INTERNET DRAFT Y. Serbest Internet Engineering Task Force SBC Document: Ray Qiu draft-serbest-l2vpn-vpls-mcast-03.txt Venu Hemige July 2005 Alcatel Category: Informational Rob Nath Expires: January 2006 Riverstone Supporting IP Multicast over VPLS Status of this memo By submitting this Internet-Draft, we represent that any applicable patent or other IPR claims of which we are aware have been disclosed, or will be disclosed, and any of which we become aware will be disclosed in accordance with RFC 3668. This document is an Internet-Draft and is in full conformance with Sections 5 and 6 of RFC 3667 and Section 5 of RFC 3668. Internet-Drafts are working documents of the Internet Engineering Task Force (IETF), its areas, and its working groups. Note that other groups may also distribute working documents as Internet- Drafts. 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." The list of current Internet-Drafts can be accessed at http://www.ietf.org/ietf/1id-abstracts.txt. The list of Internet-Draft Shadow Directories can be accessed at http://www.ietf.org/shadow.html. IPR Disclosure Acknowledgement By submitting this Internet-Draft, each author represents that any applicable patent or other IPR claims of which he or she is aware have been or will be disclosed, and any of which he or she becomes aware will be disclosed, in accordance with Section 6 of BCP 79. Abstract In Virtual Private LAN Service (VPLS), the PE devices provide a logical interconnect such that CE devices belonging to a specific VPLS instance appear to be connected by a single LAN. A VPLS solution performs replication for multicast traffic at the ingress PE devices. When replicated at the ingress PE, multicast traffic wastes bandwidth when 1. Multicast traffic is sent to sites with no members, and 2. Pseudo wires to different sites go through a shared path. This document is addressing the former by IGMP and PIM snooping. [Page 1] Internet Draft draft-serbest-l2vpn-vpls-mcast-03.txt July, 2005 Conventions used in this document The key words MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in RFC 2119. Table of Contents 1. Contributing Authors............................................3 2. Introduction....................................................3 3. Overview of VPLS................................................4 4. Multicast Traffic over VPLS.....................................5 5. Constraining of IP Multicast in a VPLS..........................6 5.1. IPv6 Considerations...........................................7 5.2. General Rules for IGMP/PIM Snooping in VPLS...................7 5.3. IGMP Snooping for VPLS........................................8 5.3.1. Discovering Multicast Routers...............................9 5.3.2. IGMP Snooping Protocol State................................9 5.3.3. IGMP Join..................................................10 5.3.4. IGMP Leave.................................................14 5.3.5. Failure Scenarios..........................................15 5.3.6. Scaling Considerations for IGMP Snooping...................16 5.3.7. Downstream Proxy Behavior..................................16 5.3.8. Upstream Proxy Behavior....................................17 5.4. PIM Snooping for VPLS........................................17 5.4.1. PIM Snooping State Summarization Macros....................18 5.4.2. PIM-DM.....................................................20 5.4.2.1. Discovering Multicast Routers............................20 5.4.2.2. PIM-DM Multicast Forwarding..............................21 5.4.2.3. PIM-DM Pruning...........................................21 5.4.2.4. PIM-DM Grafting..........................................22 5.4.2.5. Failure Scenarios........................................23 5.4.3. PIM-SM.....................................................23 5.4.3.1. Discovering Multicast Routers............................24 5.4.3.2. PIM-SM (*,G) Join........................................24 5.4.3.3. PIM-SM Pruning...........................................26 5.4.3.4. PIM-SM (S,G) Join........................................27 5.4.3.5. PIM-SM (S,G,rpt) Prunes..................................28 5.4.3.6. PIM-SM (*,*,RP) State....................................28 5.4.3.7. Failure Scenarios........................................28 5.4.3.8. Special Cases for PIM-SM Snooping........................28 5.4.4. PIM-SSM....................................................30 5.4.4.1. Discovering Multicast Routers............................31 5.4.4.2. Guidelines for PIM-SSM Snooping..........................31 5.4.4.3. PIM-SSM Join.............................................32 5.4.4.4. PIM-SSM Prune............................................33 [Page 2] Internet Draft draft-serbest-l2vpn-vpls-mcast-03.txt July, 2005 5.4.4.5. Failure Scenarios........................................33 5.4.4.6. Special Cases for PIM-SSM Snooping.......................33 5.4.5. Bidirectional-PIM (BIDIR-PIM)..............................33 5.4.5.1. Discovering Multicast Routers............................34 5.4.5.2. Guidelines for BIDIR-PIM Snooping........................35 5.4.5.3. BIDIR-PIM Join...........................................35 5.4.5.4. BIDIR-PIM Prune..........................................36 5.4.5.5. Failure Scenarios........................................37 5.4.6. Multicast Source Directly Connected to the VPLS Instance...37 5.5. VPLS Multicast on the Upstream PE............................37 5.5.1. Negotiating PIM Multicast capability in LDP................38 5.5.2. Exchanging PIM Hellos......................................38 5.5.3. Exchanging PIM Join/Prune States...........................39 5.5.3.1. PIM Join Suppression Issues..............................39 5.5.3.2. Resiliency against soft-state failures...................40 5.5.3.2.1. Explicit Tracking of C-Joins at the downstream PE......40 5.5.3.2.2. Refreshing PIM Join TLVs on the PWs....................41 5.5.3.3. PIM-BIDIR Considerations.................................41 5.6. Data Forwarding Rules........................................41 6. Security Considerations........................................41 7. References.....................................................42 7.1. Normative References.........................................42 7.2. Informative References.......................................42 1. Contributing Authors This document was the combined effort of several individuals. The following are the authors, in alphabetical order, who contributed to this document: Suresh Boddapati Venu Hemige Sunil Khandekar Vach Kompella Marc Lasserre Rob Nath Ray Qiu Yetik Serbest Himanshu Shah 2. Introduction In Virtual Private LAN Service (VPLS), the Provider Edge (PE) devices provide a logical interconnect such that Customer Edge (CE) devices belonging to a specific VPLS instance appear to be connected by a single LAN. Forwarding information base for particular VPLS instance is populated dynamically by source MAC address learning. This is a straightforward solution to support unicast traffic, with reasonable [Page 3] Internet Draft draft-serbest-l2vpn-vpls-mcast-03.txt July, 2005 flooding for unicast unknown traffic. Since a VPLS provides LAN emulation for IEEE bridges as wells as for routers, the unicast and multicast traffic need to follow the same path for layer-2 protocols to work properly. As such, multicast traffic is treated as broadcast traffic and is flooded to every site in the VPLS instance. VPLS solutions (i.e., [VPLS-LDP] and [VPLS-BGP]) perform replication for multicast traffic at the ingress PE devices. When replicated at the ingress PE, multicast traffic wastes bandwidth when: 1. Multicast traffic is sent to sites with no members, 2. Pseudo wires to different sites go through a shared path, and 3. Multicast traffic is forwarded along a shortest path tree as opposed to the minimum cost spanning tree. This document is addressing the first problem by IGMP and PIM snooping. Using VPLS in conjunction with IGMP and/or PIM snooping has the following advantages: - It improves VPLS to support IP multicast efficiently (not necessarily optimum, as there can still be bandwidth waste), - It prevents sending multicast traffic to sites with no members, - It keeps P routers in the core stateless, - The Service Provider (SP) does not need to perform the tasks to provide multicast service (e.g., running PIM, managing P-group addresses, managing multicast tunnels) - The SP does not need to maintain PIM adjacencies with the customers. In this document, we describe the procedures for Internet Group Management Protocol (IGMP) and Protocol Independent Multicast (PIM) snooping over VPLS for efficient distribution of IP multicast traffic. 3. Overview of VPLS In case of VPLS, the PE devices provide a logical interconnect such that CE devices belonging to a specific VPLS appear to be connected by a single LAN. End-to-end VPLS consists of a bridge module and a LAN emulation module ([L2VPN-FR]). In a VPLS, a customer site receives Layer-2 service from the SP. The PE is attached via an access connection to one or more CEs. The PE performs forwarding of user data packets based on information in the Layer-2 header, that is, MAC destination address. The CE sees a bridge. The details of VPLS reference model, which we summarize here, can be found in [L2VPN_FR]. In VPLS, the PE can be viewed as containing a Virtual Switching Instance (VSI) for each L2VPN that it serves. A CE device attaches, possibly through an access network, to a bridge module of a PE. Within the PE, the bridge module attaches, through an Emulated LAN Interface to an Emulated LAN. For each VPLS, there [Page 4] Internet Draft draft-serbest-l2vpn-vpls-mcast-03.txt July, 2005 is an Emulated LAN instance. The Emulated LAN consists of VPLS Forwarder module (one per PE per VPLS service instance) connected by pseudo wires (PW), where the PWs may be traveling through Packet Switched Network (PSN) tunnels over a routed backbone. VSI is a logical entity that contains a VPLS forwarder module and part of the bridge module relevant to the VPLS service instance [L2VPN-FR]. Hence, the VSI terminates PWs for interconnection with other VSIs and also terminates attachment circuits (ACs) for accommodating CEs. A VSI includes the forwarding information base for a L2VPN [L2VPN-FR] which is the set of information regarding how to forward Layer-2 frames received over the AC from the CE to VSIs in other PEs supporting the same L2VPN service (and/or to other ACs), and contains information regarding how to forward Layer-2 frames received from PWs to ACs. Forwarding information bases can be populated dynamically (such as by source MAC address learning) or statically (e.g., by configuration). Each PE device is responsible for proper forwarding of the customer traffic to the appropriate destination(s) based on the forwarding information base of the corresponding VSI. 4. Multicast Traffic over VPLS In VPLS, if a PE receives a frame from an Attachment Circuit (AC) with no matching entry in the forwarding information base for that particular VPLS instance, it floods the frame to all other PEs (which are part of this VPLS instance) and to directly connected ACs (other than the one that the frame is received from). The flooding of a frame occurs when: - The destination MAC address has not been learned, - The destination MAC address is a broadcast address, - The destination MAC address is a multicast address. Malicious attacks (e.g., receiving unknown frames constantly) aside, the first situation is handled by VPLS solutions as long as destination MAC address can be learned. After that point on, the frames will not be flooded. A PE is REQUIRED to have safeguards, such as unknown unicast limiting and MAC table limiting, against malicious unknown unicast attacks. There is no way around flooding broadcast frames. To prevent runaway broadcast traffic from adversely affecting the VPLS service and the SP network, a PE is REQUIRED to have tools to rate limit the broadcast traffic as well. Similar to broadcast frames, multicast frames are flooded as well, as a PE can not know where multicast members reside. Rate limiting multicast traffic, while possible, should be should be done carefully since several network control protocols relies on multicast. For one thing, layer-2 and layer-3 protocols utilize multicast for their operation. For instance, Bridge Protocol Data Units (BPDUs) use an IEEE assigned all bridges multicast MAC address, and OSPF is multicast to all OSPF routers multicast MAC address. If the rate- [Page 5] Internet Draft draft-serbest-l2vpn-vpls-mcast-03.txt July, 2005 limiting of multicast traffic is not done properly, the customer network will experience instability and poor performance. For the other, it is not straightforward to determine the right rate limiting parameters for multicast. A VPLS solution MUST NOT affect the operation of customer layer-2 protocols (e.g., BPDUs). Additionally, a VPLS solution MUST NOT affect the operation of layer-3 protocols. In the following section, we describe procedures to constrain the flooding of IP multicast traffic in a VPLS. 5. Constraining of IP Multicast in a VPLS The objective of improving the efficiency of VPLS for multicast traffic that we are trying to optimize here has the following constraints: - The service is VPLS, i.e., a layer-2 VPN, - In VPLS, ingress replication is required, - There is no layer-3 adjacency (e.g., PIM) between a CE and a PE. Under these circumstances, the most obvious approach is implementation of IGMP and PIM snooping in VPLS. Other multicast routing protocols such as DVMRP and MOSPF are outside the scope of this document. Another approach to constrain multicast traffic in a VPLS is to utilize point-multipoint LSPs (e.g., [PMP-RSVP-TE]). In such case, one has to establish a point-multipoint LSP from a source PE (i.e., the PE to which the source router is connected to) to all other PEs participating in the VPLS instance. In this case, if nothing is done, all PEs will receive multicast traffic even if they do not have any members hanging off of them. One can apply IGMP/PIM snooping, but this time IGMP/PIM snooping should be done in P routers as well. One can propose a dynamic way of establishing point-multipoint LSPs, for instance by mapping IGMP/PIM messages to RSVP-TE signaling. One should consider the effect of such approach on the signaling load and on the delay between the time the join request received and the traffic is received (this is important for IPTV application for instance). This approach is outside the scope of this document. Although, in some extremely controlled cases, such as a ring topology of PE routers with no P routers or a tree topology, the efficiency of the replication of IP multicast can be improved. For instance, spoke PWs of a hierarchical VPLS can be daisy-chained together and some replication rules can be devised. These cases are not expected to be common and will not be considered in this document. In the following sub-sections, we provide some guidelines for the implementation of IGMP and PIM snooping in VPLS. Snooping techniques [Page 6] Internet Draft draft-serbest-l2vpn-vpls-mcast-03.txt July, 2005 need to be employed on ACs at the downstream PEs. Snooping techniques can also be employed on PWs at the upstream PEs. This may work well for small to medium scale deployments. However, if there are a large number of VPLS instances with a large number of PEs per instances, then the amount of snooping required at the upstream PEs can overwhelm the upstream PEs. In section 5.5. , we provide an alternative approach using LDP to build multicast replication states on the upstream PEs. Using a reliable mechanism like LDP allows the upstream PEs to eliminate the requirement to snoop on PWs. It also eliminates the need to refresh multicast states on the upstream PEs. 5.1. IPv6 Considerations In VPLS, PEs forward Ethernet frames received from CEs and as such are agnostic of the layer-3 protocol used by the CEs. However, as an IGMP and PIM snooping switch, the PE would have to look deeper into the IP and IGMP/PIM packets and build snooping state based on that. As already stated, the scope of this document is limited to snooping IGMP/PIM packets. So, we are concerned with snooping specific IP payloads. Nonetheless, there are two IP versions a PE would have to be able to interpret. IGMP is the Group Management Protocol which applies only to IPv4. MLD [MLD] is the equivalent of IGMPv2 defined for IPv6. MLDv2 [MLDv2] is the equivalent of IGMPv3 defined for IPv6. PIM runs on top of both IPv4 and IPv6. This document mandates that a PE MUST be able to snoop IGMP and PIM encapsulated as IPv4 payloads. The PE SHOULD also be capable of snooping MLD/MLDv2 packets and PIM packets encapsulated as IPv6 payloads. If the PE cannot snoop IPv6 payloads, then it MUST NOT build any snooping state for such multicast groups and MUST simply flood any data traffic sent to such groups. This allows an IPv6- unaware PE to perform the snooping function only on IPv4 multicast groups. This is possible because an IPv4 multicast address and an IPv6 multicast address never share the same MAC address. To avoid confusion, this document describes the procedures for IGMP/PIM snooping for IPv4. The procedures described for IGMP can also be applied to MLD and MLDv2. Please refer to Section 3 of [MAGMA-SNOOP] for a list of IPv4/IPv6 differences an IGMP/MLD snooping switch has to be aware of. In addition to those differences, some of the other differences of interest are: - IPv4 multicast addresses map to multicast MAC address starting with 01:00:5E and IPv6 multicast addresses map to multicast MAC addresses starting with 33:33. So the MAC ddresses used for IPv4 and IPv6 never overlap. 5.2. General Rules for IGMP/PIM Snooping in VPLS The following rules for the correct operation of IGMP/PIM snooping MUST be followed. [Page 7] Internet Draft draft-serbest-l2vpn-vpls-mcast-03.txt July, 2005 Rule 1: IGMP and PIM messages forwarded by PEs MUST follow the split- horizon rule for mesh PWs as defined in [VPLS-LDP]. Rule 2: IGMP/PIM snooping states in a PE MUST be per VPLS instance. Rule 3: If a PE does not have any entry in a IGMP/PIM snooping state for multicast group (*,G) or (S,G), the multicast traffic to that group in the VPLS instance MUST be flooded. Rule 4: A PE MUST support PIM mode selection per VPLS instance via CLI and/or EMS. Another option could be to deduce the PIM mode from RP address for a specific multicast group. For instance, a RP address can be learned during the Designated Forwarder (DF) election stage for Bidirectional-PIM. 5.3. IGMP Snooping for VPLS IGMP is a mechanism to inform the routers on a subnet of a hosts request to become a member of a particular multicast group. IGMP is a stateful protocol. The router (i.e., the querier) regularly verifies that the hosts want to continue to participate in the multicast groups by sending periodic queries, transmitted to all hosts multicast group (IP:224.0.0.1, MAC:01-00-5E-00-00-01) on the subnet. If the hosts are still interested in that particular multicast group, they respond with membership report message, transmitted to the multicast group of which they are members. In IGMPv1 [RFC1112], the hosts simply stop responding to IGMP queries with membership reports, when they want to leave a multicast group. IGMPv2 [RFC2236] adds a leave message that a host will use when it needs to leave a particular multicast group. IGMPv3 [RFC3376] extends the report/leave mechanism beyond multicast group to permit joins and leaves to be issued for specific source/group (S,G) pairs. In IGMP snooping, a PE snoops on the IGMP protocol exchange between hosts and routers, and based on that restricts the flooding of IP multicast traffic. In the following, we explore the mechanisms involved in implementing IGMP snooping for VPLS. Please refer to Figure 1 as an example of VPLS with IGMP snooping. In the figure, Router 1 is the Querier. If multiple routers exist on a single subnet (basically that is what a VPLS instance is), they can mutually elect a designated router (DR) that will manage all of the IGMP messages for that subnet. [Page 8] Internet Draft draft-serbest-l2vpn-vpls-mcast-03.txt July, 2005 VPLS Instance +------+ AC1 +------+ +------+ AC4 +------+ | Host |-----| PE |-------------| PE |-----|Router| | 1 | | 1 |\ PW1to3 /| 3 | | 1 | +------+ +------+ \ / +------+ +------+ | \ / | | \ / | | \ /PW2to3 | | \ / | PW1to2| \ |PW3to4 | / \ | | / \PW1to4 | | / \ | | / \ | +------+ +------+ / \ +------+ +------+ | Host | | PE |/ PW2to4 \| PE | |Router| | 2 |-----| 2 |-------------| 4 |-----| 2 | +------+ AC2 +------+ +------+ AC5 +------+ | |AC3 +------+ | Host | | 3 | +------+ Figure 1 Reference Diagram for IGMP Snooping for VPLS 5.3.1. Discovering Multicast Routers A PE need to discover the multicast routers in VPLS instances. This is necessary because: - Designated Router can be different from the Querier on a LAN. - It is not always the Querier that initiates PIM joins - Multicast traffic to the LAN could arrive from a non-querying router because it could be the closest to the source. As recommended in [MAGMA-SNOOP], the PEs can discover multicast routers using Multicast Router Discovery Protocol or they can be statically configured. Since multicast routing protocols other than PIM is out scope, multicast routers can also be discovered by snooping PIM Hello packets as described in Section 5.4.2. . 5.3.2. IGMP Snooping Protocol State The IGMP snooping mechanism described here builds the following state on the PEs. [Page 9] Internet Draft draft-serbest-l2vpn-vpls-mcast-03.txt July, 2005 For each VPLS Instance o Set of Multicast Routers (McastRouters) in the VPLS instance using mechanisms listed in Section 5.2.1. o The IGMP Querying Router (Querier) in the VPLS instance. For each Group entry (*,G) or Source Filtering entry (S,G) in a VPLS instance o Set of interfaces (ACs and/or PWs) from which IGMP membership reports were received. For (*,G) entries, we will call this set igmp_include(*,G). For (S,G) entries, we will call this set igmp_include(S,G). o Set of interfaces from which IGMPv3 hosts have requested to not receive traffic from the specified sources. We will call this set igmp_exclude(S,G). On each interface I, for each (*,G) or (S,G) entry o A Group Timer (GroupTimer(*,G,I)) representing the hold-time for each downstream (*,G) report received on interface I. o A Source Timer (SrcTimer(S,G,I)) representing the hold-time for each downstream (S,G) report received on interface I. 5.3.3. IGMP Join The IGMP snooping mechanism for joining a multicast group (for all IGMP versions) works as follows: - The PE does querier election (by tracking query messages and the source IP addresses) to determine the Querier when there are multiple routers present. Additionally, the query must be received with a non-zero source-ip-address to perform the Querier election - At this point all PEs learn the place of the Querier. For instance, for PE 1 it is behind PW1to3, for PE 2 behind PW2to3, for PE 3 behind AC4, for PE 4 behind PW3to4. - The Querier sends a membership query on the LAN. The membership query can be either general query or group specific query. - PE 3 replicates the query message and forwards it to all PEs participating in the VPLS instance (i.e., PE 1, PE 2, PE 4). - PE 3 keeps a state of {[McastRouters: AC4, PW3to4], [Querier: AC4]}. - All PEs then forward the query to ACs which are part of the VPLS instance. - Suppose that all hosts (Host 1, Host 2, and Host 3) want to participate in the multicast group. [Page 10] Internet Draft draft-serbest-l2vpn-vpls-mcast-03.txt July, 2005 - Host 2 first (for the sake of the example) sends a membership report to the multicast group (e.g., IP: 239.1.1.1, MAC: 01-00-5E- 01-01-01), of which Host 2 wants to be a member. - PE 2 replicates the membership report message and forwards it to all PEs participating in the VPLS instance (i.e., PE 1, PE3, PE 4). - PE 2 notes that there is a directly connected host, which is willing to participate in the multicast group and updates its state to {[McastRouters: PW2to3, PW2to4], [Querier: PW2to3], [igmp_include(*,G):AC2; GroupTimer(*,G,AC2)=GMI]}. Guideline 1: A PE MUST forward a membership report message to ACs that are part of "McastRouters" state. This is necessary to avoid report suppression for other members in order for the PEs to construct correct states and to not have any orphan receiver hosts. There are still some scenarios that can result in orphan receivers. For instance, a multicast router and some hosts could be connected to a customer layer-2 switch, and that layer-2 switch can be connected to a PE via an AC. In such scenario, the customer layer-2 switch MUST perform IGMP snooping as well, and it MUST NOT forward the IGMP report messages coming from the PE to the hosts directly connected to it. There can be some cases such that the layer-2 switch does not have IGMP snooping capability or that device is a dummy hub/bridge. In such cases, one can statically configure the AC, through which the IGMP incapable layer-2 device is connected, to be a (S,G)/(*,G) member on the PE. This way, multicast traffic will always be sent to the hosts connected to that layer-2 device, even they do not send joins because of join suppression. Continuing with the example: - PE 2 does not forward the membership report of Host 2 to Host 3. - Per the guideline above, PE 1 does not forward the membership report of Host 2 to Host 1. - Per the guideline above, PE 3 does forward the membership report of Host 2 to Router 1 (the Querier). - PE 3 notes that there is a host in the VPLS instance, which is willing to participate in the multicast group and updates its state to {[McastRouters: AC4, PW3to4], [Querier: AC4], [igmp_include(*,G): PW2to3], [GroupTimer(*,G,PW2to3)=GMI]} regardless of the type of the query. - Let us assume that Host 1 subsequently sends a membership report to the same multicast group. [Page 11] Internet Draft draft-serbest-l2vpn-vpls-mcast-03.txt July, 2005 - PE 1 replicates the membership report message and forwards it to all PEs participating in the VPLS instance (i.e., PE 2, PE 3, PE 4). - PE 1 notes that there is a directly connected host, which is willing to participate in the multicast group. Basically, it keeps a state of {[McastRouters: PW1to3, PW1to4], [Querier: PW1to3], [igmp_include(*,G): AC1,PW1to2], [GroupTimer(*,G,AC1)=GMI]}. - Per Guideline 1, PE 2 does not forward the membership report of Host 1 to Host 2 and Host 3. - PE 3 and PE 4 receive the membership report message of Host 1 and check their states. Per Guideline 1, they send the report to Router 1 and Router 2 respectively. They also update their states to reflect Host 1. - Now, Host 3 sends a membership report to the same multicast group. - PE 2 updates its state to {[McastRouters: PW2to3, PW2to4], [Querier: PW2to3], [igmp_include(*,G): AC2,AC3,PW1to2], GroupTimer(*,G,AC3)=GMI]}. It then floods the report message to all PEs participating in the VPLS instance. Per Guideline 1, PE 3 forwards the membership report of Host 3 to Router 1, and PE 4 forwards the membership report of Host 3 to Router 2. At this point, all PEs have necessary states to ensure that no multicast traffic will be sent to sites with no members. The previous steps work the same way for IGMPv1 and IGMPv2, when the query is general or source specific. The group and source specific query for IGMPv3 is considered separately below. In IGMPv3, there is no simple membership join or leave report. IGMPv3 reports are one of IS_INCLUDE, IS_EXCLUDE, ALLOW, BLOCK, TO_INCLUDE, TO_EXCLUDE. The PEs MUST implement the "router behavior" portion of the state machine defined in Section 6 of [RFC3376]. The IGMP snooping mechanism for joining a multicast group (for IGMPv3) works as follows: - The Querier sends a membership query to the LAN. The membership query is group and source specific query with a list of sources (e.g., S1, S2, .., Sn). - PE 3 replicates the query message and forwards it to all PEs participating in the VPLS instance (i.e., PE 1, PE 2, PE 4). - PE 3 keeps a state of {[McastRouters: AC4, PW3to4], [Querier: AC4]}. [Page 12] Internet Draft draft-serbest-l2vpn-vpls-mcast-03.txt July, 2005 - All PEs then forward the query to ACs which are part of the VPLS instance. - Suppose that all hosts (Host 1, Host 2, and Host 3) want to participate in the multicast group. Host 1 and Host 2 want to subscribe to (Sn,G), and Host 3 wants to subscribe to (S3,G). - Host 2 first (for the sake of the example) sends a membership report message with group record type IS_INCLUDE for (Sn,G). - PE 2 replicates the membership report message and forwards it to all PEs participating in the VPLS instance (i.e., PE 1, PE 3, PE 4). - PE 2 notes that there is a directly connected host, which is willing to participate in the multicast group and updates its state to {[McastRouters: PW2to3, PW2to4], [Querier: PW2to3], [igmp_include(Sn,G): AC2], [SrcTimer(Sn,G,AC2)=GMI]}. - Per Guideline 1, PE 2 does not forward the membership report of Host 2 to Host 3. - Per Guideline 1, PE 1 does not forward the membership report of Host 2 to Host 1. - Per Guideline 1, PE 3 does forward the membership report of Host 2 to Router 1 (the Querier). - Per Guideline 1, PE 4 does forward the membership report of Host 2 to Router 2. - PE 3 notes that there is a host in the VPLS instance, which is willing to participate in the multicast group. Basically, it updates its state to {[McastRouters: AC4, PW3to4], [Querier: AC4], [igmp_include(Sn,G): PW2to3], [SrcTimer(Sn,G,PW2to3)=GMI]}. - Likewise, PE 4 updates its state to {[McastRouters: PW3to4, AC5], [Querier: PW3to4], [igmp_include(Sn,G):PW2to4], [SrcTimer(Sn,G,PW2to4)=GMI]}. - Let us say Host 1 now sends a membership report message with group record type IS_INCLUDE for (Sn,G). - Similar procedures are followed by PEs as explained in the previous steps. For instance, PE 1 updates its state to {[McastRouters: PW1to3, PW1to4], [Querier: PW1to3], [igmp_include(Sn,G): PW1to2, AC1], SrcTimer(Sn,G,AC1)=GMI}. PE 3 updates its state to {[McastRouters: AC4, PW3to4], [Querier: AC4], [(S1,G); Hosts: ], [igmp_include(Sn,G): PW2to3, PW1to3], [SrcTimer(Sn,G,PW1to3)=GMI]}. - Finally, Host 3 sends a membership report message with group record type IS_INCLUDE for (S3,G). - PE 2 replicates the membership report message and forwards it to all PEs participating in the VPLS instance (i.e., PE 1, PE 3, PE 4). - Per Guideline 1, PE 2 does not forward the membership report of Host 3 to Host 2. [Page 13] Internet Draft draft-serbest-l2vpn-vpls-mcast-03.txt July, 2005 - Per Guideline 1, PE 1 does not forward the membership report of Host 3 to Host 1. - Per Guideline 1, PE 3 does forward the membership report of Host 3 to Router 1. - Per Guideline 1, PE 4 does forward the membership report of Host 3 to Router 2. - All PEs update their states accordingly. For instance, PE 2 updates its state to {[McastRouters: PW2to3, PW2to4], [Querier: PW2to3], [igmp_include(S3,G): AC3], [igmp_include(Sn,G): PW1to2, AC2], [SrcTimer(S3,G,AC3)=GMI]}. PE 4 updates its state to {[McastRouters: AC5, PW3to4], [Querier: PW3to4], [igmp_include(S3,G): PW2to4], [igmp_include(Sn,G): PW1to4, PW2to4], [SrcTimer(S3,G,PW2to4)=GMI]}. At this point, all PEs have necessary states to not send multicast traffic to sites with no members. Based on above description of IGMPv3 based snooping for VPLS, one may conclude that the PEs MUST have the capability to store (S,G) state and MUST forward/replicate traffic accordingly. This is, however, not MANDATORY. A PE MAY only keep (*,G) based states rather than on a per (S,G) basis with the understanding that this will result in a less efficient IP multicast forwarding within each VPLS instance. Guideline 2: If a PE receives unsolicited report message and if it does not possess a state for that particular multicast group, it MUST flood that unsolicited membership report message to all PEs participating in the VPLS instance, as well as to the multicast router if it is locally attached. 5.3.4. IGMP Leave The IGMP snooping mechanism for leaving a multicast group works as follows: - In the case of IGMPv2, when a PE receives a leave (*,G) message from a host via its AC, it lowers the corresponding GroupTimer(*,G,AC) to "Last Member Query Time" (LMQT). - In the case of IGMPv3, when a PE receives a membership report message with group record type of IS_EXCLUDE or TO_EXCLUDE or BLOCK for (S,G) from a host via its AC, it lowers the corresponding SrcTimer(S,G,AC) for all affected (S,G)s to LMQT. In the following guideline, a "leave (*,G)/(S,G) message" also means IGMPv3 membership report message with group record type of IS_EXCLUDE or TO_EXCLUDE or BLOCK for (S,G). [Page 14] Internet Draft draft-serbest-l2vpn-vpls-mcast-03.txt July, 2005 Guideline 3: A PE MUST NOT forward a leave (*,G)/(S,G) message to ACs participating in the VPLS instance, If the PE still has locally connected hosts or hosts connected over a H-VPLS spoke in its state. Guideline 4: A PE MUST forward a leave (*,G)/(S,G) message to all PEs participating in the VPLS instance. A PE MAY forward the leave (*,G)/(S,G) message to the "McastRouters" ONLY, if there are no member hosts in its state. Guideline 5: If a PE does not receive a (*,G) membership report from an AC before GroupTimer(*,G,AC) expires, the PE MUST remove the AC from its state. In case of IGMPv3, if a PE does not receive a (S,G) membership report from an AC before the SrcTimer(S,G,AC) expires, the PE MUST remove the AC from its state. 5.3.5. Failure Scenarios Up to now, we did not consider any failures, which we will focus in this section. - In case the Querier fails (e.g., AC to the querier fails), another router in the VPLS instance will be selected as the Querier. The new Querier will be sending queries. In such circumstances, the IGMP snooping states in the PEs will be updated/overwritten by the same procedure explained above. - In case a Multicast router fails, the IGMP snooping states in the PEs will be updated/overwritten by the multicast router discovery procedures provided in Section 5.3.1. . - In case a host fails (e.g., AC to the host fails), a PE removes the host from its IGMP snooping state for that particular multicast group. Guidelines 3, 4 and 5 still apply here. - In case a PW (which is in IGMP snooping state) fails, the PEs will remove the PW from their IGMP snooping state. For instance, if PW1to3 fails, then PE 1 will remove PW1to3 from its state as the Querier connection, and PE 3 will remove PW1to3 from its state as one of the host connections. Guidelines 3, 4 and 5 still apply here. After PW is restored, the IGMP snooping states in the PEs will be updated/overwritten by the same procedure explained above. One can implement a dead timer before making any changes to IGMP snooping states upon PW failure. In that case, IGMP snooping states will be altered if the PW can not be restored before the dead timer expires. [Page 15] Internet Draft draft-serbest-l2vpn-vpls-mcast-03.txt July, 2005 5.3.6. Scaling Considerations for IGMP Snooping In scenarios where there are multiple ACs connected to a PE, it is quite likely that IGMP membership reports for the same group are received from multiple ACs. The normal behavior would be to have each of the membership reports sent to McastRouters. But in scenarios where many ACs send IGMP membership reports to the same groups, the burden on all the other PEs can be overwhelming. To make matters worse, there can be a large number of hosts on the same AC that all request IGMP membership reports to the same group. While [IGMPv2] suggests the use of report suppression, [IGMPv3] does not. Regardless, if hosts do not implement report suppression, this can be a scaling issue on the PEs. This section outlines the optimization suggested in [MAGMA-SNOOOP] to perform proxy-querying and proxy- reporting function on the PEs to avoid report explosion. For this optimization, we separate out the IGMP group state on the PEs into downstream state and upstream state. Note that the following sub-sections describe the procedures for (*,G). The same procedures must be be extended to (S,G)s. Furthermore the behavior described is for a downstream PE. While it is very important for downstream PEs to implement the proxy behavior described here, the scalability issues are not as bad on upstream PEs. Optimizing upstream PEs would be designed to alleviate the burden on the upstream CEs. Nevertheless the same procedures can be applied to upstream PEs as well as an added optimization. The only difference would be that ACs would be upstream interface(s) and PWs would be downstream interface(s) for such PEs. 5.3.7. Downstream Proxy Behavior When a IGMP membership Report for a group is received on an AC, the PE adds the AC to the corresponding igmp_include set and resets the GrpTimer to GMI. When an IGMP membership Leave for a group is received on an AC, the PE lowers the corresponding GrpTimer to LMQT and sends out a proxy group-specific query on that AC. When sending the group-specific query, the PE encodes 0.0.0.0 (or :: in case of IPv6) in the source- ip address field. If there is no other host interested in that group, then the AC is removed from the corresponding igmp_include set after the GrpTimer expires. [Page 16] Internet Draft draft-serbest-l2vpn-vpls-mcast-03.txt July, 2005 5.3.8. Upstream Proxy Behavior When the igmp_include set for a group becomes non-null, the PE sends out a proxy IGMP Join report for that group to McastRouters. When the igmp_include set for a group becomes empty, the PE sends out a proxy IGMP Leave report for that group to McastRouters. When the PE receives a general query, it replies with its current snooping state for all groups and group-sources. It also forwards the general query to all ACs thus removing the need for proxy general queries. When the PE receives a group-specific or group-source specific query, the PE does not forward such queries to the ACs. Instead it replies with a proxy report if it has snooping state for that group or group-source. When sending the proxy report, the PE encodes 0.0.0.0 (or :: in the case of IPv6) in the source-ip address field. 5.4. PIM Snooping for VPLS IGMP snooping procedures described above provide efficient delivery of IP multicast traffic in a given VPLS service when end stations are connected to the VPLS. However, when VPLS is offered as a WAN service it is likely that the CE devices are routers and would run PIM between them. To provide efficient IP multicasting in such cases, it is necessary that the PE routers offering the VPLS service do PIM snooping. This section describes the procedures for PIM snooping. PIM is a multicast routing protocol, which runs exclusively between routers. PIM shares many of the common characteristics of a routing protocol, such as discovery messages (e.g., neighbor discovery using Hello messages), topology information (e.g., multicast tree), and error detection and notification (e.g., dead timer and designated router election). On the other hand, PIM does not participate in any kind of exchange of databases, as it uses the unicast routing table to provide reverse path information for building multicast trees. There are a few variants of PIM. In PIM-DM ([PIM-DM]), multicast data is pushed towards the members similar to broadcast mechanism. PIM-DM constructs a separate delivery tree for each multicast group. As opposed to PIM-DM, other PIM versions (PIM-SM [RFC2362], PIM-SSM [PIM-SSM], and BIDIR-PIM [BIDIR-PIM]) invokes a pull methodology instead of push technique. PIM routers periodically exchange Hello messages to discover and maintain stateful sessions with neighbors. After neighbors are discovered, PIM routers can signal their intentions to join/prune specific multicast groups. This is accomplished by having downstream routers send an explicit join message (for the sake of generalization, consider Graft messages for PIM-DM as join messages) [Page 17] Internet Draft draft-serbest-l2vpn-vpls-mcast-03.txt July, 2005 to the upstream routers. The join/prune message can be group specific (*,G) or group and source specific (S,G). In PIM snooping, a PE snoops on the PIM message exchange between routers, and builds its multicast states. Based on the multicast states, it forwards IP multicast traffic accordingly to avoid unnecessary flooding. 5.4.1. PIM Snooping State Summarization Macros The following sets are defined to help build the forwarding state on a PE. Some sets may apply only to a subset of the PIM Protocol variants as noted along with the definition of the sets. pim_joins(*,G) = Set of all downstream interfaces on which PIM (*,G) Joins are received. This set applies only to PIM-SM, PIM-SSM, PIM-BIDIR. pim_joins(S,G) = Set of all downstream interfaces on which PIM (S,G) Joins are received. This set applies only to PIM-SM, PIM-SSM. All_Pim_DM_OifList = If the upstream interface (the interface towards the upstream PIM neighbor) is a PW, then this set is the set of all ACs on which there are PIM neighbors. If the upstream interface is an AC, then this is the set of all interfaces (both AC and PW) on which there are PIM neighbors. This set applies only to PIM-DM. pim_prunes(S,G) = Set of all downstream interfaces on which PIM (S,G) prunes are received. This set applies only to PIM-DM. pim_prunes(S,G,rpt) = Set of all downstream interfaces on which PIM (S,G,rpt) prunes are received. This set applies only to PIM-SM. Pim_oiflist(*,G) = Set of interfaces that PIM contributes to the list of outgoing interfaces to which data traffic must be forwarded on a (*,G) match. Pim_oiflist(S,G) = Set of interfaces that PIM contributes to the list of outgoing interfaces to which data traffic must be forwarded on an (S,G) match. Note that pim_oiflist is not the complete list of outgoing interfaces (oiflist). IGMP/MLD also contribute to this list. For PIM-DM, pim_oiflist(S,G) = All_Pim_DM_OifList (-) pim_prunes(S,G) [Page 18] Internet Draft draft-serbest-l2vpn-vpls-mcast-03.txt July, 2005 For PIM-SM and PIM-SSM, Pim_inherited_oiflist(S,G,rpt) = pim_joins(*,G) (-) pim_prunes(S,G,rpt) pim_oiflist(*,G) = pim_joins(*,G) pim_oiflist(S,G) = pim_inherited_oiflist(S,G,rpt) (+) pim_joins(S,G) For PIM-BIDIR, Pim_oiflist(*,G) = DF(RP(G)) + pim_joins(*,G) Where DF(RP(G)) is the AC/PW towards the router that is the designated forwarder for RP(G). In the following, the mechanisms involved for implementing PIMv2 ([RFC2362]) snooping in VPLS are specified. PIMv1 is out of the scope of this document. Please refer to Figure 2 as an example of VPLS with PIM snooping. VPLS Instance +------+ AC1 +------+ +------+ AC4 +------+ |Router|-----| PE |-------------| PE |-----|Router| | 1 | | 1 |\ PW1to3 /| 3 | | 4 | +------+ +------+ \ / +------+ +------+ | \ / | | \ / | | \ /PW2to3 | | \ / | PW1to2| \ |PW3to4 | / \ | | / \PW1to4 | | / \ | | / \ | +------+ +------+ / \ +------+ +------+ |Router| | PE |/ PW2to4 \| PE | |Router| | 2 |-----| 2 |-------------| 4 |-----| 5 | +------+ AC2 +------+ +------+ AC5 +------+ | |AC3 +------+ |Router| | 3 | +------+ Figure 2 Reference Diagram for PIM Snooping for VPLS [Page 19] Internet Draft draft-serbest-l2vpn-vpls-mcast-03.txt July, 2005 In the following sub-sections, snooping mechanisms for each variety of PIM are specified. 5.4.2. PIM-DM The characteristics of PIM-DM is flood and prune behavior. Shortest path trees are built as a multicast source starts transmitting. In Figure 2, the multicast source is behind Router 4, and all routers have at least one receiver except Router 3 and Router 5. 5.4.2.1. Discovering Multicast Routers The PIM-DM snooping mechanism for neighbor discovery works as follows: - To establish PIM neighbor adjacencies, PIM multicast routers (all routers in this example) send PIM Hello messages to the ALL PIM Routers group address (IPv4: 224.0.0.13, MAC: 01-00-5E-00-00- 0D) on every PIM enabled interfaces. The IPv6 ALL PIM Routers group is "ff02::d". In addition, PIM Hello messages are used to elect Designated Router for a multi-access network. In PIM-DM, the DR acts as the Querier if IGMPv1 is used. Otherwise, DR has no function in PIM-DM. Guideline 6: PIM Hello messages MUST be flooded in the VPLS instance. A PE MUST populate its "PIM Neighbors" list according to the snooping results. This is a general PIM snooping guideline and applies to all variants of PIM snooping. Guideline 7: For PIM-DM only. pim_oiflist(S,G) is populated with All_Pim_DM_Interfaces (the ACs/PWs in the "PIM Neighbors" list). Changes to the "PIM Neighbors" list MUST be replicated to All_Pim_DM_Interfaces. - Every router starts sending PIM Hello messages. Per Guideline 6, every PE replicates Hello messages and forwards them to all PEs participating in the VPLS instance. - Based on PIM Hello exchanges PE routers populate PIM snooping states as follows. PE 1: {[(,); Source:; Flood to: AC1, PW1to2, PW1to3, PW1to4], [PIM Neighbors: (Router 1,AC1), (Router 2,Router 3,PW1to2), (Router 4,PW1to3), (Router 5,PW1to4)] }, PE 2: {[(,); Source:; Flood to: AC2, AC3, PW1to2, PW2to3, PW2to4], [PIM Neighbors: (Router 1,PW1to2), (Router 2,AC2), (Router 3,AC3), (Router 4,PW2to3), (Router 5,PW2to4)]}, PE 3: {[(,); Source:; Flood to: AC4, PW1to3, PW2to3, PW3to4], [PIM Neighbors: (Router 1,PW1to3), (Router 2,Router 3,PW2to3), (Router 4,AC4), (Router [Page 20] Internet Draft draft-serbest-l2vpn-vpls-mcast-03.txt July, 2005 5,PW3to4)]}, PE 4: {[(,); Source:; Flood to: AC5, PW1to4, PW2to4, PW3to4], [PIM Neighbors: (Router 1,PW1to4), (Router 2,Router 3,PW2to4), (Router 4,PW3to4), (Router 5,AC5)]}. The original pim_oiflist(S,G) is populated with ACs/PWs in the PIM neighbor list per Guideline 7.. - PIM Hello messages contain a Holdtime value, which tells the receiver when to expire the neighbor adjacency (which is three times the Hello period). Guideline 8: If a PE does not receive a Hello message from a router within its Holdtime, the PE MUST remove that router from the PIM snooping state. If a PE receives a Hello message from a router with Holdtime value set to zero, the PE MUST remove that router from the PIM snooping state immediately. PEs MUST track the Hello Holdtime value per PIM neighbor. 5.4.2.2. PIM-DM Multicast Forwarding The PIM-DM snooping mechanism for multicast forwarding works as follows: - When the source starts sending traffic to multicast group (S,G), PE 3 updates its state to PE 3: {[(S,G) ; Source: (Router 4,AC4); pim_oiflist(S,G): PW1to3, PW2to3, PW3to4], [PIM Neighbors: (Router 1,PW1to3), (Router 2,Router 3,PW2to3), (Router 4,AC4), (Router 5,PW3to4)]}. AC4 is removed from the pim_oiflist list for (S,G), since it is where the multicast traffic comes from. Guideline 9: Multicast traffic MUST be replicated per PW and AC basis, i.e., even if there are more than one PIM neighbor behind a PW/AC, only one replication MUST be sent to that PW/AC. - PE 3 replicates the multicast traffic and sends it to the other PE routers in its pim_oiflist(S,G). - Consequently, all PEs update their states as follows. PE 1: {[(S,G); Source: (Router 4,PW1to3); pim_oiflist(S,G): AC1], [PIM Neighbors: (Router 1,AC1), (Router 2,Router 3,PW1to2), (Router 4,PW1to3), (Router 5,PW1to4)]}, PE 2: {[(S,G); Source: (Router 4,PW2to3); pim_oiflist(S,G): AC2, AC3], [PIM Neighbors: (Router 1,PW1to2), (Router 2,AC2), (Router 3,AC3), (Router 4,PW2to3), (Router 5,PW2to4)]}, PE 4: {[(S,G); Source: (Router 4,PW3to4); pim_oiflist(S,G): AC5], [PIM Neighbors: (Router 1,PW1to4), (Router 2,Router 3,PW2to4), (Router 4,PW3to4), (Router 5,AC5)]}. 5.4.2.3. PIM-DM Pruning [Page 21] Internet Draft draft-serbest-l2vpn-vpls-mcast-03.txt July, 2005 At this point all the routers (Router 1, Router 2,Router 3, Router 5) receive the multicast traffic. - However, Router 3 and Router 5 do not have any members for that multicast group, so they send prune messages to leave the multicast group to the ALL PIM Routers group. PE 2 updates its state to PE 2: {[(S,G); Source: (Router 4,PW2to3); pim_prunes(S,G): AC3, pim_oiflist(S,G): AC2], [PIM Neighbors: (Router 1,PW1to2), (Router 2,AC2), (Router 3,AC3), (Router 4,PW2to3), (Router 5,PW2to4)]}. PE 4 also removes Router 3 and Router 5 from its state as well. Guideline 10:.The PIM-DM prune message MUST be forwarded towards the upstream PE only if pim_oiflist(S,G) became empty as a result of the received prune message. If pim_oiflist(S,G) was already null when the PIM-DM prune was received, then the prune MUST NOT be forwarded upstream. - PE 2 does not forward the prune message per Guideline 10. PE 4 updates its state to PE 4: {[(S,G); Source: (Router 4,AC4); pim_prunes(S,G): AC5, pim_oiflist(S,G):], [PIM Neighbors: (Router 1,PW1to4), (Router 2,Router 3,PW2to4), (Router4, PW3to4). - PIM-DM prune messages contain a Holdtime value, which specifies how many seconds the prune state should last. Guideline 11: For PIM-DM only. A PE MUST keep the prune state for a PW/AC according to the Holdtime in the prune message, unless a corresponding Graft message is received. - Upon receiving the prune messages, each PE 3 updates its state accordingly to PE 3: {[(S,G); Source: (Router 4,AC4); pim_prunes(S,G): PW2to4, pim_oiflist(S,G): PW1to3, PW2to3], [PIM Neighbors: (Router 1,PW1to3), (Router 2,Router 3,PW2to3), (Router 4,AC4), (Router 5, PW3to4)]}. Guideline 12: For PIM-DM only. To avoid overriding joins, a PE SHOULD suppress the PIM prune messages to directly connected routers (i.e., ACs), as long as there is a PW/AC in its corresponding pim_oiflist(S,G). - In this case, PE 1, PE 2, and PE 3 do not forward the prune messages to their directly connected routers. 5.4.2.4. PIM-DM Grafting The multicast traffic is now flowing only to points in the network where receivers are present. [Page 22] Internet Draft draft-serbest-l2vpn-vpls-mcast-03.txt July, 2005 Guideline 13: For PIM-DM only. A PE MUST remove the AC/PW from its corresponding prune state (pim_prunes(S,G)) when it receives a graft message from the AC/PW. That is, the corresponding AC/PW MUST be added to the pim_oiflist(S,G) list. Guideline 14: For PIM-DM only. PIM-DM graft messages MUST be forwarded based on the destination MAC address. If the destination MAC address is 01-00-5E-00-00-0D, then the graft message MUST be flooded in the VPLS instance. PIM-DM graft messages MUST NOT be flooded if pim_oiflist is non-null. - For the sake of example, suppose now Router 3 has a receiver the multicast group (S,G). Assuming Router 3 sends a graft message in IP unicast to Router 4 to restart the flow of multicast traffic. PE 2 updates its state to PE 2: {[(S,G); Source: (Router 4,PW2to3); pim_prunes(S,G): , pim_oiflist(S,G): AC2, AC3], [PIM Neighbors: (Router 1,PW1to2), (Router 2,AC2), (Router 3,AC3), (Router 4,PW2to3), (Router 5,PW2to4)]}. PE 2 then forwards the graft message to PE 3 according to Guideline 14. - Upon receiving the graft message, PE 3 updates its state accordingly to PE 3: {[(S,G); Source: (Router 4,AC4); pim_prunes(S,G): PW3to4, pim_oiflist(S,G): PW1to3, PW2to3], [PIM Neighbors: (Router 1,PW1to3), (Router 2,Router 3,PW2to3), (Router 4,AC4), (Router 5,PW3to4)]}. 5.4.2.5. Failure Scenarios Guideline 15: PIM Assert messages MUST be flooded in the VPLS instance. Guideline 16: If an AC/PW goes down, a PE MUST remove it from its PIM snooping state. Failures can be easily handled in PIM-DM snooping, as it uses push technique. If an AC or a PW goes down, PEs in the VPLS instance will remove it from their snooping state (if the AC/PW is not already pruned). After the AC/PW comes back up, it will be automatically added to the snooping state by PE routers, as all PWs/ACs MUST be in the snooping state, unless they are pruned later on. 5.4.3. PIM-SM The key characteristics of PIM-SM is explicit join behavior. In this model, the multicast traffic is only sent to locations that specifically request it. The root node of a tree is the Rendezvous Point (RP) in case of a shared tree or the first hop router that is [Page 23] Internet Draft draft-serbest-l2vpn-vpls-mcast-03.txt July, 2005 directly connected to the multicast source in the case of a shortest path tree. In Figure 2, the RP is behind Router 4, and all routers have at least one member except Router 3 and Router 5. As in the case with IGMPv3 snooping, we assume that the PEs have the capability to store (S,G) states for PIM-SM snooping and forward/replicate traffic accordingly. This is not mandatory. An implementation, can fall back to (*,G) states, if its hardware can not support it. In such case, the efficiency of multicast forwarding will be less. 5.4.3.1. Discovering Multicast Routers The PIM-SM snooping mechanism for neighbor discovery works the same way as the procedure defined in PIM-DM section, with the exception of PIM-DM only guidelines. - Based on PIM Hello exchanges PE routers populate PIM snooping states as follows. PE 1: {[(,); Flood to:], [PIM Neighbors: (Router 1,AC1), (Router 2,Router 3,PW1to2), (Router 4,PW1to3), (Router 5,PW1to4)]}, PE 2: {[(,); Flood to:], [PIM Neighbors: (Router 1,PW1to2), (Router 2,AC2), (Router 3,AC3), (Router 4,PW2to3), (Router 5,PW2to4)]}, PE 3: {[(,); Flood to:], [PIM Neighbors: (Router 1,PW1to3), (Router 2,Router 3,PW2to3), (Router 4,AC4), (Router 5,PW3to4)]}, PE 4: {[(,); Flood to:], [PIM Neighbors: (Router 1,PW1to4), (Router 2,Router 3,PW2to4), (Router 4,PW3to4), (Router 5,AC5)]}. To reduce the amount of PIM Join/Prune traffic in the VPLS network, it is important that Explicit-Tracking capability be disabled between the CEs. If a CE advertises tracking support, it is recommended that the PEs modify the tracking-support option in CE Hello packets before forwarding them to ensure that tracking support is disabled between the CEs. Otherwise, the mechanism listed for "JP_Optimization" throughout the PIM-SM and PIM-SSM sections of this document MUST NOT be employed. NOTE: The examples below are for scenarios where JP_Optimization is not employed. For PIM-SM to work properly, all routers within the domain must use the same mappings of group addresses to RP addresses. Currently, there are three methods for RP discovery: 1. Static RP configuration, 2, Auto-RP, and 3. PIMv2 Bootstrap Router mechanism. 5.4.3.2. PIM-SM (*,G) Join [Page 24] Internet Draft draft-serbest-l2vpn-vpls-mcast-03.txt July, 2005 The PIM-SM snooping mechanism for joining a multicast group (*,G) works as follows: Guideline 18: PIM-SM join messages MUST be sent only to the remote PE, which is connected to the router to which the Join is addressed. JP_Optimization: The PIM-SM join message MUST be forwarded towards the upstream CE only if pim_joins(*,G) became non-empty as a result of the received join message. If pim_joins(*,G) was already non-null when the PIM-SM join was received, then the join MUST NOT be forwarded upstream. PIM-SM join messages MUST be sent only to the remote PE, which is connected to the router to which the Join is addressed. The remote PE can be determined by the "Upstream Neighbor Address" field of the Join message. The "Upstream Neighbor Address" can be correlated to a PW or an AC in the "PIM Neighbors" state. By Guideline 18, we are ensuring that the other routers that are part of the VPLS instance do not receive the PIM join messages and will initiate their own join messages if they are interested in receiving that particular multicast traffic. - Assume Router 1 wants to join the multicast group (*,G) sends a join message for the multicast group (*,G). PE 1 sends the join message to PE 3 by Guideline 18. Guideline 19: A PE MUST add a PW/AC to its pim_joins(*,G) list, if it receives a (*,G) join message from the PW/AC. - PE 1 updates their states as follows: PE 1: {[pim_joins(*,G): AC1], [PIM Neighbors: (Router 1,AC1), (Router 2,Router 3,PW1to2), (Router 4,PW1to3), (Router 5,PW1to4)]}. A periodic refresh mechanism is used in PIM-SM to maintain the proper state. PIM-SM join messages contain a Holdtime value, which specifies for how many seconds the join state should be kept. Guideline 20: If a PE does not receive a refresh join message from a PW/AC within its Holdtime, the PE MUST remove the PW/AC from its pim_joins(*,G) list. - All PEs update their states accordingly as follows: PE 1: {[pim_joins(*,G): AC1], [PIM Neighbors: (Router 1,AC1), (Router 2,Router 3,PW1to2), (Router 4,PW1to3), (Router 5,PW1to4)]}, PE 2: {[(,); Flood to: ], [PIM Neighbors: (Router 1,PW1to2), (Router 2,AC2), (Router 3,AC3), (Router 4,PW2to3), (Router 5,PW2to4)]}, PE 3: {[pim_joins(*,G): PW1to3], [PIM Neighbors: (Router 1,PW1to3), [Page 25] Internet Draft draft-serbest-l2vpn-vpls-mcast-03.txt July, 2005 (Router 2,Router 3,PW2to3), (Router 4,AC4), (Router 5,PW3to4)]}, PE 4: {[(,); Flood to: ], [PIM Neighbors: (Router 1,PW1to4), (Router 2,Router 3,PW2to4), (Router 4,PW3to4), (Router 5,AC5)]}. - After Router 2 joins the same multicast group, the states become as follows: PE 1: {[pim_joins(*,G): AC1], [PIM Neighbors: (Router 1,AC1), (Router 2,Router 3,PW1to2), (Router 4,PW1to3), (Router 5,PW1to4)]}, PE 2: {[pim_joins(*,G): AC2], [PIM Neighbors: (Router 1,PW1to2), (Router 2,AC2), (Router 3,AC3), (Router 4,PW2to3), (Router 5,PW2to4)]}, PE 3: {[pim_joins(*,G): PW1to3, PW2to3], [PIM Neighbors: (Router 1,PW1to3), (Router 2,Router 3,PW2to3), (Router 4,AC4), (Router 5,PW3to4)]}, PE 4: {[(,); Flood to: ], [PIM Neighbors: (Router 1,PW1to4), (Router 2,Router 3,PW2to4), (Router 4,PW3to4), (Router 5,AC5)]}. - For the sake of example, Router 3 joins the multicast group. PE 2 sends the join message to PE 3. - Next Router 5 joins the group, and the states are updated accordingly: PE 1: {[pim_joins(*,G): AC1], [PIM Neighbors: (Router 1,AC1), (Router 2,Router 3,PW1to2), (Router 4,PW1to3), (Router 5,PW1to4)]}, PE 2: {[pim_joins(*,G): AC2, AC3], [PIM Neighbors: (Router 1,PW1to2), (Router 2,AC2), (Router 3,AC3), (Router 4,PW2to3), (Router 5,PW2to4)]}, PE 3: {[pim_joins(*,G): PW1to3, PW2to3, PW3to4], [PIM Neighbors: (Router 1,PW1to3), (Router 2,Router 3,PW2to3), (Router 4,AC4), (Router 5,PW3to4)]}, PE 4: {[pim_joins(*,G): AC5],[PIM Neighbors: (Router 1,PW1to4), (Router 2,Router 3,PW2to4), (Router 4,PW3to4), (Router 5,AC5)]} At this point, all PEs have necessary states to not send multicast traffic to sites with no members. 5.4.3.3. PIM-SM Pruning The PIM-SM snooping mechanism for leaving a multicast group works as follows: - Assume Router 5 sends a prune message. Guideline 21: PIM-SM prune messages MUST be flooded in the VPLS instance. JP_Optimization: Instead of the above guideline, a PE MUST forward prune messages only towards the upstream CE and only if pim_joins(*,G) became empty as a result of the received prune message. If pim_joins(*,G) is non-empty after receiving the prune message, the PE MUST NOT forward the prune message. Guideline 22: A PE MUST remove a PW/AC from its pim_joins(*,G) list if it receives a (*,G) prune message from the PW/AC. A prune-delay timer SHOULD be implemented to support prune override. [Page 26] Internet Draft draft-serbest-l2vpn-vpls-mcast-03.txt July, 2005 However, the prune-delay timer is not required if there is only one PIM neighbor on that AC/PW on which the prune was received. - PE 4 floods the (*,G) prune to the VPLS instance per Guideline 21. PE routers participating in the VPLS instance also forward the (*,G) prune to the ACs, which are connected to the VPLS instance. The states are updated as follows: PE 1: {[pim_joins(*,G): AC1], [PIM Neighbors: (Router 1,AC1), (Router 2,Router 3,PW1to2), (Router 4,PW1to3), (Router 5,PW1to4)]}, PE 2: {[pim_joins(*,G): AC2, AC3], [PIM Neighbors: (Router 1,PW1to2), (Router 2,AC2), (Router 3,AC3), (Router 4,PW2to3), (Router 5,PW2to4)]}, PE 3: {[pim_joins(*,G): PW1to3, PW2to3], [PIM Neighbors: (Router 1,PW1to3), (Router 2,Router 3,PW2to3), (Router 4,AC4), (Router 5,PW3to4)]}, PE 4: {[(,); Flood to: ],[PIM Neighbors: (Router 1,PW1to4), (Router 2,Router 3,PW2to4), (Router 4,PW3to4), (Router 5,AC5)]}. In PIM-SM snooping, prune messages are flooded by PE routers. In such implementation, PE routers may receive overriding join messages, which will not affect anything. 5.4.3.4. PIM-SM (S,G) Join The PIM-SM snooping mechanism for source and group specific join works as follows: Guideline 23: A PE MUST add a PW/AC to its pim_joins(S,G) list if it receives a (S,G) join message from the PW/AC. The PE MUST forward the received join message towards the upstream CE. JP_Optimization: The PE MUST forward the Join message towards the upstream neighbor only if the pim_joins(S,G) list becomes non- empty as a result of the received join. If the pim_joins(S,G) list was non-empty prior to receiving the join message, then the PE MUST NOT forward the join message. Guideline 24: A PE MUST remove a PW/AC from its pim_joins(S,G) list if it receives a (S,G) prune message from the PW/AC. The PE MUST flood the prune message in the VPLS instance. A prune-delay timer SHOULD be implemented to support prune override on the downstream AC/PW. However, the prune-delay timer is not required if there is only one PIM neighbor on that AC/PW on which the prune was received. JP_Optimization: Instead of flooding the prune message in the VPLS instance, the PE MUST forward the prune message towards the upstream neighbor only if the pim_joins(S,G) list becomes empty as a result of the received prune. If the pim_joins(S,G) list remains [Page 27] Internet Draft draft-serbest-l2vpn-vpls-mcast-03.txt July, 2005 non-empty after receiving the prune message, then the PE MUST NOT forward the prune message. Guideline 25: A PE MUST prefer (S,G) state to (*,G), if both S and G match. 5.4.3.5. PIM-SM (S,G,rpt) Prunes Guideline 28: When a PE receives a Prune(S,G,rpt) on an AC/PW, it MUST add the AC/PW to the pim_prunes(S,G,rpt) list. Additionally, if pim_snoop_inherited_olist(S,G,rpt) becomes empty, the PE MUST forward the Prune(S,G,rpt) towards the upstream neighbor. If pim_snoop_inherited_olist(S,G,rpt) is still non-empty, then the PE MUST NOT forward the Prunes(S,G,rpt). 5.4.3.6. PIM-SM (*,*,RP) State PIM-SM defines a (*,*,RP) state which is used when traffic needs to cross multicast domains. A (*,*,RP) receiver requests all multicast traffic within a PIM domain to be sent to it. If the two multicast domains are both PIM-SM, they can use MSDP to leak multicast routes. But, if one is PIM-SM and the other is PIM-DM (hence, MSDP can not be used), then the border router would initiate a (*,*,RP) join to all RPs in the PIM-SM domain. If the customers will configure multiple and different PIM domains, PIM-SM snooping MUST support (*,*,RP) state as well. Depending on how likely scenario this is, future versions may include (*,*,RP) states. 5.4.3.7. Failure Scenarios Failures can be easily handled in PIM-SM snooping, as it employs state-refresh technique. PEs in the VPLS instance will remove any entry for non-refreshing routers from their states. 5.4.3.8. Special Cases for PIM-SM Snooping There are some special cases to consider for PIM-SM snooping. First one is the RP-on-a-stick. The RP-on-a-stick scenario may occur when the Shortest Path Tree and the Shared Tree shares a common Ethernet segment, as all routers will be connected over a multicast access network (i.e., VPLS). Such a scenario will be handled by PIM-SM rules (particularly, the incoming interface can not also appear in the outgoing interface list) very nicely. Second scenario is the turnaround router. The turnaround router scenario occurs when shortest path tree and shared tree share a common path. The router at which these tree merge is the turnaround router. PIM-SM handles this case by proxy (S,G) join implementation by the turnaround router. [Page 28] Internet Draft draft-serbest-l2vpn-vpls-mcast-03.txt July, 2005 There can be some scenarios where CE routers can receive duplicate multicast traffic. Let us consider the scenario in Figure 3. +------+ AC3 +------+ | PE2 |-----| R3 | /| | | | / +------+ +------+ / | | / | | /PW1to2 | | / | +-----+ / |PW2to3 | Src | / | +-----+ / | | / | | / | | +------+ +------+ / +------+ +------+ | R1 | | PE1 |/ PW1to3 | PE3 | | R4 | | |-----| |-------------| |-----| | +------+ AC1 +------+ +------+ AC4 +------+ | | |AC2 |AC5 +------+ +------+ | R2 | | R5 | +---+ | | | |-----|RP | +------+ +------+ +---+ Figure 3 CE Routers Receive Duplicate Traffic In the scenario depicted in Figure 3, both R1 and R2 has two ECMP routes to reach the source Src. Hence, R1 may pick R3 as its next hop ("Upstream Neighbor"), and R2 may pick R4 as its next hop. As a result, both R1 and R2 will receive duplicate traffic. [Page 29] Internet Draft draft-serbest-l2vpn-vpls-mcast-03.txt July, 2005 This issue can be solved as follows. PEs can keep the PW/AC that the join message is forwarded to (upstream PW/AC) in "Flood to" list in addition to the PW/AC that the join message is received (downstream PWAC). If the traffic arrives from a different PW/AC, that traffic is not forwarded downstream. Hence, in the example depicted in Figure 3 where source is dual homed to R3 and R4, R1 will receive (S,G) traffic if it comes from PW1to2, and R2 will receive (S,G) traffic if it comes from PW1to3. Again, in Figure 3, R1 may send (S,G) join to R3 and R2 may send (*,G) join to the RP behind R5. In this scenario as well, both R1 and R2 will receive duplicate traffic, as Guideline 25 will be no help to prevent it. In this case, where R1 joins for (S,G), and R2 joins for (*,G), we can do the following. The PEs SHOULD keep the upstream PW/AC in the state as described above. In addition, the PEs need to act on (S,G,RPT) prunes and remove the related upstream PW/AC from "Flood to" list of (S,G) state copied from (*,G) state. As a result, Ces will not receive duplicate traffic. However, there will still be bandwidth waste as the egress PE takes care of the duplicate traffic problem. We can further enhance the proposal by triggering Assert mechanism in CE routers. The PE which detects the duplicate traffic problem can simply remove the snooping state for that particular multicast group, and can send out "flush" message to other PEs participating in the VPLS instance. In return, other PEs also flush their snooping state for that multicast group. As a result, all the PEs will flood the multicast traffic in the VPLS instance (by Rule 3). Consequently, CEs will do Assert. The flush message TLV can be sent over the targeted LDP sessions running among PEs. Future versions will include the details. 5.4.4. PIM-SSM The key characteristics of PIM-SSM is explicit join behavior, but it eliminates the shared tree and the rendezvous point in PIM-SM. In this model, a shortest path tree for each (S,G) is built with the first hop router (that is directly connected to the multicast source) being the root node. PIM-SSM is ideal for one-to-many multicast services. In Figure 2, S1 is behind Router 1, and S4 is behind Router 4. Routers 2 and 4 want to join (S1,G), while Router 5 wants to join (S4,G). We assume that the PEs have the capability to store (S,G) states for PIM-SSM snooping and constrain multicast flooding scope accordingly. An implementation, can fall back to (*,G) states, if its hardware can not support it. In such case, the efficiency of multicast forwarding will be less. [Page 30] Internet Draft draft-serbest-l2vpn-vpls-mcast-03.txt July, 2005 5.4.4.1. Discovering Multicast Routers The PIM-SSM snooping mechanism for neighbor discovery works the same way as the procedure defined in PIM-DM section, with the exception of PIM-DM only guidelines. - Based on PIM Hello exchanges PE routers populate PIM snooping states as follows. PE 1: {[(,); Flood to:], [PIM Neighbors: (Router 1,AC1), (Router 2,Router 3,PW1to2), (Router 4,PW1to3), (Router 5,PW1to4)]}, PE 2: {[(,); Flood to:], [PIM Neighbors: (Router 1,PW1to2), (Router 2,AC2), (Router 3,AC3), (Router 4,PW2to3), (Router 5,PW2to4)]}, PE 3: {[(,); Flood to:], [PIM Neighbors: (Router 1,PW1to3), (Router 2,Router 3,PW2to3), (Router 4,AC4), (Router 5,PW3to4)]}, PE 4: {[(,); Flood to:], [PIM Neighbors: (Router 1,PW1to4), (Router 2,Router 3,PW2to4), (Router 4,PW3to4), (Router 5,AC5)]}. 5.4.4.2. Guidelines for PIM-SSM Snooping PIM-SSM snooping is actually simpler than PIM-SM and only the following guidelines (some of which are repetitions from PIM-SM section) apply. Guideline 28: A PE MUST add a PW/AC to its (S,G) pim_joins(S,G) list if it receives a (S,G) join message from the PW/AC. Guideline 29: PIM-SSM join messages MUST be sent only to the remote PE, which is connected to the router to which the Join is addressed. JP_Optimization: The PE MUST forward the Join message towards the upstream neighbor only if the pim_joins(S,G) list becomes non- empty as a result of the received join. If the pim_joins(S,G) list was non-empty prior to receiving the join message, then the PE MUST NOT forward the join message. Guideline 30: PIM prune messages MUST be flooded in the VPLS instance. A prune-delay timer SHOULD be implemented to support prune override on the downstream AC/PW. However, the prune-delay timer is not required if there is only one PIM neighbor on that AC/PW on which the prune was received. JP_Optimization: Instead of flooding the prune message in the VPLS instance, the PE MUST forward the prune message towards the upstream neighbor only if the pim_joins(S,G) list becomes empty as a result of the received prune. If the pim_joins(S,G) list remains non-empty after receiving the prune message, then the PE MUST NOT forward the prune message. [Page 31] Internet Draft draft-serbest-l2vpn-vpls-mcast-03.txt July, 2005 Guideline 31: If A PE does not receive a refresh join message from a PW/AC within its Holdtime, the PE MUST remove the PW/AC from its pim_joins(S,G) list. Guideline 32: A PE MUST remove a PW/AC from its pim_joins(S,G) list if it receives a (S,G) prune message from the PW/AC. A prune-delay timer SHOULD be implemented to support prune override. 5.4.4.3. PIM-SSM Join The PIM-SSM snooping mechanism for joining a multicast group works as follows: - Assume Router 2 requests to join the multicast group (S1,G). - PE 2 updates its state, and then sends the join message to PE 1. - All PEs update their states as follows: PE 1: {[(S1,G); Flood to: PW1to2], [PIM Neighbors: (Router 1,AC1), (Router 2,Router 3,PW1to2), (Router 4,PW1to3), (Router 5,PW1to4)]}, PE 2: {[pim_joins(S1,G): AC2], [PIM Neighbors: (Router 1,PW1to2), (Router 2,AC2), (Router 3,AC3), (Router 4,PW2to3), (Router 5,PW2to4)]}, PE 3: {[(,); Flood to: ], [PIM Neighbors: (Router 1,PW1to3), (Router 2,Router 3,PW2to3), (Router 4,AC4), (Router 5,PW3to4)]}, PE 4: {[(,); Flood to: ], [PIM Neighbors: (Router 1,PW1to4), (Router 2,Router 3,PW2to4), (Router 4,PW3to4), (Router 5,AC5)]}. - Next, assume Router 4 sends a join (S1,G) message. Following the same procedures, all PEs update their states as follows: PE 1: {[pim_joins(S1,G): PW1to2, PW1to3], [PIM Neighbors: (Router 1,AC1), (Router 2,Router 3,PW1to2), (Router 4,PW1to3), (Router 5,PW1to4)]}, PE 2: {[pim_joins(S1,G): AC2], [PIM Neighbors: (Router 1,PW1to2), (Router 2,AC2), (Router 3,AC3), (Router 4,PW2to3), (Router 5,PW2to4)]}, PE 3: {[pim_joins(S1,G): AC4], [PIM Neighbors: (Router 1,PW1to3), (Router 2,Router 3,PW2to3), (Router 4,AC4), (Router 5,PW3to4)]}, PE 4: {[(,); Flood to: ], [PIM Neighbors: (Router 1,PW1to4), (Router 2,Router 3,PW2to4), (Router 4,PW3to4), (Router 5,AC5)]}. - Then, assume Router 5 requests to join the multicast group (S4,G). After the same procedures are applied, all PEs update their states as follows: PE 1: {[pim_joins(S1,G): PW1to2, PW1to3], [PIM Neighbors: (Router 1,AC1), (Router 2,Router 3,PW1to2), (Router 4,PW1to3), (Router 5,PW1to4)]}, PE 2: {[pim_joins(S1,G): AC2], [PIM Neighbors: (Router 1,PW1to2), (Router 2,AC2), (Router 3,AC3), (Router 4,PW2to3), (Router 5,PW2to4)]}, PE 3: {[pim_joins(S1,G): AC4], [pim_joins(S4,G): PW3to4], [PIM Neighbors: (Router 1,PW1to3), (Router 2,Router 3,PW2to3), (Router [Page 32] Internet Draft draft-serbest-l2vpn-vpls-mcast-03.txt July, 2005 4,AC4), (Router 5,PW3to4)]}, PE 4: {[pim_joins(S4,G): AC5], [PIM Neighbors: (Router 1,PW1to4), (Router 2,Router 3,PW2to4), (Router 4,PW3to4), (Router 5,AC5)]}. 5.4.4.4. PIM-SSM Prune At this point, all PEs have necessary states to not send multicast traffic to sites with no members. The PIM-SSM snooping mechanism for leaving a multicast group works as follows: Assume Router 2 sends a (S1,G) prune message to leave the multicast group. The prune message gets flooded in the VPLS instance. All PEs update their states as follows: PE 1: {[pim_joins(S1,G): PW1to3], [PIM Neighbors: (Router 1,AC1), (Router 2,Router 3,PW1to2), (Router 4,PW1to3), (Router 5,PW1to4)]}, PE 2: {[Deletes (S1,G) state], [PIM Neighbors: (Router 1,PW1to2), (Router 2,AC2), (Router 3,AC3), (Router 4,PW2to3), (Router 5,PW2to4)]}, PE 3: {[(S1,G); Flood to: AC4], [(S4,G); Flood to: PW3to4], [PIM Neighbors: (Router 1,PW1to3), (Router 2,Router 3,PW2to3), (Router 4,AC4), (Router 5,PW3to4)]}, PE 4: {[(S4,G); Flood to: AC5], [PIM Neighbors: (Router 1,PW1to4), (Router 2,Router 3,PW2to4), (Router 4,PW3to4), (Router 5,AC5)]}. In PIM-SSM snooping, prune messages are flooded by PE routers. In such implementation, PE routers may receive overriding join messages, which will not affect anything. 5.4.4.5. Failure Scenarios Similar to PIM-SSM snooping, failures can be easily handled in PIM- SSM snooping, as it employs state-refresh technique. The PEs in the VPLS instance will remove entry for non-refreshing routers from their states. 5.4.4.6. Special Cases for PIM-SSM Snooping The scenarios with duplicate traffic as depicted in Figure 3 apply to PIM-SSM snooping as well. Again, the issue can be solved by the method described in Section 5.4.3.8. . 5.4.5. Bidirectional-PIM (BIDIR-PIM) BIDIR-PIM is a variation of PIM-SM. The main differences between PIM-SM and Bidirectional-PIM are as follows: - There are no source-based trees, and source-specific multicast is not supported (i.e., no (S,G) states) in BIDIR-PIM. - Multicast traffic can flow up the shared tree in BIDIR-PIM. - To avoid forwarding loops, one router on each link is elected as the Designated Forwarder (DF) for each RP in BIDIR-PIM. [Page 33] Internet Draft draft-serbest-l2vpn-vpls-mcast-03.txt July, 2005 The main advantage of BIDIR-PIM is that it scales well for many-to- many applications. However, the lack of source-based trees means that multicast traffic is forced to remain on the shared tree. In Figure 2, the RP for (*,G4) is behind Router 4, and the RP for (*,G1) is behind Router 1. Router 2 and Router 4 want to join (*,G1), whereas Router 5 wants to join (*,G4). On the VPLS instance, Router 4 is the DF for the RP of (*,G4), and Router 1 is the DF of the RP for (*,G1). 5.4.5.1. Discovering Multicast Routers The PIM-SSM snooping mechanism for neighbor discovery works the same way as the procedure defined in PIM-DM section, with the exception of PIM-DM only guidelines. - Based on PIM Hello exchanges PE routers populate PIM snooping states as follows. PE 1: { [PIM Neighbors: (Router 1,AC1), (Router 2,Router 3,PW1to2), (Router 4,PW1to3), (Router 5,PW1to4)] }, PE 2: { [PIM Neighbors: (Router 1,PW1to2), (Router 2,AC2), (Router 3,AC3), (Router 4,PW2to3), (Router 5,PW2to4)] }, PE 3: {[PIM Neighbors: (Router 1,PW1to3), (Router 2,Router 3,PW2to3), (Router 4,AC4), (Router 5,PW3to4)]}, PE 4: { [PIM Neighbors: (Router 1,PW1to4), (Router 2,Router 3,PW2to4), (Router 4,PW3to4), (Router 5,AC5)]}. For BIDIR-PIM to work properly, all routers within the domain must know the address of the RP. There are three methods to do that: 1. Static RP configuration, 2, Auto-RP, and 3. PIMv2 Bootstrap. Guideline 17 applies here as well. During RP discovery time, PIM routers elect DF per subnet for each RP. The algorithm to elect the DF is as follows: all PIM neighbors in a subnet advertise their unicast route to elect the RP and the router with the best route is elected. Guideline 33: All PEs MUST snoop the DF elections messages and determine the DF for AC/PW towards the DF (DF(RP)) MUST be added to the oiflist for each (*,G) whose RP(G) is RP. When DF(RP) changes., the oiflist must be updated accordingly, the oiflist must be updated accordingly - In Figure 2, there is one RP (call it RPA) behind Router 5. Based on DF election messages, PE routers populate PIM snooping states as follows: PE 1: {[PIM Neighbors: (Router 1,AC1), (Router 2,Router 3,PW1to2), (Router 4,PW1to3), (Router 5,PW1to4)], [DF(RPA): PW1to3], PE 2: {[PIM Neighbors: (Router 1,PW1to2), (Router 2,AC2), (Router 3,AC3), (Router 4,PW2to3), (Router 5,PW2to4)], [DF(RPA): PW2to3]}, PE 3: {[PIM Neighbors: (Router [Page 34] Internet Draft draft-serbest-l2vpn-vpls-mcast-03.txt July, 2005 1,PW1to3), (Router 2,Router 3,PW2to3), (Router 4,AC4), (Router 5,PW3to4)], [DF(RPA): AC5]}, PE 4: {[PIM Neighbors: (Router 1,PW1to4), (Router 2,Router 3,PW2to4), (Router 4,PW3to4), (Router 5,AC5)], [DF(RPA): PW3to4]}. 5.4.5.2. Guidelines for BIDIR-PIM Snooping The BIDIR-PIM snooping for Join and Prune messages is similar to PIM-SM and the following guidelines (some of which are repetitions from PIM-SM section) apply. Guideline 34: A PE MUST add a PW/AC to its pim_joins(*,G) list if it receives a (*,G) join message from the PW/AC. Guideline 35: BIDIR-PIM join messages MUST be flooded to all PEs in the VPLS instance. BIDIR-PIM join messages received on remote PEs MUST be forwarded only towards the router to which the Join is addressed. Guideline 36: BIDIR-PIM prune messages MUST be flooded in the VPLS instance. Guideline 37: If A PE does not receive a refresh join message from a PW/AC within its Holdtime, the PE MUST remove the PW/AC from its pim_joins(*,G) list. Guideline 38: A PE MUST remove a PW/AC from its pim_joins(*,G) list if it receives a (*,G) prune message from the PW/AC. A prune-delay timer SHOULD be implemented to support prune override. 5.4.5.3. BIDIR-PIM Join The BIDIR-PIM snooping mechanism for joining a multicast group works as follows: - As before, assume the RP for both G1 and G4 (RPA) is behind Router 4. Assume Router 2 wants to join the multicast group (*,G1). PE 2 sends the join message to the other PEs. All PEs update their states as follows: PE 1: { [PIM Neighbors: (Router 1,AC1), (Router 2,Router 3,PW1to2), (Router 4,PW1to3), (Router 5,PW1to4)], [DF(RPA): PW1to3], [pim_joins(*,G1): PW1to2]}, PE 2: {[PIM Neighbors: (Router 1,PW1to2), (Router 2,AC2), (Router 3,AC3), (Router 4,PW2to3), (Router 5,PW2to4)], [DF(RPA): PW2to3], [pim_joins(*,G1): AC2]}, PE 3: {[PIM Neighbors: (Router 1,PW1to3), (Router 2,Router 3,PW2to3), (Router 4,AC4), (Router 5,PW3to4)], [DF(RPA): AC4], [pim_joins(*,G1): PW2to3]}, PE 4: { [PIM Neighbors: (Router 1,PW1to4), (Router 2,Router 3,PW2to4), (Router [Page 35] Internet Draft draft-serbest-l2vpn-vpls-mcast-03.txt July, 2005 4,PW3to4), (Router 5,AC5)], [DF(RPA): PW3to4], [pim_joins(*,G1): PW2to4]}. - Next, assume Router 4 wants to join the multicast group (*,G1). All PEs update their states as follows: PE 1: {[PIM Neighbors: (Router 1,AC1), (Router 2,Router 3,PW1to2), (Router 4,PW1to3), (Router 5,PW1to4)], [DF(RPA): PW1to3], [pim_joins(*,G1): PW1to2, PW1to3]}, PE 2: {[PIM Neighbors: (Router 1,PW1to2), (Router 2,AC2), (Router 3,AC3), (Router 4,PW2to3), (Router 5,PW2to4)], [DF(RPA): PW2to3], [pim_joins(*,G1): AC2, PW2to3}, PE 3: {[PIM Neighbors: (Router 1,PW1to3), (Router 2,Router 3,PW2to3), (Router 4,AC4), (Router 5,PW3to4)], [DF(RPA): AC4], pim_joins(*,G1): PW2to3, AC4]}, PE 4: {[PIM Neighbors: (Router 1,PW1to4), (Router 2,Router 3,PW2to4), (Router 4,PW3to4), (Router 5,AC5)], [DF(RPA): PW3to4], [pim_joins(*,G1): PW2to4, PW3to4]}. - Then, assume Router 5 wants to join the multicast group (*,G4). Following the same procedures, all PEs update their states as follows: PE 1: {[PIM Neighbors: (Router 1,AC1), (Router 2,Router 3,PW1to2), (Router 4,PW1to3), (Router 5,PW1to4)], [DF(RPA): PW1to3], [pim_joins(*,G1): PW1to2, PW1to3], [pim_joins(*,G4): PW1to4]}, PE 2: {[PIM Neighbors: (Router 1,PW1to2), (Router 2,AC2), (Router 3,AC3), (Router 4,PW2to3), (Router 5,PW2to4)], [DF(RPA): PW2to3], [pim_joins(*,G1): AC2, PW2to3], [pim_joins(*,G4): PW2to4]}, PE 3: {[PIM Neighbors: (Router 1,PW1to3), (Router 2,Router 3,PW2to3), (Router 4,AC4), (Router 5,PW3to4)], [DF(RPA): AC4], pim_joins(*,G1): PW2to3, AC4], pim_joins(*,G4): PW3to4]}, PE 4: {[PIM Neighbors: (Router 1,PW1to4), (Router 2,Router 3,PW2to4), (Router 4,PW3to4), (Router 5,AC5)], [DF(RPA): PW3to4], [pim_joins(*,G1): PW2to4, PW3to4], [pim_joins(*,G4): AC5]}. 5.4.5.4. BIDIR-PIM Prune At this point, all PEs have necessary states to not send multicast traffic to sites with no members. One example of the BIDIR-PIM snooping mechanism for leaving a multicast group works as follows: - Assume Router 2 wants to leave the multicast group (*,G1) and sends a (*,G1) prune message. The prune message gets flooded in the VPLS instance. All PEs update their states as follows: PE 1: {[PIM Neighbors: (Router 1,AC1), (Router 2,Router 3,PW1to2), (Router 4,PW1to3), (Router 5,PW1to4)], [DF(RPA): PW1to3], [pim_joins(*,G1): PW1to3], [pim_joins(*,G4): PW1to4]}, PE 2: {[PIM Neighbors: (Router 1,PW1to2), (Router 2,AC2), (Router 3,AC3), (Router 4,PW2to3), (Router 5,PW2to4)], [DF(RPA): PW2to3], [Page 36] Internet Draft draft-serbest-l2vpn-vpls-mcast-03.txt July, 2005 [pim_joins(*,G1): PW2to3], [pim_joins(*,G4): PW2to4]}, PE 3: {[PIM Neighbors: (Router 1,PW1to3), (Router 2,Router 3,PW2to3), (Router 4,AC4), (Router 5,PW3to4)], [DF(RPA): AC4], [pim_joins(*,G1): AC1], [pim_joins(*,G4): PW3to4]}, PE 4: {[PIM Neighbors: (Router 1,PW1to4), (Router 2,Router 3,PW2to4), (Router 4,PW3to4), (Router 5,AC5)], [DF(RPA): PW3to4], [pim_joins(*,G1): PW3to4], [pim_joins(*,G4): AC5]}. 5.4.5.5. Failure Scenarios Once again, failures can be easily handled in BIDIR-PIM snooping, as it employs state-refresh technique. PEs in the VPLS instance will remove entry for non-refreshing routers from their states. 5.4.6. Multicast Source Directly Connected to the VPLS Instance If there is a source in the CE network that connects directly into the VPLS instance, then multicast traffic from that source MUST be sent to all PIM routers on the VPLS instance apart from the outgoing interface list for the corresponding snooping state. If there is already (S,G)/(*,G) snooping state that is formed on any PE, this will not happen per the current forwarding rules and guidelines. The (S,G)/(*,G) state may not send traffic towards all the routers. So, in order to determine if traffic needs to be flooded to all routers, a PE must be able to determine if the traffic came from a host on that LAN. There are three ways to address this problem: - The PE would have to do ARP snooping to determine if a source is directly connected. - Another option is to have configuration on all PEs to say there are CE sources that are directly connected to the VPLS instance and disallow snooping for the groups for which the source is going to send traffic. This way traffic from that source to those groups will always be flooded within the provider network. - A third option is to require that sources of CE multicast routers must appear behind a router. 5.5. VPLS Multicast on the Upstream PE An implementation MAY use native PIM Snooping procedures on the upstream PE to build multicast state as described in the previous sections. But snooping on PWs may overwhelm the upstream PEs. In this section, we propose an alternate approach for building multicast states on the upstream PE and thus avoid snooping on the PWs. As discussed earlier in the previous sections, VPLS Multicast for the various flavors of PIM requires only the downstream PE(s) and the upstream PE to build multicast state for a given multicast flow. Join/Prune messages need only be sent towards the upstream CE and it is wasteful to distribute and/or build multicast state on all PEs. [Page 37] Internet Draft draft-serbest-l2vpn-vpls-mcast-03.txt July, 2005 Unless otherwise noted, an upstream PE in this section refers to the PE to which the upstream RPF neighbor in the C-instance is connected. [VPLS-LDP] already uses LDP as the infrastructure to build PWs between the PEs and to exchange VPLS information. It suits VPLS multicast very well to leverage this existing infrastructure to send PIM multicast state information from the downstream PE to the upstream PE. While the procedures described here are extensible to other IP multicast protocols as well, we will define the procedures for PIM in this draft. The rules and procedures for this are described below. 5.5.1. Negotiating PIM Multicast capability in LDP When a PE is capable of exchanging PIM Multicast states using LDP, it signals this capability to its peers. When the two ends of a PW are both capable of exchanging PIM control information using LDP, the procedures described in the following sub-sections are employed. Otherwise, the following procedures are simply skipped. If a LDP-multicast capable PE determines that the other end of a PW is LDP-multicast capable, it SHOULD turn off native PIM snooping procedures on that PW. Otherwise, it MAY employ PIM snooping procedures to build multicast states. Packet format for exchanging PIM Multicast capability information in LDP will be defined in a future revision of the draft. One option is to define a bit in the LDP Hello Message to signal this capability. 5.5.2. Exchanging PIM Hellos We introduce a new PIM Hello TLV to carry the PIM Hellos received at the downstream PEs to all the other PEs using LDP. This TLV is sent in the LDP Address Message to all other PEs. The scope of this TLV is the VPLS instance specified in the FEC TLV in the LDP Address Message. When a new PIM Hello is received at the downstream PE, it is sent to all other PEs using the PIM Hello TLV. Subsequently, if a PIM Hello received from a C-router is the same as the previous Hello received from that C-router, it SHOULD NOT be sent in the PIM Hello TLV. If a PIM Hello received from a C-router is different from the previous received Hello from that C-router, the PE MUST send it in the PIM Hello TLV to all PEs. If the downstream PE ages out a PIM Hello, it MUST send a PIM Hello TLV with zero hold time to remove the Hello state on all other PEs. The downstream PE MUST also forward the PIM Hello on all PWs. When a PE receives a PIM Hello TLV for a C-router, it replaces the old PIM Hello information with the new one received in the TLV. The upstream PE never ages out a PIM Hello state. It MUST remove a PIM [Page 38] Internet Draft draft-serbest-l2vpn-vpls-mcast-03.txt July, 2005 Hello state only when it receives a PIM Hello TLV with zero hold-time OR when the PW is torn down. When a PE receives a new PIM Hello TLV, then all multicast states with that PW as the RPF interface MUST be refreshed with PIM Join/Prune TLVs. Packet format for the PIM Hello TLV will be defined in a future revision of this draft. 5.5.3. Exchanging PIM Join/Prune States We introduce a new PIM Join/Prune TLV to advertise C-Join/Prune messages received at a downstream PE to the upstream PE. This TLV is sent in the LDP Address Message to the upstream PE. The scope of this TLV is the VPLS instance specified in the FEC TLV in the LDP Address Message. When a downstream PE receives a new C-Join/Prune message for a multicast state, it MUST send the C-Join/Prune message in a PIM Join/Prune TLV to the upstream PE. If it receives a C-Join/Prune message information different than what was received before (e.g. newer hold-time, change in Join/Prune state, different RPF-neighbor field, etc), the PE MUST send the Join/Prune message in a PIM Join/Prune TLV to the new upstream PE. Note that if a C-Join is received for a new upstream PE, it may not imply that the state on the old upstream PE needs to be torn down. Different C-Routers may be sending C-Joins to different upstream RPF neighbors. If the downstream PE cannot determine who the upstream PE is, then it SHOULD save the Join/Prune state and send the PIM Join/Prune TLV when it is able to determine who the upstream PE is i.e. when it receives a PIM Hello TLV on a PW from the corresponding C-RPF-neighbor. The downstream PE MUST also forward the C-Join/Prune message. These packets will not be snooped at the upstream PE(s) and are intended only for the upstream C-router(s). Packet format for the PIM Join/Prune TLV will be defined in a future revision of this draft. 5.5.3.1. PIM Join Suppression Issues For VPLS Multicast to work, the C-routers MUST disable PIM Join suppression. However, it is our understanding that existing deployments from several vendors do not support the capability to disable PIM Join suppression. If that is so, then VPLS Multicast simply does not work if we multicast the C-Joins to all C-routers. Also, the provider has no control over the configuration on a C- router (to ensure that C-Join Suppression is disabled). [Page 39] Internet Draft draft-serbest-l2vpn-vpls-mcast-03.txt July, 2005 If the downstream PE determines that PIM Join suppression is active in a VPLS instance, then it MUST unicast-forward the C-Joins towards the RPF-neighbor field in the C-Join. This allows the C-Join to not be seen by other C-routers. Since we recommend that it unicast- forward the C-Join/Prune packets, it is important to ensure that the PIM control packets are received in order at the upstream C-router. To achieve this, the same ordering restriction that apply to broadcast and unknown frames apply to PIM control packets. 5.5.3.2. Resiliency against soft-state failures PIM is a soft-state protocol. So it is possible for packets to get dropped. Even though the Join/Prune exchange between the PEs is reliable; if certain packets are not received at the downstream PE, it can leave stale state on the upstream PE. Specifically, an RPF change on a downstream C-Router results in a C-Prune message to be sent to the old RPF-neighbor and a C-Join to be sent to the new RPF- neighbor. If the C-Prune message is not received at the downstream PE, then the downstream PE will not be able to forward that message to the upstream PE. This will result in stale state in the upstream PE. An implementation MUST implement one of the following two procedures to handle this. 5.5.3.2.1. Explicit Tracking of C-Joins at the downstream PE This method allows us to not require refreshes on PWs and yet achieve resiliency from soft-state failures. If this method is used, then the hold-time encoded in the PIM Join/Prune TLVs MUST be set to infinity. This is the recommended method since it eliminates refreshes on PWs. For each C-Join(S,G) received at the downstream PE on an AC, the downstream PE MUST keep the following state per Upstream C-Router: - The Set of Upstream C-Router Addresses o Per Upstream C-Router, a set of: . Downstream C-Router Address . Downstream Join Expiry Timer When a C-Join received at the downstream PE results in the set of Downstream C-Routers for a (C-Source, C-Group, C-RPF-Neighbor) to become non-empty, then a PIM Join TLV MUST be sent to the Upstream PE. If the set of Downstream C-Routers for a (C-Source, C-Group, C-RPF- Neighbor) becomes empty, then a PIM Prune TLV MUST be sent to the corresponding Upstream PE. This may happen as a result of a received C-Prune message or as a result of the Downstream Join Timer Expiry. [Page 40] Internet Draft draft-serbest-l2vpn-vpls-mcast-03.txt July, 2005 5.5.3.2.2. Refreshing PIM Join TLVs on the PWs If this method is employed, the downstream PE MUST forward received C-Joins in the form of PIM Join TLVs on the PWs at periodic intervals. The refresh interval across the PWs should be configurable in multiples of the C-Join refresh interval. If this refresh multiple is N, then every Nth C-Join refresh for a given multicast state MUST also be sent as a PIM Join TLV to the upstream PE. The HoldTime field in the PIM Join TLV MUST be set to ((N * HoldTime in C-Join) + 20) seconds. 5.5.3.3. PIM-BIDIR Considerations Unlike other PIM variants, in PIM-BIDIR, a traffic source need not be behind the RPF-neighbor. Traffic can come from any AC/PW and it MUST be forwarded by the switches. Following are the deviations from the procedures defined earlier to handle PIM-BIDIR. PIM-BIDIR Join/Prune TLVs MUST be forwarded to all PEs instead of just the upstream PE towards the RP. PIM BIDIR Join/Prune Packets MUST also be multicast-forwarded as is on all PWs. 5.6. Data Forwarding Rules The final list of outgoing interfaces for a given (S,G) or (*,G) is computed by combining the IGMP and PIM state summarization macros. OifList(*,G) = igmp_include(*,G) (+) pim_oiflist(*,G) Oiflist(S,G) = igmp_include(*,G) (-) igmp_exclude(S,G) (+) Igmp_include(S,G) (+) pim_oiflist(S,G) If PIM Snooping is active for a given (*,G) or (S,G), then the PE also tracks the upstream AC/PW as the RPF interface. Data traffic MUST be forwarded ONLY IF traffic arrives on the RPF interface. If data traffic arrives on any other interface, then the following rules apply: - If the traffic arrives on an AC and the PE determines that the traffic is coming from a directly connected source, then the rules described in Section 5.4.6. apply. - Otherwise, it could be a PIM ASSERT scenario. Then the rules described in Section 5.4.3.8. apply. In the presence of only IGMP Snooping state, there is no RPF interface that can be remembered. In such a scenario, traffic should simply be forwarded to the Oiflist after performing source interface pruning. 6. Security Considerations [Page 41] Internet Draft draft-serbest-l2vpn-vpls-mcast-03.txt July, 2005 Security considerations provided in VPLS solution documents (i.e., [VPLS-LDP] and [VPLS-BGP) apply to this document as well. 7. References 7.1. Normative References 7.2. Informative References [VPLS-LDP] Lasserre, M, et al. "Virtual Private LAN Services over MPLS", work in progress [VPLSD-BGP] Kompella, K, et al. "Virtual Private LAN Service", work in progress [L2VPN-FR] Andersson, L, et al. "L2VPN Framework", work in progress [PMP-RSVP-TE] Aggarwal, R, et al. "Extensions to RSVP-TE for Point to Multipoint TE LSPs", work in progress [RFC1112] Deering, S., "Host Extensions for IP Multicasting", RFC 1112, August 1989. [RFC2236] Fenner, W., "Internet Group Management Protocol, Version 2", RFC 2236, November 1997. [RFC3376] Cain, B., et al. "Internet Group Management Protocol, Version 3", RFC 3376, October 2002. [MAGMA-SNOOP] Christensen, M., et al. "Considerations for IGMP and MLD Snooping Switches", work in progress [PIM-DM] Deering, S., et al. "Protocol Independent Multicast Version 2 Dense Mode Specification", draft-ietf- pim-v2-dm-03.txt, June 1999. [RFC2362] Estrin, D, et al. "Protocol Independent Multicast- Sparse Mode (PIM-SM): Protocol Specification", RFC 2362, June 1998. [PIM-SSM] Holbrook, H., et al. "Source-Specific Multicast for IP", work in progress [BIDIR-PIM] Handley, M., et al. "Bi-directional Protocol Independent Multicast (BIDIR-PIM)", work in progress Authors' Addresses Yetik Serbest SBC Labs 9505 Arboretum Blvd. Austin, TX 78759 Yetik_serbest@labs.sbc.com Ray Qiu Alcatel North America 701 East Middlefield Rd. Mountain View, CA 94043 Ray.Qiu@alcatel.com Venu Hemige [Page 42] Internet Draft draft-serbest-l2vpn-vpls-mcast-03.txt July, 2005 Alcatel North America 701 East Middlefield Rd. Mountain View, CA 94043 Venu.hemige@alcatel.com Rob Nath Riverstone Networks 5200 Great America Parkway Santa Clara, CA 95054 Rnath@riverstonenet.com Suresh Boddapati Alcatel North America 701 East Middlefield Rd. Mountain View, CA 94043 Suresh.boddapati@alcatel.com Sunil Khandekar Alcatel North America 701 East Middlefield Rd. Mountain View, CA 94043 Sunil.khandekar@alcatel.com Vach Kompella Alcatel North America 701 East Middlefield Rd. Mountain View, CA 94043 Vach.kompella@alcatel.com Marc Lasserre Riverstone Networks Marc@riverstonenet.com Himanshu Shah Ciena hshah@ciena.com Intellectual Property Statement The IETF takes no position regarding the validity or scope of any Intellectual Property Rights or other rights that might be claimed to pertain to the implementation or use of the technology described in this document or the extent to which any license under such rights might or might not be available; nor does it represent that it has made any independent effort to identify any such rights. Information on the procedures with respect to rights in RFC documents can be found in BCP 78 and BCP 79. Copies of IPR disclosures made to the IETF Secretariat and any assurances of licenses to be made available, or the result of an attempt made to obtain a general license or permission for the use of [Page 43] Internet Draft draft-serbest-l2vpn-vpls-mcast-03.txt July, 2005 such proprietary rights by implementers or users of this specification can be obtained from the IETF on-line IPR repository at http://www.ietf.org/ipr. The IETF invites any interested party to bring to its attention any copyrights, patents or patent applications, or other proprietary rights that may cover technology that may be required to implement this standard. Please address the information to the IETF at ietf- ipr@ietf.org. Full copyright statement Copyright (C) The Internet Society (2005). This document is subject to the rights, licenses and restrictions contained in BCP 78, and except as set forth therein, the authors retain all their rights. This document and the information contained herein are provided on an "AS IS" basis and THE CONTRIBUTOR, THE ORGANIZATION HE/SHE REPRESENTS OR IS SPONSORED BY (IF ANY), THE INTERNET SOCIETY AND THE INTERNET ENGINEERING TASK FORCE DISCLAIM ALL WARRANTIES, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO ANY WARRANTY THAT THE USE OF THE INFORMATION HEREIN WILL NOT INFRINGE ANY RIGHTS OR ANY IMPLIED WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. [Page 44]