero on transmission. Ignored upon receipt. R: MUST be 0. R indicates both that the encoding format of the record is that of a Source Aggregated Assert Record, but also that all assert records represented by the Source Aggregated Assert Record have R=0 and are therefore (S,G) assert records according to the definition of R in , Section 4.9.6. Source Address, Metric Preference, Metric: As specified in Section 4.9.6 of . Source Address MUST NOT be zero. Number of Groups: The number of Group Address fields. Group Address: As specified in Section 4.9.6 of .

RP Aggregated Assert Record An RP Aggregation Assert record aggregates (*,G) assert records with the same Metric Preference and Metric. Typically this is the case for all (*,G) using the same RP, but the encoding is not limited to only (*,G) using the same RP because the RP address is not encoded as it is also not present in assert records.

R: MUST be 1. R indicates both that the encoding format of the record is that of an RP Aggregated Assert Record, and that all assert records represented by the RP Aggregated Assert Record have R=1 and are therefore (*,G) assert records according to the definition of R in , Section 4.9.6. Metric Preference, Metric: As specified in Section 4.9.6 of . Reserved: Set to zero on transmission. Ignored upon receipt. Number of Group Records: The number of packed Group Records. A record consists of a Group Address and a Source Address list with a number of sources. The format of each Group Record is:

Group Address and Reserved: As specified in Section 4.9.6 of . Reserved: Set to zero on transmission. Ignored upon receipt. Number of Sources: The Number of Sources is corresponding to the number of Source Address fields in the Group Record. If this number is 0, the Group Record indicates one assert record with a Source Address of 0. If this number is not 0 and one of the (*,G) assert records to be encoded should have the Source Address 0, then 0 needs to be encoded as one of the Source Address fields. Source Address: As specified in Section 4.9.6 of . But there can be multiple Source Address fields in the Group Record.

IANA Considerations IANA is requested to assign a new code point from the "PIM-Hello Options" registry for the Packed Assert Capability as follows:

IANA is requested to assign two Flag Bits in the Assert message from the "PIM Message Types" registry as follows:

[RFC-Editor note: If IANA can not assign the requested two bits 0 and 1, then the figures showing those two bits need to be fixed to show P and A in the actual locations IANA assigns - aka: the bits shown are "placeholders" according to the requested bits in this section.]

Security Considerations The security considerations of apply to the extensions defined in this document. This document packs multiple assert records in a single message. As described in Section 6.1 of , a forged Assert message could cause the legitimate designated forwarder to stop forwarding traffic to the LAN. The effect may be amplified when using a PackedAssert message.

Acknowledgments The authors would like to thank: Stig Venaas for the valuable contributions of this document, Alvaro Retana for his thorough and constructive RTG AD review, Ines Robles for her Gen-ART review, Tommy Pauly for his transport area review, Robert Sparks for his SecDir review and Shuping Peng for her RtgDir review.

Working Group considerations [RFC-Editor: please remove this section].

Open Issues

Changelog This document is hosted starting with -06 on https://github.com/toerless/pim-assert-packing.

draft-ietf-pim-assert-packing-09 For details of review discussion/replies, see review reply emails in (https://github.com/toerless/pim-assert-packing/tree/main/emails)j review Alvaro Retana: Reintroduced ref to PIM-DM, fixed typos, downgraded MAY->may for "sufficient". IANA Expert Review / Stig Venaas: Removed Count field from message headers as it complicates parsing and is unnecessary. Added "Zero" field to make packed asserts received by a non-packed-assert-capable-router guaranteed to fail ("Reserved" address family type). Changed from RFC8736 to RFC8736bis so that we can use the word "Unassigned" in the IANA table. Review Shuping Peng Changed explanation of how assert packing works from "layer" to "alternative to packetization via PIM Assert Message. Fixed various typos, expanded term etc.. Review Robert Sparks: Moved Intro explanations of how one could avoid asserts (but how its problematic) to appendix. Applied textual nits found. Removed quotes around term "sufficient" for easier readbility. Review Tommy Paul: Transport related concern explained in reply, but no additional explanations in text because the question referred to basic PIM behavior expected to be understood by readers: No discovery of loss/trigger for retransmission, just restransmission of same message element after discover of ongoing duplicates and/or expiry of timers.

draft-ietf-pim-assert-packing-08 Included changes from review of Alvaro Retana (https://mailarchive.ietf.org/arch/msg/pim/GsKq9bB2a6yDdM9DvAUGCWthdEI) Please see the following emails discussing the changes: https://raw.githubusercontent.com/toerless/pim-assert-packing/main/emails/07-alvaro-review-reply.txt

draft-ietf-pim-assert-packing-07 Included changes from review of Alvaro Retana (https://mailarchive.ietf.org/arch/msg/pim/Cp4o5glUFge2b84X9CQMwCWZjAk/) Please see the following emails discussing the changes: https://raw.githubusercontent.com/toerless/pim-assert-packing/main/emails/05-alvaro-review-reply.txt https://raw.githubusercontent.com/toerless/pim-assert-packing/main/emails/07-pim-wg-announce.txt

draft-ietf-pim-assert-packing-06 This version was converted from txt format into markdown for better editing later, but is otherwise text identical to -05. It was posted to DataTracker to make diffs easier. Functional changes:

In many standards track documents several words are used to signify the requirements in the specification. These words are often capitalized. This document defines these words as they should be interpreted in IETF documents. This document specifies an Internet Best Current Practices for the Internet Community, and requests discussion and suggestions for improvements. This document specifies Protocol Independent Multicast - Sparse Mode (PIM-SM). PIM-SM is a multicast routing protocol that can use the underlying unicast routing information base or a separate multicast-capable routing information base. It builds unidirectional shared trees rooted at a Rendezvous Point (RP) per group, and it optionally creates shortest-path trees per source.This document obsoletes RFC 4601 by replacing it, addresses the errata filed against it, removes the optional (*,*,RP), PIM Multicast Border Router features and authentication using IPsec that lack sufficient deployment experience (see Appendix A), and moves the PIM specification to Internet Standard. RFC 2119 specifies common key words that may be used in protocol specifications. This document aims to reduce the ambiguity by clarifying that only UPPERCASE usage of the key words have the defined special meanings. Cisco Systems, Inc. Futurewei Technologies, Inc. The PIM version 2 messages share a common message header format. The common header definition contains eight reserved bits. This document specifies how these bits may be used by individual message types and creates a registry containing the per-message-type usage. This document also extends the PIM type space by defining three new message types. For each of the new types, four of the previously reserved bits are used to form an extended type range. This document updates RFCs 7761 and 3973 by defining the use of the currently Reserved field in the PIM common header. This document further updates RFCs 7761 and 3973, along with RFCs 5015, 5059, 6754, and 8364, by specifying the use of the currently reserved bits for each PIM message. This document obsoletes RFC 8736. This document specifies Protocol Independent Multicast - Dense Mode (PIM-DM). PIM-DM is a multicast routing protocol that uses the underlying unicast routing information base to flood multicast datagrams to all multicast routers. Prune messages are used to prevent future messages from propagating to routers without group membership information. This memo defines an Experimental Protocol for the Internet community. This document describes the MVPN (Multicast in BGP/MPLS IP VPNs) solution designed and deployed by Cisco Systems. The procedures specified in this document are largely a subset of the generalized MVPN framework recently standardized by the IETF. However, as the deployment of the procedures specified herein predates the publication of IETF standards (in some cases by over five years), an implementation based on these procedures differs in some respects from a fully standards-compliant implementation. These differences are pointed out in the document. This document defines a Historic Document for the Internet community.

Use case examples The PIM Assert mechanism can only be avoided by designing the network to be without transit subnets with multiple upstream routers. For example, an L2 ring between routers can sometimes be reconfigured to be a ring of point-to-point subnets connected by the routers. These L2/L3 topology changes are undesirable though, when they are only done to enable IP multicast with PIM because they increase the cost of introducing IP multicast with PIM. These designs are also not feasible when specific L2 technologies are needed. For example various L2 technologies for rings provide sub 50msec failover mechanisms, something not possible equally with an L3 subnet based ring. Likewise, IEEE Time Sensitive Networking mechanisms would require an L2 topology that can not simply be replaced by an L3 topology. L2 sub-topologies can also significantly reduce the cost of deployment. The following subsections give examples of the type of network and use-cases in which subnets with asserts have been observerd or are expected to require scaling as provided by this specification.

Enterprise network When an Enterprise network is connected through a layer-2 network, the intra-enterprise runs layer-3 PIM multicast. The different sites of the enterprise are equivalent to the PIM connection through the shared LAN network. Depending upon the locations and amount of groups there could be many asserts on the first-hop routers.

Video surveillance Video surveillance deployments have migrated from analog based systems to IP-based systems oftentimes using multicast. In the shared LAN network deployments, when there are many cameras streaming to many groups there may be issues with many asserts on first-hop routers.

Financial Services Financial services extensively rely on IP Multicast to deliver stock market data and its derivatives, and current multicast solution PIM is usually deployed. As the number of multicast flows grow, there are many stock data with many groups may result in many PIM asserts on a shared LAN network from publisher to the subscribers.

IPTV broadcast Video PIM DR deployments are often used in host-side network for IPTV broadcast video services. Host-side access network failure scenario may be benefitted by assert packing when many groups are being used. According to the DR will be elected to forward multicast traffic in the shared access network. When the DR recovers from a failure, the original DR starts to send traffic, and the current DR is still forwarding traffic. In the situation multicast traffic duplication maybe happen in the shared access network and can trigger the assert progress.

MVPN MDT As described in , MDT (Multicast Distribution Tree) is used as tunnels for MVPN. The configuration of multicast-enabled VRF (VPN routing and forwarding) or interface that is in a VRF changing may cause many assert packets to be sent in a same time.

Special L2 services Additionally, future backhaul, or fronthaul, networks may want to connect L3 across an L2 underlay supporting Time Sensitive Networks (TSN). The infrastructure may run DetNet over TSN. These transit L2 LANs would have multiple upstreams and downstreams. This document is taking a proactive approach to prevention of possible future assert issues in these types of environments.