Internet Engineering Task Force (IETF) T. Morin, Ed. Request for Comments: 6517 France Telecom - Orange Category: Informational B. Niven-Jenkins, Ed. ISSN: 2070-1721 BT Y. Kamite NTT Communications R. Zhang Alcatel-Lucent N. Leymann Deutsche Telekom N. Bitar Verizon February 2012 Mandatory Features in a Layer 3 Multicast BGP/MPLS VPN Solution
AbstractMore that one set of mechanisms to support multicast in a layer 3 BGP/MPLS VPN has been defined. These are presented in the documents that define them as optional building blocks. To enable interoperability between implementations, this document defines a subset of features that is considered mandatory for a multicast BGP/MPLS VPN implementation. This will help implementers and deployers understand which L3VPN multicast requirements are best satisfied by each option. Status of This Memo This document is not an Internet Standards Track specification; it is published for informational purposes. This document is a product of the Internet Engineering Task Force (IETF). It represents the consensus of the IETF community. It has received public review and has been approved for publication by the Internet Engineering Steering Group (IESG). Not all documents approved by the IESG are a candidate for any level of Internet Standard; see Section 2 of RFC 5741. Information about the current status of this document, any errata, and how to provide feedback on it may be obtained at http://www.rfc-editor.org/info/rfc6517.
Copyright Notice Copyright (c) 2012 IETF Trust and the persons identified as the document authors. All rights reserved. This document is subject to BCP 78 and the IETF Trust's Legal Provisions Relating to IETF Documents (http://trustee.ietf.org/license-info) in effect on the date of publication of this document. Please review these documents carefully, as they describe your rights and restrictions with respect to this document. Code Components extracted from this document must include Simplified BSD License text as described in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the Simplified BSD License. 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 3 2. Terminology . . . . . . . . . . . . . . . . . . . . . . . . . 3 3. Examining Alternative Mechanisms for MVPN Functions . . . . . 4 3.1. MVPN Auto-Discovery . . . . . . . . . . . . . . . . . . . 4 3.2. S-PMSI Signaling . . . . . . . . . . . . . . . . . . . . . 5 3.3. PE-PE Exchange of C-Multicast Routing . . . . . . . . . . 7 3.3.1. PE-PE C-Multicast Routing Scalability . . . . . . . . 7 3.3.2. PE-CE Multicast Routing Exchange Scalability . . . . . 10 3.3.3. Scalability of P Routers . . . . . . . . . . . . . . . 10 3.3.4. Impact of C-Multicast Routing on Inter-AS Deployments 10 3.3.5. Security and Robustness . . . . . . . . . . . . . . . 11 3.3.6. C-Multicast VPN Join Latency . . . . . . . . . . . . . 12 3.3.7. Conclusion on C-Multicast Routing . . . . . . . . . . 14 3.4. Encapsulation Techniques for P-Multicast Trees . . . . . . 15 3.5. Inter-AS Deployments Options . . . . . . . . . . . . . . . 16 3.6. BIDIR-PIM Support . . . . . . . . . . . . . . . . . . . . 19 4. Co-Located RPs . . . . . . . . . . . . . . . . . . . . . . . . 20 5. Avoiding Duplicates . . . . . . . . . . . . . . . . . . . . . 21 6. Existing Deployments . . . . . . . . . . . . . . . . . . . . . 21 7. Summary of Recommendations . . . . . . . . . . . . . . . . . . 22 8. Security Considerations . . . . . . . . . . . . . . . . . . . 22 9. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . 23 10. References . . . . . . . . . . . . . . . . . . . . . . . . . . 23 10.1. Normative References . . . . . . . . . . . . . . . . . . . 23 10.2. Informative References . . . . . . . . . . . . . . . . . . 23 Appendix A. Scalability of C-Multicast Routing Processing Load . 25 A.1. Scalability with an Increased Number of PEs . . . . . . . 26 A.1.1. SSM Scalability . . . . . . . . . . . . . . . . . . . 26 A.1.2. ASM Scalability . . . . . . . . . . . . . . . . . . . 35 A.2. Cost of PEs Leaving and Joining . . . . . . . . . . . . . 37 Appendix B. Switching to S-PMSI . . . . . . . . . . . . . . . . . 40
RFC6513] include multiple alternative mechanisms for some of the required building blocks of the solution. However, they do not identify which of these mechanisms are mandatory to implement in order to ensure interoperability. Not defining a set of mandatory-to-implement mechanisms leads to a situation where implementations may support different subsets of the available optional mechanisms that do not interoperate, which is a problem for the numerous operators having multi-vendor backbones. The aim of this document is to leverage the already expressed requirements [RFC4834] and study the properties of each approach to identify mechanisms that are good candidates for being part of a core set of mandatory mechanisms that can be used to provide a base for interoperable solutions. This document goes through the different building blocks of the solution and concludes which mechanisms an implementation is required to implement. Section 7 summarizes these requirements. Considering the history of the multicast VPN proposals and implementations, it is also useful to discuss how existing deployments of early implementations [RFC6037] [MVPN] can be accommodated and provide suggestions in this respect. RFC6513] and [RFC4834]. As a reminder, in Section 3.1 of [RFC6513], the "C-" and "P-" notations are used to disambiguate between the provider scope and the scope of a specific VPN customer; for instance, "C-PIM" designates a PIM protocol instance in a VPN or VRF, while "P-PIM" would designate the instance of PIM eventually deployed by the provider across its core between P and PE routers. Other acronyms used in this document include the following: o LSP: Label Switched Path o P2MP: Point to Multipoint o MP2MP: Multipoint to Multipoint o GRE: Generic Routing Encapsulation o mLDP: Multicast LDP
o I-PMSI: Inclusive Provider Multiservice Interface o MI-PMSI: Multidirectional Inclusive Provider Multiservice Interface o S-PMSI: Selective Provider Multiservice Interface o SSM: Source-Specific Multicast o ASM: Any-Source Multicast o PIM-SM: PIM Sparse Mode o PIM-SSM: PIM Sparse Mode in SSM Mode o BIDIR-PIM: Bidirectional PIM o AS: Autonomous System o ASBR: Autonomous System Border Router o VRF: VPN Routing and Forwarding o PE: Provider Edge o CE: Customer Edge o RPA: Rendezvous Point Address o RPL: Rendezvous Point Link Additionally, 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 [RFC2119]. RFC6513] proposes two different mechanisms for Multicast VPN (MVPN) auto-discovery: 1. BGP-based auto-discovery 2. "PIM/shared P-tunnel": discovery done through the exchange of PIM Hellos by C-PIM instances, across an MI-PMSI implemented with one shared P-tunnel per VPN (using ASM or MP2MP LDP)
Both solutions address Section 5.2.10 of [RFC4834], which states that "the operation of a multicast VPN solution SHALL be as light as possible, and providing automatic configuration and discovery SHOULD be a priority when designing a multicast VPN solution. Particularly, the operational burden of setting up multicast on a PE or for a VR/ VRF SHOULD be as low as possible". The key consideration is that PIM-based discovery is only applicable to deployments using a shared P-tunnel to instantiate an MI-PMSI (it is not applicable if only P2P, PIM-SSM, and P2MP LDP/RSVP-TE P-tunnels are used, because contrary to ASM and MP2MP LDP, building these types of P-tunnels cannot happen before the auto-discovery has been done). In contrast, the BGP-based auto-discovery does not place any constraint on the type of P-tunnel that would have to be used. BGP-based auto-discovery is independent of the type of P-tunnel used, thus satisfying the requirement in Section 126.96.36.199 of [RFC4834] that "a multicast VPN solution SHOULD be designed so that control and forwarding planes are not interdependent". Additionally, it is to be noted that a number of service providers have chosen to use SSM-based P-tunnels for the default multicast distribution trees within their current deployments, therefore relying already on some BGP-based auto-discovery. Moreover, when shared P-tunnels are used, the use of BGP auto- discovery would allow inconsistencies in the addresses/identifiers used for the shared P-tunnel to be detected (e.g., the same shared P-tunnel identifier being used for different VPNs with distinct BGP route targets). This is particularly attractive in the context of inter-AS VPNs where the impact of any misconfiguration could be magnified and where a single service provider may not operate all the ASes. Note that this technique to detect some misconfiguration cases may not be usable during a transition period from a shared-P-tunnel auto-discovery to a BGP-based auto-discovery. Thus, the recommendation is that implementation of the BGP-based auto-discovery is mandated and should be supported by all MVPN implementations. RFC6513] proposes two mechanisms for signaling that multicast flows will be switched to a Selective PMSI (S-PMSI): 1. a UDP-based TLV protocol specifically for S-PMSI signaling (described in Section 7.4.2)
2. a BGP-based mechanism for S-PMSI signaling (described in Section 7.4.1) Section 5.2.10 of [RFC4834] states that "as far as possible, the design of a solution SHOULD carefully consider the number of protocols within the core network: if any additional protocols are introduced compared with the unicast VPN service, the balance between their advantage and operational burden SHOULD be examined thoroughly". The UDP-based mechanism would be an additional protocol in the MVPN stack, which isn't the case for the BGP-based S-PMSI switching signaling, since (a) BGP is identified as a requirement for auto-discovery and (b) the BGP-based S-PMSI switching signaling procedures are very similar to the auto-discovery procedures. Furthermore, the UDP-based S-PMSI switching signaling mechanism requires an MI-PMSI, while the BGP-based protocol does not. In practice, this mean that with the UDP-based protocol, a PE will have to join to all P-tunnels of all PEs in an MVPN, while in the alternative where BGP-based S-PMSI switching signaling is used, it could delay joining a P-tunnel rooted at a PE until traffic from that PE is needed, thus reducing the amount of state maintained on P routers. S-PMSI switching signaling approaches can also be compared in an inter-AS context (see Section 3.5). The proposed BGP-based approach for S-PMSI switching signaling provides a good fit with both the segmented and non-segmented inter-AS approaches (see Section 3.5). By contrast, while the UDP-based approach for S-PMSI switching signaling appears to be usable with segmented inter-AS tunnels, key advantages of the segmented approach are lost: o ASes are no longer independent in their ability to choose when S-PMSIs tunnels will be triggered in their AS (and thus control the amount of state created on their P routers). o ASes are no longer independent in their ability to choose the tunneling technique for the P-tunnels used for an S-PMSI. o In an inter-AS option B context, an isolation of ASes is obtained as PEs in one AS don't have (direct) exchange of routing information with PEs of other ASes. This property is not preserved if UDP-based S-PMSI switching signaling is used. By contrast, BGP-based C-multicast switching signaling does preserve this property. Given all the above, it is the recommendation of the authors that BGP is the preferred solution for S-PMSI switching signaling and should be supported by all implementations.
If nothing prevents a fast-paced creation of S-PMSI, then S-PMSI switching signaling with BGP would possibly impact the route reflectors (RRs) used for MVPN routes. However, such a fast-paced behavior would have an impact on P and PE routers resulting from S-PMSI tunnels signaling, which will be the same independent of the S-PMSI signaling approach that is used and which is certainly best to avoid by setting up proper mechanisms. The UDP-based S-PMSI switching signaling protocol can also be considered, as an option, given that this protocol has been in deployment for some time. Implementations supporting both protocols would be expected to provide a per-VRF (VPN Routing and Forwarding) configuration knob to allow an implementation to use the UDP-based TLV protocol for S-PMSI switching signaling for specific VRFs in order to support the co-existence of both protocols (for example, during migration scenarios). Apart from such migration-facilitating mechanisms, the authors specifically do not recommend extending the already proposed UDP-based TLV protocol to new types of P-tunnels. RFC6513] proposes multiple mechanisms for PE-PE exchange of customer multicast routing information (C-multicast routing): 1. Full per-MVPN PIM peering across an MI-PMSI (described in Section 188.8.131.52) 2. Lightweight PIM peering across an MI-PMSI (described in Section 184.108.40.206) 3. The unicasting of PIM C-Join/Prune messages (described in Section 220.127.116.11) 4. The use of BGP for carrying C-multicast routing (described in Section 3.4.2) Section 4.2.4 of [RFC4834] recommends that "a multicast VPN solution SHOULD support several hundreds of PEs per multicast VPN, and MAY usefully scale up to thousands" and Section 4.2.5 states that "a solution SHOULD scale up to thousands of PEs having multicast service enabled".
Scalability with an increased number of VPNs per PE, or with an increased amount of multicast state per VPN, are also important but are not focused on in this section since we didn't identify differences between the various approaches for these matters: all others things equal, the load on PE due to C-multicast routing increases roughly linearly with the number of VPNs per PE and with the amount of multicast state per VPN. This section presents conclusions related to PE-PE C-multicast routing scalability. Appendix A provides more detailed explanations on the differences in ways PIM-based approaches and the BGP-based approach handle the C-multicast routing load, along with quantified evaluations of the amount of state and messages with the different approaches. Many points made in this section are detailed in Appendix A.1. At high scales of multicast deployment, the first and third mechanisms require the PEs to maintain a large number of PIM adjacencies with other PEs of the same multicast VPN (which implies the regular exchange of PIM Hellos with each other) and to periodically refresh C-Join/Prune states, resulting in an increased processing cost when the number of PEs increases (as detailed in Appendix A.1). The second approach is less subject to this, and the fourth approach is not subject to this. The third mechanism would reduce the amount of C-Join/Prune processing for a given multicast flow for PEs that are not the upstream neighbor for this flow but would require "explicit tracking" state to be maintained by the upstream PE. It also isn't compatible with the "Join suppression" mechanism. A possible way to reduce the amount of signaling with this approach would be the use of a PIM refresh-reduction mechanism. Such a mechanism, based on TCP, is being specified by the PIM IETF Working Group ([PIM-PORT]); its use in a multicast VPN context is not described in [RFC6513], but it is expected that this approach will provide a scalability similar to the BGP-based approach without RRs. The second mechanism would operate in a similar manner to full per- MVPN PIM peering except that PIM Hello messages are not transmitted and PIM C-Join/Prune refresh-reduction would be used, thereby improving scalability, but this approach has yet to be fully described. In any case, it seems that it only improves one thing among the things that will impact scalability when the number of PEs increases. The first and second mechanisms can leverage the "Join suppression" behavior and thus improve the processing burden of an upstream PE, sparing the processing of a Join refresh message for each remote PE
joined to a multicast stream. This improvement requires all PEs of a multicast VPN to process all PIM Join and Prune messages sent by any other PE participating in the same multicast VPN whether they are the upstream PE or not. The fourth mechanism (the use of BGP for carrying C-multicast routing) would have a comparable drawback of requiring all PEs to process a BGP C-multicast route only interesting a specific upstream PE. For this reason, Section 16 of [RFC6514] recommends the use of the Route Target constrained BGP distribution [RFC4684] mechanisms, which eliminate this drawback by having only the interested upstream PE receive a BGP C-multicast route. Specifically, when Route Target constrained BGP distribution is used, the fourth mechanism reduces the total amount of the C-multicast routing processing load put on the PEs by avoiding any processing of customer multicast routing information on the "unrelated" PEs that are neither the joining PE nor the upstream PE. Moreover, the fourth mechanism further reduces the total amount of message processing load by avoiding the use of periodic refreshes and by inheriting BGP features that are expected to improve scalability (for instance, providing a means to offload some of the processing burden associated with customer multicast routing onto one or many BGP route reflectors). The advantages of the fourth mechanism come at a cost of maintaining an amount of state linear with the number of PEs joined to a stream. However, the use of route reflectors allows this cost to be spread among multiple route reflectors, thus eliminating the need for a single route reflector to maintain all this state. However, the fourth mechanism is specific in that it offers the possibility of offloading customer multicast routing processing onto one or more BGP route reflector(s). When this is used, there is a drawback of increasing the processing load placed on the route reflector infrastructure. In the higher scale scenarios, it may be required to adapt the route reflector infrastructure to the MVPN routing load by using, for example: o a separation of resources for unicast and multicast VPN routing: using dedicated MVPN route reflector(s) (or using dedicated MVPN BGP sessions or dedicated MVPN BGP instances), and o the deployment of additional route reflector resources, for example, increasing the processing resources on existing route reflectors or deployment of additional route reflectors. The most straightforward approach is to consider the introduction of route reflectors dedicated to the MVPN service and dimension them
according to the need of that service (but doing so is not required and is left as an operator engineering decision). Section 18.104.22.168 of [RFC4834]). 5.2.6 and 5.2.8 of [RFC4834]. In an inter-AS option B context, an isolation of ASes is obtained as PEs in one AS don't have (direct) exchange of routing information with PEs of other ASes. This property is not preserved if PIM-based
PE-PE C-multicast routing is used. By contrast, the fourth option (BGP-based C-multicast routing) does preserve this property. Additionally, the authors note that the proposed BGP-based approach for C-multicast routing provides a good fit with both the segmented and non-segmented inter-AS approaches. By contrast, though the PIM- based C-multicast routing is usable with segmented inter-AS tunnels, the inter-AS scalability advantage of the approach is lost, since PEs in an AS will see the C-multicast routing activity of all other PEs of all other ASes. RFC5796] and, in any case, would require significant additional operations for the provider to be usable in a multicast VPN context. The robustness of the infrastructure, especially the existing infrastructure providing unicast VPN connectivity, is key. The C-multicast routing function, especially under load, will compete with the unicast routing infrastructure. With the PIM-based approaches, the unicast and multicast VPN routing functions are expected to only compete in the PE, for control plane processing resources. In the case of the BGP-based approach, they will compete on the PE for processing resources and in the route reflectors (supposing they are used for MVPN routing). In both cases, mechanisms will be required to arbitrate resources (e.g., processing priorities). In the case of PIM-based procedures, this arbitration occurs between the different control plane routing instances in the PE. In the case of the BGP-based approach, this is likely to require using distinct BGP sessions for multicast and unicast (e.g., through the use of dedicated MVPN BGP route reflectors or the use of a distinct session with an existing route reflector). Multicast routing is dynamic by nature, and multicast VPN routing has to follow the VPN customers' multicast routing events. The different approaches can be compared on how they are expected to behave in scenarios where multicast routing in the VPNs is subject to an intense activity. Scalability of each approach under such a load is detailed in Appendix A.2, and the fourth approach (BGP-based) used in conjunction with the Route Target Constraint mechanisms [RFC4684] is the only one having a cost for join/leave operations independent of the number of PEs in the VPN (with one exception detailed in
Appendix A.2) and state maintenance not concentrated on the upstream PE. On the other hand, while the BGP-based approach is likely to suffer a slowdown under a load that is greater than the available processing resources (because of possibly congested TCP sockets), the PIM-based approaches would react to such a load by dropping messages, with failure-recovery obtained through message refreshes. Thus, the BGP- based approach could result in a degradation of join/leave latency performance typically spread evenly across all multicast streams being joined in that period, while the PIM-based approach could result in increased join/leave latency, for some random streams, by a multiple of the time between refreshes (e.g., tens of seconds), and possibly in some states, the adjacency may timeout, resulting in disruption of multicast streams. The behavior of the PIM-based approach under such a load is also harder to predict, given that the performance of the "Join suppression" mechanism (an important mechanism for this approach to scale) will itself be impeded by delays in Join processing. For these reasons, the BGP-based approach would be able to provide a smoother degradation and more predictable behavior under a highly dynamic load. In fact, both an "evenly spread degradation" and an "unevenly spread larger degradation" can be problematic, and what seems important is the ability for the VPN backbone operator to (a) limit the amount of multicast routing activity that can be triggered by a multicast VPN customer and (b) provide the best possible independence between distinct VPNs. It seems that both of these can be addressed through local implementation improvements and that both the BGP-based and PIM-based approaches could be engineered to provide (a) and (b). It can be noted though that the BGP approach proposes ways to dampen C-multicast route withdrawals and/or advertisements and thus already describes a way to provide (a), while nothing comparable has yet been described for the PIM-based approaches (even though it doesn't appear difficult). The PIM-based approaches rely on a per-VPN data plane to carry the MVPN control plane and thus may benefit from this first level of separation to solve (b). Section 5.1.3 of [RFC4834] states that the "group join delay [...] is also considered one important QoS parameter. It is thus RECOMMENDED that a multicast VPN solution be designed appropriately in this regard". In a multicast VPN context, the "group join delay" of interest is the time between a CE sending a PIM Join to its PE and
the first packet of the corresponding multicast stream being received by the CE. It is to be noted that the C-multicast routing procedures will only impact the group join latency of a said multicast stream for the first receiver that is located across the provider backbone from the multicast source-connected PE (or the first <n> receivers in the specific case where a specific Upstream Multicast Hop selection algorithm is used, which allows <n> distinct PEs to be selected as the Upstream Multicast Hop by distinct downstream PEs). The different approaches proposed seem to have different characteristics in how they are expected to impact join latency: o The PIM-based approaches minimize the number of control plane processing hops between a new receiver-connected PE and the source-connected PE and, being datagram-based, introduce minimal delay, thereby possibly having a join latency as good as possible depending on implementation efficiency. o Under degraded conditions (packet loss, congestion, or high control plane load) the PIM-based approach may impact the latency for a given multicast stream in an all-or-nothing manner: if a C-multicast routing PIM Join packet is lost, latency can reach a high time (a multiple of the periodicity of PIM Join refreshes). o The BGP-based approach uses TCP exchanges, which may introduce an additional delay depending on BGP and TCP implementation but which would typically result, under degraded conditions (such packet loss, congestion, or high control plane load), in a comparably lower increase of latency spread more evenly across the streams. o As shown in Appendix A, the BGP-based approach is particular in that it removes load from all the PEs (without putting this load on the upstream PE for a stream); this improvement of background load can bring improved performance when a PE acts as the upstream PE for a stream and thus benefits join latency. This qualitative comparison of approaches shows that the BGP-based approach is designed for a smoother degradation of latency under degraded conditions such as packet loss, congestion, or high control plane load. On the other hand, the PIM-based approaches seem to structurally be able to reach the shorter "best-case" group join latency (especially compared to deployment of the BGP-based approach where route reflectors are used). Doing a quantitative comparison of latencies is not possible without referring to specific implementations and benchmarking procedures and
would possibly expose different conclusions, especially for best-case group join latency for which performance is expected to vary with PIM and BGP implementations. We can also note that improving a BGP implementation for reduced latency of route processing would not only benefit multicast VPN group join latency but the whole BGP-based routing, which means that the need for good BGP/RR performance is not specific to multicast VPN routing. Last, C-multicast join latency will be impacted by the overall load put on the control plane, and the scalability of the C-multicast routing approach is thus to be taken into account. As explained in Section 3.3.1 and Appendix A, the BGP-based approach will provide the best scalability with an increased number of PEs per VPN, thereby benefiting group join latency in such higher-scale scenarios. RFC6037]; therefore, it is deployed and operating in MVPNs today. The fourth approach may or may not end up being preferred for a said deployment, but because the first approach has been in deployment for some time, the support for this mechanism will in any case be helpful to facilitate an eventual migration from a deployment using mechanism close to the first approach. Consequently, at the present time, implementations are recommended to support both the fourth (BGP-based) and first (full per-MVPN PIM peering) mechanisms. Further experience on deployments of the fourth approach is needed before some best practices can be defined. In the meantime, this recommendation would enable a service provider to choose between the first and the fourth mechanisms, without this choice being constrained by vendor implementation choices. A service provider can also take into account the peculiarities of its own deployment context by pondering the weight of the different factors into account.
Section 22.214.171.124 of [RFC4834] states, "In a multicast VPN solution extending a unicast layer 3 PPVPN solution, consistency in the tunneling technology has to be favored: such a solution SHOULD allow the use of the same tunneling technology for multicast as for unicast. Deployment consistency, ease of operation, and potential migrations are the main motivations behind this requirement". Current unicast VPN deployments use a variety of LDP, RSVP-TE, and GRE/IP-Multicast for encapsulating customer packets for transport across the provider core of VPN services. In order to allow the same encapsulations to be used for unicast and multicast VPN traffic, it is recommended that multicast VPN standards should recommend that implementations support multicast VPNs and all the P2MP variants of the encapsulations and signaling protocols that they support for unicast and for which some multipoint extension is defined, such as mLDP, P2MP RSVP-TE, and GRE/IP-multicast. All three of the above encapsulation techniques support the building of P2MP multicast P-tunnels. In addition, mLDP and GRE/ IP-ASM-Multicast implementations may also support the building of MP2MP multicast P-tunnels. The use of MP2MP P-tunnels may provide some scaling benefits to the service provider as only a single MP2MP P-tunnel need be deployed per VPN, thus reducing by an order of magnitude the amount of multicast state that needs to be maintained by P routers. This gain in state is at the expense of bandwidth optimization, since sites that do not have multicast receivers for multicast streams sourced behind a said PE group will still receive packets of such streams, leading to non-optimal bandwidth utilization across the VPN core. One thing to consider is that the use of MP2MP multicast P-tunnel will require additional configuration to define the same P-tunnel identifier or multicast ASM group address in all PEs (it has been noted that some auto-configuration could be possible
for MP2MP P-tunnels, but this is not currently supported by the auto- discovery procedures). (It has been noted that C-multicast routing schemes not covered in [RFC6513] could expose different advantages of MP2MP multicast P-tunnels; this is out of the scope of this document.) MVPN services can also be supported over a unicast VPN core through the use of ingress PE replication whereby the ingress PE replicates any multicast traffic over the P2P tunnels used to support unicast traffic. While this option does not require the service provider to modify their existing P routers (in terms of protocol support) and does not require maintaining multicast-specific state on the P routers in order for the service provider to be able deploy a multicast VPN service, the use of ingress PE replication obviously leads to non-optimal bandwidth utilization, and it is therefore unlikely to be the long-term solution chosen by service providers. However, ingress PE replication may be useful during some migration scenarios or where a service provider considers the level of multicast traffic on their network to be too low to justify deploying multicast-specific support within their VPN core. All proposed approaches for control plane and data plane can be used to provide aggregation amongst multicast groups within a VPN and amongst different multicast VPNs, and potentially reduce the amount of state to be maintained by P routers. However, the latter (the aggregation amongst different multicast VPNs) will require support for upstream-assigned labels on the PEs. Support for upstream- assigned labels may require changes to the data plane processing of the PEs, and this should be taken into consideration by service providers considering the use of aggregate PMSI tunnels for the specific platforms that the service provider has deployed.
The first scenario can be considered the "simplest" because the network is wholly managed by a single service provider under a single strategy and is therefore likely to use a consistent set of technologies across each AS. The second scenario may be more complex than the first because the strategy and technology choices made for each AS may have been different due to their differing histories, and the service provider may not have unified (or may be unwilling to unify) the strategy and technology choices for each AS. The third scenario is the most complex because in addition to the complexity of the second scenario, the ASes are managed by different service providers and therefore may be subject to a different trust model than the other scenarios. Section 5.2.6 of [RFC4834] states that "a solution MUST support inter-AS multicast VPNs, and SHOULD support inter-provider multicast VPNs", "considerations about co-existence with unicast inter-AS VPN Options A, B, and C (as described in Section 10 of [RFC4364]) are strongly encouraged", and "a multicast VPN solution SHOULD provide inter-AS mechanisms requiring the least possible coordination between providers, and keep the need for detailed knowledge of providers' networks to a minimum -- all this being in comparison with corresponding unicast VPN options". Section 8 of [RFC6513] addresses these requirements by proposing two approaches for MVPN inter-AS deployments: 1. Non-segmented inter-AS tunnels where the multicast tunnels are end-to-end across ASes, so even though the PEs belonging to a given MVPN may be in different ASes, the ASBRs play no special role and function merely as P routers (described in Section 8.1). 2. Segmented inter-AS tunnels where each AS constructs its own separate multicast tunnels that are then 'stitched' together by the ASBRs (described in Section 8.2). (Note that an inter-AS deployment can alternatively rely on Option A -- so-called "back-to-back" VRFs -- that option is not considered in this section given that it can be used without any inter-AS-specific mechanism.) Section 5.2.6 of [RFC4834] also states, "Within each service provider, the service provider SHOULD be able on its own to pick the most appropriate tunneling mechanism to carry (multicast) traffic among PEs (just like what is done today for unicast)". The segmented approach is the only one capable of meeting this requirement.
The segmented inter-AS solution would appear to offer the largest degree of deployment flexibility to operators. However, the non- segmented inter-AS solution can simplify deployment in a restricted number of scenarios. [RFC6037] only supports the non-segmented inter-AS solution; therefore, the non-segmented inter-AS solution is likely to be useful to some operators for backward compatibility and during migration from [RFC6037] to [RFC6513]. The following is a comparison matrix between the "segmented inter-AS P-tunnels" and "non-segmented inter-AS P-tunnels" approaches: o Scalability for I-PMSIs: The "segmented inter-AS P-tunnels" approach is more scalable, because of the ability of an ASBR to aggregate multiple intra-AS P-tunnels used for I-PMSI within its own AS into one inter-AS P-tunnel to be used by other ASes. Note that the I-PMSI scalability improvement brought by the "segmented inter-AS P-tunnels" approach is higher when segmented P-tunnels have a granularity of source AS (see item below). o Scalability for S-PMSIs: The "segmented inter-AS P-tunnels" approach, when used with the BGP-based C-multicast routing approach, provides flexibility in how the bandwidth/state trade- off is handled, to help with scalability. Indeed, in that case, the trade-off made for a said (C-S,C-G) in a downstream AS can be made more in favor of scalability than the trade-off made by the neighbor upstream AS, thanks to the ability to aggregate one or more S-PMSIs of the upstream AS in one I-PMSI tunnel in a downstream AS. o Configuration at ASBRs: Depending on whether segmented P-tunnels have a granularity of source ASBR or source AS, the "segmented inter-AS P-tunnels" approach would require respectively the same or additional configuration on ASBRs as the "non-segmented inter-AS P-tunnels" approach. o Independence of tunneling technology from one AS to another: The "segmented inter-AS P-tunnels" approach provides this; the "non- segmented inter-AS P-tunnels" approach does not. o Facilitated coexistence with, and migration from, existing deployments and lighter engineering in some scenarios: The "non- segmented inter-AS P-tunnels" approach provides this; the "segmented inter-AS P-tunnels" approach does not. The applicability of segmented or non-segmented inter-AS tunnels to a given deployment or inter-provider interconnect will depend on a number of factors specific to each service provider. However, given the different elements reminded above, it is the recommendation of
the authors that all implementations should support the segmented inter-AS model. Additionally, the authors recommend that implementations should consider supporting the non-segmented inter-AS model in order to facilitate coexistence with, and migration from, existing deployments, and to provide a lighter engineering in a restricted set of scenarios, although it is recognized that initial implementations may only support one or the other. Section 11 of the current MVPN specification [RFC6513] defines three methods to support BIDIR-PIM, as RECOMMENDED in [RFC4834]: 1. Standard DF election procedure over an MI-PMSI 2. VPN Backbone as the RPL (Section 11.1) 3. Partitioned Sets of PEs (Section 11.2) Method (1) is naturally applied to deployments using "Full per-MVPN PIM peering across an MI-PMSI" for C-multicast routing, but as indicated in [RFC6513], Section 11, the DF election may not work well in an MVPN environment, and an alternative to DF election would be desirable. The advantage of methods (2) and (3) is that they do not require running the DF election procedure among PEs. Method (2) leverages the fact that in BIDIR-PIM, running the DF election procedure is not needed on the RPL. This approach thus has the benefit of simplicity of implementation, especially in a context
where BGP-based C-multicast routing is used. However, it has the drawback of putting constraints on how BIDIR-PIM is deployed, which may not always match the requirements of MVPN customers. Method (3) treats an MVPN as a collection of sets of multicast VRFs, all PEs in a set having the same reachability information towards C-RPA but distinct from PEs in other sets. Hence, with this method, C-Bidir packet loops in MVPN are resolved by the ability to partition a VPN into disjoint sets of VRFs, each having a distinct view of the converged network. The partitioning approach to BIDIR-PIM requires either upstream-assigned MPLS labels (to denote the partition) or a unique MP2MP LSP per partition. The former is based on PE Distinguisher Labels that have to be distributed using auto-discovery BGP routes, and their handling requires the support for upstream assigned labels and context label lookups [RFC5331]. The latter, using MP2MP LSP per partition, does not have these constraints but is restricted to P-tunnel types supporting MP2MP connectivity (such as mLDP [RFC6388]). This approach to C-Bidir can work with PIM-based or BGP-based C-multicast routing procedures and is also generic in the sense that it does not impose any requirements on the BIDIR-PIM service offering. Given the above considerations, method (3) "Partitioned Sets of PEs" is the RECOMMENDED approach. In the event where method (3) is not applicable (lack of support for upstream assigned labels or for a P-tunnel type providing MP2MP connectivity), then method (1) "Standard DF election procedure over an MI-PMSI" and (2) "VPN Backbone as the RPL" are RECOMMENDED as interim solutions, (1) having the advantage over (2) of not putting constraints on how BIDIR-PIM is deployed and the drawbacks of only being applicable when PIM-based C-multicast is used and of possibly not working well in an MVPN environment. Section 126.96.36.199 of [RFC4834] states, "In the case of PIM-SM in ASM mode, engineering of the RP function requires the deployment of specific protocols and associated configurations. A service provider may offer to manage customers' multicast protocol operation on their behalf. This implies that it is necessary to consider cases where a customer's RPs are outsourced (e.g., on PEs). Consequently, a VPN solution MAY support the hosting of the RP function in a VR or VRF". However, customers who have already deployed multicast within their networks and have therefore already deployed their own internal RPs
are often reluctant to hand over the control of their RPs to their service provider and make use of a co-located RP model, and providing RP-collocation on a PE will require the activation of Multicast Source Discovery Protocol (MSDP) or the processing of PIM Registers on the PE. Securing the PE routers for such activity requires special care and additional work and will likely rely on specific features to be provided by the routers themselves. The applicability of the co-located RP model to a given MVPN will thus depend on a number of factors specific to each customer and service provider. It is therefore the recommendation that implementations should support a co-located RP model but that support for a co-located RP model within an implementation should not restrict deployments to using a co-located RP model: implementations MUST support deployments when activation of a PIM RP function (PIM Register processing and RP- specific PIM procedures) or a VRF MSDP instance is not required on any PE router and where all the RPs are deployed within the customers' networks or CEs. Section 9.1.1 of [RFC6513] "Discarding Packets from Wrong PE", allowing fully avoiding duplicates.
associated with such mechanisms. However, if there happen to be cases where they could be viably implemented relatively simply, such mechanisms may help improve migration management. RFC6513]. Note well that the implementation of the non-mandatory alternative mechanisms is not precluded. Recommendations are: o that BGP-based auto-discovery be the mandated solution for auto- discovery; o that BGP be the mandated solution for S-PMSI switching signaling; o that implementations support both the BGP-based and the full per- MVPN PIM peering solutions for PE-PE exchange of customer multicast routing until further operational experience is gained with both solutions; o that implementations use the "Partitioned Sets of PEs" approach for BIDIR-PIM support; o that implementations implement the P2MP variants of the P2P protocols that they already implement, such as mLDP, P2MP RSVP-TE, and GRE/IP-Multicast; o that implementations support segmented inter-AS tunnels and consider supporting non-segmented inter-AS tunnels (in order to maintain backward compatibility and for migration); o that implementations MUST support deployments when the activation of a PIM RP function (PIM Register processing and RP-specific PIM procedures) or VRF MSDP instance is not required on any PE router; and o that implementations support the procedures described in Section 9.1.1 of [RFC6513].
Section 3.6 on BIDIR-PIM Support. [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate Requirement Levels", BCP 14, RFC 2119, March 1997. [RFC6513] Rosen, E., Ed. and R. Aggarwal, Ed., "Multicast in MPLS/ BGP IP VPNs", RFC 6513, February 2012. [RFC6514] Aggarwal, R., Rosen, E., Morin, T., and Y. Rekhter, "BGP Encodings and Procedures for Multicast in MPLS/BGP IP VPNs", RFC 6514, February 2012. [MVPN] Aggarwal, R., "Base Specification for Multicast in BGP/ MPLS VPNs", Work in Progress, June 2004. [PIM-PORT] Farinacci, D., Wijnands, I., Venaas, S., and M. Napierala, "A Reliable Transport Mechanism for PIM", Work in Progress, October 2011. [RFC4364] Rosen, E. and Y. Rekhter, "BGP/MPLS IP Virtual Private Networks (VPNs)", RFC 4364, February 2006. [RFC4684] Marques, P., Bonica, R., Fang, L., Martini, L., Raszuk, R., Patel, K., and J. Guichard, "Constrained Route Distribution for Border Gateway Protocol/MultiProtocol Label Switching (BGP/MPLS) Internet Protocol (IP) Virtual Private Networks (VPNs)", RFC 4684, November 2006. [RFC4834] Morin, T., Ed., "Requirements for Multicast in Layer 3 Provider-Provisioned Virtual Private Networks (PPVPNs)", RFC 4834, April 2007. [RFC5331] Aggarwal, R., Rekhter, Y., and E. Rosen, "MPLS Upstream Label Assignment and Context-Specific Label Space", RFC 5331, August 2008.
[RFC5796] Atwood, W., Islam, S., and M. Siami, "Authentication and Confidentiality in Protocol Independent Multicast Sparse Mode (PIM-SM) Link-Local Messages", RFC 5796, March 2010. [RFC6037] Rosen, E., Cai, Y., and IJ. Wijnands, "Cisco Systems' Solution for Multicast in BGP/MPLS IP VPNs", RFC 6037, October 2010. [RFC6388] Wijnands, IJ., Minei, I., Kompella, K., and B. Thomas, "Label Distribution Protocol Extensions for Point-to- Multipoint and Multipoint-to-Multipoint Label Switched Paths", RFC 6388, November 2011.
advertising this route anymore. Similarly, a route reflector that had advertised this route to its client PEs before will withdraw this route when its (other) client PEs and its route reflectors peers are no longer advertising this route. In this context, the "did the last receiver leave?" question can be said to be answered by the route reflector(s). Furthermore, the BGP route distribution can leverage more than one route reflector: if multiple route reflectors are used with PEs being distributed (as clients) among these route reflectors, the "did the last receiver leave?" question is partly answered by each of these route reflectors. We can see that the "last receiver leaves" question is a part of the work that the C-multicast routing building block has to address, and the different approaches significantly differ. The different approaches for handling C-multicast routing can indeed result in a different amount of processing and how this processing is spread among the different functions. These differences can be better estimated by quantifying the amount of message processing and state maintenance. Though the type of processing, messages, and states may vary with the different approaches, we propose here a rough estimation of the load of PEs, in terms of number of messages processed and number of control plane states maintained. A "message processed" is a message being parsed, a lookup being done, and some action being taken (such as, for instance, updating a control plane or data plane state or discarding the information in the message). A "state maintained" is a multicast state kept in the control plane memory of a PE, related to an interface or a PE being subscribed to a multicast stream (note that a state will be counted on an equipment as many times as the number of protocols in which it is present, e.g., two times when present both as a PIM state and a BGP route). Note that here we don't compare the data plane states on PE routers, which wouldn't vary between the different options chosen.
o One SSM multicast stream is considered. o Only the intra-AS case is concerned (with the segmented inter-AS tunnels and BGP-based C-multicast routing, #mvpn_PE and #R_PE should refer to the PEs of the MVPN in the AS, not to all PEs of the MVPN). o The scenario is as follows: * One PE joins the multicast stream (because of a new receiver- connected site has sent a Join on the PE-CE link), followed by a number of additional PEs that also join the same multicast stream, one after the other; we evaluate the processing required for the addition of each PE. * A period of time T passes, without any PE joining or leaving (baseline). * All PEs leave, one after the other, until the last one leaves; we evaluate the processing required for the leave of each PE. o The parameters used are: * #mvpn_PE: the number of PEs in the MVPN * #R_PE: the number of PEs joining the multicast stream * #RR: the number of route reflectors * T_PIM_r: the time between two refreshes of a PIM Join (default is 60s) The estimation unit used is the "message.equipment" (or "m.e"): one "message.equipment" corresponds to "one equipment processing one message" (10 m.e being "10 equipments processing each one message", "5 messages each processed by 2 equipments", or "1 message processed by 10 equipment", etc.). Similarly, for the amount of control plane state, the unit used is "state.equipment" or "s.e". This accounts for the fact that a message (or a state) can be processed (or maintained) by more than one node. We distinguish three different types of equipments: the upstream PE for the considered multicast stream, the RR (if any), and the other PEs (which are not the upstream PE).
The numbers or orders of magnitude given in the tables in the following subsections are totals across all equipments of a same type, for each type of equipment, in the "m.e" and "s.e" units defined above. Additionally: o For PIM, only Join and Prune messages are counted: * The load due to PIM Hellos can be easily computed separately and only depends on the number of PEs in the VPN. * Message processing related to the PIM Assert mechanism is also not taken into account, for the sake of simplicity. o For BGP, all advertisements and withdrawals of C-multicast Source Tree Join routes are considered (Source-Active auto-discovery routes are not used in an SSM context); following the recommendation in Section 16 of [RFC6514], the case where the Route Target Constraint mechanisms [RFC4684] is not used is not covered. (Note that for all options provided for C-multicast routing, the procedures to set up and maintain a shortest path tree toward the source of an SSM group are the same as the procedures used to set up and maintain a shortest path tree toward an RP or a non-SSM source; the results of this section are thus re-used in Appendix A.1.2.)
RFC4684], these RRs would then send this message to only the RRs that have the upstream PE as a client. None of the other RRs and none of the other PEs will receive this message. Thus, for a message associated with a given MVPN, the total number of RRs that would need to process this message only depends on the number of RRs that maintain C-multicast routes for that MVPN and that have either the receiver-connected PE or the source-connected PE as their clients and is independent of the total number of RRs or the total number of PEs. In practice, for a given MVPN, a PE would be a client of just 2 RRs (for redundancy, an RR cluster would typically have 2 RRs). Therefore, in practice the message would need to be processed by at most 4 RRs (2 RRs if both the downstream PE and the upstream PE are the clients of the same RRs). Thus, the number of RRs that have to process a given message is at most 4. Since RRs in different RR clusters have a full Internal BGP (iBGP) mesh among themselves, each RR in the RR cluster that contains the upstream PE would receive the message from each of the RRs in the RR cluster that contains the downstream PE. Given 2 RRs per cluster, the total number of messages processed by all the RRs is 6. Additionally, as soon as there is a receiver-connected PE in each RR cluster, the number of RRs processing a C-multicast route tends quickly toward 2 (taking into account that a PE peering to RRs will be made redundant).
+------------+----------+--------------+-----------+----------------+ | | upstream | other PEs | RRs (#RR) | total across | | | PE (1) | (total | | all equipments | | | | across | | | | | | (#mvpn_PE-1) | | | | | | PEs) | | | +------------+----------+--------------+-----------+----------------+ | first PE | 2 m.e | 2 m.e | 6 m.e | 10 m.e | | joins | | | | | +------------+----------+--------------+-----------+----------------+ | for *each* | between | 2 m.e | (at most) | (at most) 10 | | additional | 0 and 2 | | 6 m.e | m.e tending | | PE joining | m.e | | tending | toward 4 m.e | | | | | toward 2 | | | | | | m.e | | +------------+----------+--------------+-----------+----------------+ | baseline | 0 | 0 | 0 | 0 | | processing | | | | | | over a | | | | | | period T | | | | | +------------+----------+--------------+-----------+----------------+ | for *each* | between | 2 m.e | (at most) | (at most) 10 | | PE leaving | 0 and 2 | | 6 m.e | m.e tending | | | m.e | | tending | toward 4 m.e | | | | | toward 2 | | +------------+----------+--------------+-----------+----------------+ | the last | 2 m.e | 2 m.e | 6 m.e | 10 m.e | | PE leaves | | | | | +------------+----------+--------------+-----------+----------------+ | total for | at most | #R_PE x 4 | (at most) | at most 10 x | | #R_PE PEs | 2 x #RRs | m.e | 6 x #R_PE | #R_PE + 2 x | | | m.e (see | | m.e | #RRs m.e | | | note | | (tending | (tending | | | below) | | toward 2 | toward 6 x | | | | | x #R_PE | #R_PE + #RRs | | | | | m.e) | m.e ) | +------------+----------+--------------+-----------+----------------+ | total | 4 s.e | 2 x #R_PE | approx. 2 | approx. 4 | | state | | s.e | #R_PE + | #R_PE + #RRx | | maintained | | | #RR x | #clusters + 4 | | | | | #clusters | m.e | | | | | s.e | | +------------+----------+--------------+-----------+----------------+ Message Processing and State Maintenance - BGP-Based Procedures
Note on the total of m.e on the upstream PE: o There are as many "message.equipment"s on the upstream PE as the number of times the RRs of the cluster of the upstream PE need to re-advertise the C-multicast (C-S,C-G) route; such a re- advertisement is not useful for the upstream PE, because the behavior of the upstream PE for a said (VPN associated to the Route Target, C-S,C-G) will not depend on the precise attributes carried by the route (other than the Route Target, of course) but will happen in some cases due to how BGP processes these routes. Indeed, a BGP peer will possibly re-advertise a route when its current best path changes for the said NLRI if the set of attributes to advertise also changes. o Let's look at the different relevant attributes and when they can influence when a re-advertisement of a C-multicast route will happen: * next-hop and originator-id: A new PE joining will not mechanically result in a need to re-advertise a C-multicast route because as the RR aggregates C-multicast routes with the same NLRI received from PEs in its own cluster (Section 11.4 of [RFC6514]), the RR rewrites the values of these attributes; however, the advertisements made by different RRs peering with the RRs in the cluster of the upstream PE may lead to updates of the value of these attributes. * cluster-list: The value of this attribute only varies between clusters, changes of the value of this attributes does not "follow" PE advertisements, and only advertisements made by different RRs may possibly lead to updates of the value of this attribute. * local-pref: The value of this attribute is determined locally; this is true both for the routes advertised by each PE (which could all be configured to use the same value) and for a route that results from the aggregation by an RR of the route with the same NLRI advertised by the PEs of his cluster (the RRs could also be configured to use a local pref independent of the local_pref of the routes advertised to him). Thus, this attribute can be considered to result in a need to re-advertise a C-multicast route. * Other BGP attributes do not have a particular reason to be set for C-multicast routes in intra-AS, and if they were, an RR (or, for attributes relevant for inter-AS, an ASBR) would also overwrite these values when aggregating these routes.
o Given the above, for a said C-multicast Source Tree Join (S,G) NLRI, what may force an RR to re-advertise the route with different attributes to the upstream PE would be the case of an RR of another cluster advertising a route better than its current best route, because of the values of attributes specific to that RR (next-hop, originator-id, cluster-list) but not because of anything specific to the PEs behind that RR. If we consider our (#R_PE -1) joining a said (C-S,C-G), one after the other after the first PE joining, some of these events may thus lead to a re- advertisement to the upstream PE, but the number of times this can happen is at worse the number of RRs in clusters having receivers (plus one because of the possible advertisement of the same route by a PE of the local cluster). o Given that we look at scalability with an increased number of PEs in this section, we need to consider the possibility that all clusters may have a client PE with a receiver. We also need to consider that the two RRs of the cluster of the upstream PE may need to re-advertise the route. With this in mind, we know that 2x#RRs is an upper bound to the number of updates made by RRs to the upstream PE, for the considered C-multicast route.
The conclusions that can be drawn from the above are as follows: o In the PIM-based approach, any message will be processed by all PEs, including those that are neither upstream nor downstream for the message; as a result, the total number of messages to process is in O(#mvpn_PE x #R_PE), i.e., O(#mvpn_PE ^ 2) if the proportion of receiver PEs is considered constant when the number of PEs increases. The refreshes of Join messages introduce a linear factor not changing the order of magnitude, but which can be significant for long-lived streams; o The BGP-based approach requires an amount of message processing in O(#R_PE) lower than the PIM-based approach. The amount is independent of the duration of streams. o State maintenance is of the same order of magnitude for all approaches: O(#R_PE), but the repartition is different: * The PIM-based approach fully spreads, and minimizes, the amount of state (one state per PE). * The BGP-based procedures spread all the state on the set of route reflectors. Appendix A.1.1 are reused in this section, for the parts that are common to the setup and maintenance of states related to a source tree or a shared tree. When PIM-SM is used in a VPN and an ASM multicast group is joined by some PEs (#R_PEs) with some sources sending toward this multicast group address, we can note the following: PEs will generally have to maintain one shared tree, plus one source tree for each source sending toward G; each tree resulting in an amount of processing and state maintenance similar to what is described in the scenario in Appendix A.1.1, with the same differences in order of magnitudes between the different approaches when the number of PEs is high. An exception to this is when, for a said group in a VPN among the PIM instances in the customer routers and VRFs, none would switch to the shortest path tree (SPT) (SwitchToSptDesired always false): in that case, the processing and state maintenance load is the one required for maintenance of the shared tree only. It has to be noted that this scenario is dependent on customer policy. To compare the resulting load in that case, between PIM-based approaches and the
BGP-based approach configured to use inter-site shared trees, the scenario in Appendix A.1.1 can be used with #R_PEs joining a (C-*, C-G) ASM group instead of an SSM group, and the same differences in order of magnitude remain true. In the case of the BGP-based approach used without inter-site shared trees, we must take into account the load resulting from the fact that to build the C-PIM shared tree, each PE has to join the source tree to each source; using the notations of Appendix A.1.1, this adds an amount of load (total load across all equipments) that is proportional to #R_PEs and the number of sources. The order of magnitude with an increasing number of PEs is thus unchanged, and the differences in order of magnitude also remain the same. Additionally, to the maintenance of trees, PEs have to ensure some processing and state maintenance related to individual sources sending to a multicast group; the related procedures and behaviors largely may differ depending on which C-multicast routing protocol is used, how it is configured, how the multicast source discovery mechanism is used in the customer VPN, and which SwitchToSptDesired policy is used. However, the following can be observed: o When BGP-based C-multicast routing is used: * Each PE will possibly have to process and maintain a BGP Source-Active auto-discovery route for (some or all) sources of an ASM group. The number of Source-Active auto-discovery routes will typically be one but may be related to the number of upstream PEs in the following cases: when inter-site shared trees are used and simultaneously more than one PE is used as the upstream PE for SPT (C-S,C-G) trees and when inter-site shared trees are used and there are multiple PEs that are possible upstream for this (S,G). * This results in message processing and state maintenance (total across all the equipments) linearly dependent on the number of PEs in the VPN (#mvpn_PE) for each source, independent of the number of PEs joined to the group. * Depending on whether or not inter-site shared trees are used, on the SwitchToSptDesired policy in the PIM instances in the customer routers and VRFs, and on the relative locations of sources and RPs, this will happen for all (S,G) of an ASM group or only for some of them and will be done in parallel to the maintenance of shared and/or source trees or at the first join of a PE on a source tree.
o When PIM-based C-multicast routing is used, depending on the SwitchToSptDesired policy in the PIM instances in the customer routers and VRFs and depending on the relative locations of sources and RPs, there are: * Possible control plane state transitions triggered by the reception of (S,G) packets. Such events would induce processing on all PEs joined to G. * Possible PIM Assert messages specific to (S,G). This would induce a message processing on each PE of the VPN for each PIM Assert message. Given the above, the additional processing that may happen for each individual source sending to the group, beyond the maintenance of source and shared trees, does not change the order of magnitude identified above. Appendix A.1.1 is done based on a use case where each PE with receivers has joined and left once. Drawing scalability-related conclusions for other patterns of changes of the set of receiver-connected PEs can be done by considering the cost of each approach for "a new PE joining" and "a PE leaving". For the "PIM LAN Procedure" approach, in the case of a single SSM or SPT tree, the total amount of message processing across all nodes depends linearly on the number of PEs in the VPN when a PE joins such a tree. For the "BGP-based" approach: o In the case of a single SSM tree, the total amount of message processing across all nodes is independent of the number of PEs, for "a new PE" joining and "a PE leaving"; it also depends on how route reflectors are meshed, but not on linear dependency. o In the case of an SPT tree for an ASM group, BGP has additional processing due to possible Source-Active auto-discovery routes: * When BGP-based C-multicast routing is used with inter-site shared trees, for the first PE joining (and the last PE leaving) a said SPT, the processing of the corresponding Source-Active auto-discovery routes results in a processing cost linearly dependent on the number of PEs in the VPN. For
subsequent PEs joining (and non-last PE leaving), there is no processing due to advertisement or withdrawal of Source-Active auto-discovery routes. * When BGP-based C-multicast routing is used without inter-site shared trees, the processing of Source-Active auto-discovery routes for an (S,G) happens independently of PEs joining and leaving the SPT for (S,G). In the case of a new PE having to join a shared tree for an ASM group G, we see the following: o The processing due to the PE joining the shared tree itself is the same as the processing required to set up an SSM tree, as described before (note that this does not happen when BGP-based C-multicast routing is used without inter-site shared trees). o For each source for which the PE joins the SPT, the resulting processing cost is the same as one SPT tree, as described before. * The conditions under which a PE will join the SPT for a said (C-S,C-G) are the same between the BGP-based with inter-site shared tree approach and the PIM-based approach, and depend solely on the SwitchToSptDesired policy in the PIM instances in the customer routers in the sites connected to the PE and/or in the VRF. * The conditions under which a PE will join the SPT for a said (C-S,C-G) differ between the BGP-based without inter-site shared trees approach and the PIM-based approach. * The SPT for a said (S,G) can be joined by the PE in the following cases: + as soon as one router, or the VPN VRF on the PE, has SwitchToSptDesired(S,G) being true + when BGP-based routing is used and configured to not use inter-site shared trees * Said differently, the only case where the PE will not join the SPT for (S,G) is when all routers in the sites of the VPN connected to the PE, or the VPN VRF itself, will never have SwitchToSptDesired(S,G) being true, with the additional condition that inter-site shared trees are used when BGP-based C-multicast routing is used.
Thus, when one PE joins a group G to which n sources are sending traffic, we note the following with regards to the dependency of the cost (in total amount of processing across all equipments) to the number of PEs: o In the general case (where any router in the site of the VPN connected to the PE, or the VRF itself, may have SwitchToSptDesired(S,G) being true): * For the "PIM LAN Procedure" approach, the cost is linearly dependent on the number of PEs in the VPN and linearly dependent on the number of sources. * For the "BGP-based" approach, the cost is linearly dependent on the number of sources, and, in the sub-case of the BGP-based approach used with inter-site shared trees, is also dependent on the number of PEs in the VPN only if the PE is the first to join the group or the SPT for some source sending to the group. o Else, under the assumption that routers in the sites of the VPN connected to the PE, and the VPN VRF itself, will never have the policy function SwitchToSptDesired(S,G) being possibly true, then: * In the case of the PIM-based approach, the cost is linearly dependent on the number of PEs in the VPN, and there is no dependency on the number of sources. * In the case of the BGP-based approach with inter-site shared trees, the cost is linearly dependent on the number of RRs, and there is no dependency on the number of sources. * In the case of the BGP-based approach without inter-site shared trees, the cost is linearly dependent on the number of RRs and on the number of sources. Hence, with the PIM-based approach, the overall cost across all equipments of any PE joining an ASM group G is always dependent on the number of PEs (same for a PE that leaves), while the BGP-based approach has a cost independent of the number of PEs. An exception is the first PE joining the ASM group for the BGP-based approach used without inter-site shared trees; in that case, there is a dependency with the number of PEs. On the dependency with the number of sources, without making any assumption on the SwitchToSptDesired policy on PIM routers and VRFs of a VPN, we see that a PE joining an ASM group may induce a processing cost linearly dependent on the number of sources. Apart from this general case, under the condition where the
SwitchToSptDesired is always false on all PIM routers and VRFs of the VPN, then with the PIM-based approach, and with the BGP-based approach used with inter-site shared trees, the cost in amount of messages processed will be independent of the number of sources (it has to be noted that this condition depends on customer policy). RFC6513] and is here for reference only.) In early versions of the document that became [RFC6513], two approaches were proposed for how a source PE can decide when to start transmitting customer multicast traffic on a S-PMSI: 1. The source PE sends multicast packets for the (C-S,C-G) on both the I-PMSI P-multicast tree and the S-PMSI P-multicast tree simultaneously for a pre-configured period of time, letting the receiver PEs select the new tree for reception before switching to only the S-PMSI. 2. The source PE waits for a pre-configured period of time after advertising the (C-S,C-G) entry bound to the S-PMSI before fully switching the traffic onto the S-PMSI-bound P-multicast tree. The first alternative had essentially two drawbacks: o (C-S,C-G) traffic is sent twice for some period of time, which would appear to be at odds with the motivation for switching to an S-PMSI in order to optimize the bandwidth used by the multicast tree for that stream. o It is unlikely that the switchover can occur without packet loss or duplication if the transit delays of the I-PMSI P-multicast tree and the S-PMSI P-multicast tree differ. By contrast, the second alternative has none of these drawbacks and satisfies the requirement in Section 5.1.3 of [RFC4834], which states that "a multicast VPN solution SHOULD as much as possible ensure that client multicast traffic packets are neither lost nor duplicated, even when changes occur in the way a client multicast data stream is carried over the provider network". The second alternative also happens to be the one used in existing deployments. Consistent with this analysis, only the second alternative is discussed in [RFC6513].