| Internet-Draft | bgp-mcast-controller | July 2023 |
| Zhang, et al. | Expires 11 January 2024 | [Page] |
This document specifies a way that one or more centralized controllers can use BGP to set up multicast distribution trees (identified by either IP source/destination address pair, or mLDP FEC) in a network. Since the controllers calculate the trees, they can use sophisticated algorithms and constraints to achieve traffic engineering. The controllers directly signal dynamic replication state to tree nodes, leading to very simple multicast control plane on the tree nodes, as if they were using static routes. This can be used for both underlay and overlay multicast trees, including replacing BGP-MVPN signaling.¶
The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in BCP 14 [RFC2119] [RFC8174] when, and only when, they appear in all capitals, as shown here.¶
This Internet-Draft is submitted in full conformance with the provisions of BCP 78 and BCP 79.¶
Internet-Drafts are working documents of the Internet Engineering Task Force (IETF). Note that other groups may also distribute working documents as Internet-Drafts. The list of current Internet-Drafts is at https://datatracker.ietf.org/drafts/current/.¶
Internet-Drafts are draft documents valid for a maximum of six months and may be updated, replaced, or obsoleted by other documents at any time. It is inappropriate to use Internet-Drafts as reference material or to cite them other than as "work in progress."¶
This Internet-Draft will expire on 11 January 2024.¶
Copyright (c) 2023 IETF Trust and the persons identified as the document authors. All rights reserved.¶
This document is subject to BCP 78 and the IETF Trust's Legal Provisions Relating to IETF Documents (https://trustee.ietf.org/license-info) in effect on the date of publication of this document. Please review these documents carefully, as they describe your rights and restrictions with respect to this document. Code Components extracted from this document must include Revised BSD License text as described in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the Revised BSD License.¶
[I-D.ietf-bess-bgp-multicast] describes a way to use BGP as a replacement signaling for PIM [RFC7761] or mLDP [RFC6388]. The BGP-based multicast signaling described there provides a mechanism for setting up both (s,g)/(*,g) multicast trees (as PIM does, but optionally with labels) and labeled (MPLS) multicast tunnels (as mLDP does). Each router on a tree performs essentially the same procedures as it would perform if using PIM or mLDP, but all the inter-router signaling is done using BGP.¶
These procedures allow the routers to set up a separate tree for each individual multicast (x,g) flow where the 'x' could be either 's' or '*', but they also allow the routers to set up trees that are used for more than one flow. In the latter case, the trees are often referred to as "multicast tunnels" or "multipoint tunnels", and specifically in this document they are mLDP tunnels (except that they are set up with BGP signaling). While it actually does not have to be restricted to mLDP tunnels, mLDP FEC is conveniently borrowed to identify the tunnel. In the rest of the document, the term tree and tunnel are used interchangeably.¶
The trees/tunnels are set up using the "receiver-initiated join" technique of PIM/mLDP, hop by hop from downstream routers towards the root. The BGP messages of MCAST-TREE SAFI are either sent hop by hop between downstream routers and their upstream neighbors, or can be reflected by Route Reflectors (RRs).¶
As an alternative to each hop independently determining its upstream router and signaling upstream towards the root (following PIM/mLDP model), the entire tree can be calculated by a centralized controller, and the signaling can be entirely done from the controller using the same MCAST-TREE SAFI. For that, some additional procedures and optimizations are specified in this document.¶
[I-D.ietf-bess-bgp-multicast] uses S-PMSI, Leaf, and Source Active Auto-Discovery (A-D) routes because the main procedures and concepts are borrowed from the BGP-MVPN [RFC6514]. While the same Leaf A-D routes can be used to signal replication state to tree nodes from controllers, this document introduces a new route type "Replication State" for the same functionality, so that familiarity with the BGP-MVPN concepts is not required.¶
While it is outside the scope of this document, signaling from the controllers could be done via other means as well, like Netconf or any other SDN methods.¶
Each router could establish direct BGP sessions with one or more controllers, or it could establish BGP sessions with RRs who in turn peer with controllers. For the same tree/tunnel, each controller may independently calculate the tree/tunnel and signal the routers on the tree/tunnel using MCAST-TREE Replication State routes. How the calculation is done are outside the scope of this document.¶
On each router, BGP route selection rules will lead to one controller's route for the tree/tunnel being selected as the active route and used for setting up forwarding state. As long as all the routers on a tree/tunnel consistently pick the same controller's routes for the tree/tunnel, the setup should be consistent. If the tree/tunnel is labeled, different labels will be used from different controllers so there is no traffic loop issue even if the routers do not consistently select the same controller's routes. In the unlabeled case, to ensure the consistency the selection SHOULD be solely based on the identifier of the controller.¶
Another consistency issue is when a bidirectional tree/tunnel needs to be re-routed. Because this is no longer triggered hop-by-hop from downstream to upstream, it is possible that the upstream change happens before the downstream, causing traffic loop. In the unlabeled case, there is no good solution (other than that the controller issues upstream change only after it gets acknowledgement from downstream). In the labeled case, as long as a new label is used there should be no problem.¶
Besides the traffic loop issue, there could be transient traffic loss before both the upstream and downstream's forwarding state are updated. This could be mitigated if the upstream keep sending traffic on the old path (in addition to the new path) and the downstream keep accepting traffic on the old path (but not on the new path) for some time. It is a local matter when for the downstream to switch to the new path - it could be data driven (e.g., after traffic arrives on the new path) or timer driven.¶
For each tree, multiple disjoint instances could be calculated and signaled for live-live protection. Different labels are used for different instances, so that the leaves can differentiate incoming traffic on different instances. As far as transit routers are concerned, the instances are just independent. Note that the two instances are not expected to share common transit routers (it is otherwise outside the scope of this document/revision).¶
When a router receives a Replication State route, the re-advertisement is blocked if a configured import RT matches the RT of the route, which indicates that this router is the target and consumer of the route hence it should not be re-advertised further. The routes includes the forwarding information in the form of Tunnel Encapsulation Attributes (TEA) [RFC9012], with enhancements specified in this document.¶
Suppose that for a particular tree, there are two downstream routers D1 and D2 for a particular upstream router U. A controller C sends one Replication State route to U, with the Tree Node's IP Address field (see Section 3.4) set to U's IP address and the TEA specifying both the two downstreams and its upstream (see Section 3.1.5). In this case, the Originating Router's Address field of the Replication State route is set to the controller's address. Note that for a TEA attached to a unicast NLRI, only one of the tunnels in a TEA is used for forwarding a particular packet, while all the tunnels in a TEA are used to reach multiple endpoints when it is attached to a multicast NLRI.¶
It could be that U may need to replicate to many downstream routers, say D1 through D1000. In that case, it may not be possible to encode all those branches in a single TEA, or may not be optimal to update a large TEA when a branch is added/removed. In that case, C may send multiple Replication State routes, each with a different RD and a different TEA that encodes a subset of the branches. This provides a flexible way to optimize the encoding of large number of branches and incremental updates of branches.¶
Notice that, in case of labeled trees, the (x,g) or mLDP FEC signaling is actually not needed to transit routers but only needed to tunnel root/leaves. However, for consistency among the root/leaf/transit nodes, and for consistency with the hop-by-hop signaling, the same signaling (with tree identification encoded in the NLRI) is used to all routers.¶
Nonetheless, a new NLRI route type of the MCAST-TREE SAFI is defined to encode label/SID instead of tree identification in the NLRI, for scenarios where there is really no need to signal tree identification, e.g. as described in Section 2. On a tunnel root, the tree's binding SID can be encoded in the NLRI.¶
For a tree node to acknowledge to the controller that it has received the signaling and installed corresponding forwarding state, it advertises a corresponding Replication State route, with the Originating Router's IP Address set to itself and with a Route Target to match the controller. For comparison, the tree signaling Replication State route from the controller has the Originating Router's IP Address set to the controller and the Route Target matching the tree node. The two Replication State routes (for controller to signal to a tree node and for a tree node to acknowledge back) differ only in those two aspects.¶
With the acknowledgement Replication State routes, the controller knows if tree setup is complete. The information can be used for many purposes, e.g. the controller may instruct the ingress to start forwarding traffic onto a tree only after it knows that the tree setup has completed.¶
In the case of labeled multicast signaled hop by hop towards the root, whether it's (x,g) multicast or "mLDP" tunnel, labels are assigned by a downstream router and advertised to its upstream router (from traffic direction point of view). In the case of controller based signaling, routers do not originate tree join routes anymore, so the controllers have to assign labels on behalf of routers, and there are three options for label assignment:¶
Assignment from each router's SRLB is no different from each router assigning labels from its own local label space in the hop-by-hop signaling case. The assignments for one router is independent of assignments for another router, even for the same tree.¶
Assignment from the controller's local label space is upstream-assigned [RFC5331]. It is used if the controller does not learn the common SRGB or each router's SRLB. Assignment from the SRGB [RFC8402] is only meaningful if all SRGBs are the same and a single common label is used for all the routers on a tree in case of unidirectional tree/tunnel (Section 1.4.1). Otherwise, assignment from SRLB is preferred.¶
The choice of which of the options to use depends on many factors. An operator may want to use a single common label per tree for ease of monitoring and debugging, but that requires explicit RPF checking and either common SRGB or upstream assigned labels, which may not be supported due to either the software or hardware limitations (e.g. label imposition/disposition limits). In an SR network, assignment from the common SRGB if it's required to use a single common label per unidirectional tree, or otherwise assignment from SRLB is a good choice because it does not require support for context label spaces.¶
MPLS labels only have local significance. For an LSP that goes through a series of routers, each router allocates a label independently and it swaps the incoming label (that it advertised to its upstream) to an outgoing label (that it received from its downstream) when it forwards a labeled packet. Even if the incoming and outgoing labels happen to be the same on a particular router, that is just incidental.¶
With Segment Routing, it is becoming a common practice that all routers use the same SRGB so that a SID maps to the same label on all routers. This makes it easier for operators to monitor and debug their network. The same concept applies to multicast trees as well - a common per-tree label can be used for a router to receive traffic from its upstream neighbor and replicate traffic to all its downstream neighbor.¶
However, a common per-tree label can only be used for unidirectional trees. In case of bidirectional trees, the common label needs to be per- <tree, direction>. Additionally, unless the entire tree is updated for every tree node to use a new common per-tree or per-<tree, direction>label with any change in the tree (no matter how small and local the change is), it requires each router to do explicit RPF check, so that only packets from its expected upstream neighbor are accepted. Otherwise, traffic loop may form during topology changes, because the forwarding state update is no longer ordered.¶
Traditionally, p2mp mpls forwarding does not require explicit RPF check as a downstream router advertises a label only to its upstream router and all traffic with that incoming label is presumed to be from the upstream router and accepted. When a downstream router switches to a different upstream router a different label will be advertised, so it can determine if traffic is from its expected upstream neighbor purely based on the label. Now with a single common label used for all routers on a tree to send and receive traffic with, a router can no longer determine if the traffic is from its expected neighbor just based on that common tree label. Therefore, explicit RPF check is needed. Instead of interface based RPF checking as in PIM case, neighbor based RPF checking is used - a label identifying the upstream neighbor precedes the common tree label and the receiving router checks if that preceding neighbor label matches its expected upstream neighbor. Notice that this is similar to what's described in Section "9.1.1 Discarding Packets from Wrong PE" of RFC 6513 (an egress PE discards traffic sent from a wrong ingress PE). The only difference is one is used for label based forwarding and the other is used for (s,g) based forwarding..¶
Both the common per-tree label and the neighbor label are allocated either from the common SRGB or from the controller's local label space. In the latter case, an additional label identifying the controller's label space is needed, as described in the following section.¶
In this case in the multicast packet's label stack the tree label and upstream neighbor label (if used in case of single common-label per tree) are preceded by a downstream-assigned "context label". The context label identifies a context-specific label space (the controller's local label space), and the upstream-assigned label that follows it is looked up in that space.¶
This specification requires that, in case of upstream-assignment from a controller's local label space, each router D to assign, corresponding to each controller C, a context label that identifies the upstream-assigned label space used by that controller. This label, call it Lc-D, is communicated by D to C via BGP-LS [RFC 7752].¶
Suppose a controller is setting up unidirectional tree T. It assigns that tree the label Lt, and assigns label Lu to identify router U which is the upstream of router D on tree T. C needs to tell U: "to send a packet on the given tree/tunnel, one of the things you have to do is push Lt onto the packet's label stack, then push Lu, then push Lc-D onto the packet's label stack, then unicast the packet to D". Controller C also needs to inform router D of the correspondence between <Lc-D, Lu, Lt> and tree T.¶
To achieve that, when C sends a Replication State route, for each tunnel in the TEA, it may include a label stack Sub-TLV [RFC9012], with the outer label being the context label Lc-D (received by the controller from the corresponding downstream), the next label being the upstream neighbor label Lu, and the inner label being the label Lt assigned by the controller for the tree. The router receiving the route will use the label stacks to send traffic to its downstreams.¶
For C to signal the expected label stack for D to receive traffic with, we overload a tunnel TLV in the TEA of the Replication State route sent to D - if the tunnel TLV has a RPF sub-TLV (Section 3.1.5), then it indicates that this is actually for receiving traffic from the upstream.¶
For the controller to calculate a tree, it needs to determine the root and leaves of the tree. This may be based on provisioning (static or dynamically programmed), or based on BGP signaling as described in the following two sections.¶
In both of the following cases, the BGP updates are targeted at the controller, via an address specific Route Target with Global Administration Field set to the controller's address and the Local Administration Field set to 0.¶
In this case, the PIM Last Hop Routers (LHRs) with interested receivers or mLDP tunnel leaves encode a Leaf A-D route ([I-D.ietf-bess-bgp-multicast]) with the Upstream Router's IP Address field set to the controller's address and the Originating Router's IP Address set to the address of the LHR or the P2MP tunnel leaf. The encoded PIM SSM source or mLDP FEC provides root information and the Originating Router's IP Address provides leaf information.¶
In this case, the First Hop Routers (FHRs) originate Source Active routes which provides root information, and the LHRs originate Leaf A-D routes, encoded as in the PIM-SSM case except that it is (*,G) instead of (S,G). The Leaf A-D routes provide leaf information.¶
An end to end multicast tree may span multiple routing domains, and the setup of the tree in each domain may be done differently as specified in [I-D.ietf-bess-bgp-multicast]. This section discusses a few aspects specific to controller signaling.¶
Consider two adjacent domains each with its own controller in the following configuration where router B is an upstream node of C for a multicast tree:¶
|
domain 1 | domain 2
|
ctrlr1 | ctrlr2
/\ | /\
/ \ | / \
/ \ | / \
A--...-B--|--C--...-D
|
¶
In the case of native (un-labeled) IP multicast, nothing special is needed. Controller 1 signals B to send traffic out of B-C link while Controller 2 signals C to accept traffic on the B-C link.¶
In the case of labeled IP multicast or mLDP tunnel, the controllers may be able to coordinate their actions such that Controller 1 signals B to send traffic out of B-C link with label X while Controller 2 signals C to accept traffic with the same label X on the B-C link. If the coordination is not possible, then C needs to use hop-by-hop BGP signaling to signal towards B, as specified in [I-D.ietf-bess-bgp-multicast].¶
The configuration could also be as following, where router B borders both domain 1 and domain 2 and is controlled by both controllers:¶
|
domain 1 | domain 2
|
ctrlr1 | ctrlr2
/\ | /\
/ \ | / \
/ \ | / \
/ \|/ \
A--...---B--...---C
|
¶
As discussed in Section 1.2, when B receives signaling from both Controller 1 and Controller 2, only one of the routes would be selected as the best route and used for programming the forwarding state of the corresponding segment. For B to stitch the two segments together, it is expected for B to know by provisioning that it is a border router so that B will look for the other segment (represented by the signaling from the other controller) and stitch the two together.¶
Multicast with BGP signaling from controllers can be an alternative to BGP-MVPN [RFC6514]. It is an attractive option especially when the controller can easily determine the source and leaf information.¶
With BGP-MVPN, distributed signaling is used for the following:¶
Based on the above signaled information, an ingress PE builds forwarding state to forward traffic arriving on the PE-CE interface to the provider tunnel (and local interfaces if there are local downstream receivers), and an egress PE builds forwarding state to forward traffic arriving on a provider tunnel to local interfaces with downstream receivers.¶
Notice that multicast with BGP signaling from controllers essentially programs "static" forwarding state onto multicast tree nodes. As long as a controller can determine how a C-multicast flow should be forwarded on ingress/egress PEs, it can signal to the ingress/egress PEs using the procedures in this document to set up forwarding state, removing the need of the above-mentioned distributed signaling and processing.¶
For the controller to learn the egress PEs for a C-multicast tree (so that it can set up or find a corresponding provider tunnel), the egress PEs advertise MCAST-TREE Leaf A-D routes (Section 1.5.1) towards the controller to signal its desire to join C-multicast trees, each with an appropriate RD and an extended community derived from the Route Target for the VPN ([I-D.ietf-idr-rt-derived-community]) so that the controller knows which VPN it is for. The controller then advertises corresponding MCAST-TREE Replication State routes to set up C-multicast forwarding state on ingress and egress PEs. To encode the provider tunnel information in the MCAST-TREE Replication State route for an ingress PE, the TEA can explicitly list all replication branches of the tunnel, or just the binding SID for the provider tunnel in the form of Segment List tunnel type, if the tunnel has a binding SID.¶
The Replication State route may also have a PMSI Tunnel Attribute (PTA) attached to specify the provider tunnel while the TEA specifies the local PE-CE interfaces where traffic need to be sent out. This not only allows provider tunnel without a binding SID (e.g., in a non-SR network) to be specified without explicitly listing its replication branches, but also allows the service controller for MVPN overlay state to be independent of provider tunnel setup (which could be from a different transport controller or even without a controller).¶
However, notice that if the service controller and transport controller are different, then the service controller needs to signal the transport controller the tree information: identification, set of leaves, and applicable constraints. While this can be achieved (see Section 1.5.1), it is easier for the service and transport controller to be the same.¶
Depending on local policy, a PE may add PE-CE interfaces to its replication state based on local signaling (e.g., IGMP/PIM) instead of completely relying on signaling from controllers.¶
If dynamic switching between inclusive and selective tunnels based on data rate is needed, the ingress PE can advertise/withdraw S-PMSI routes targeted only at the controllers, without PMSI Tunnel Attribute attached. The controller then updates relevant MCAST-TREE Replication State routes to update C-multicast forwarding states on PEs to switch to a new tunnel.¶
A TEA may encode a list of tunnels. A TEA attached to an MCAST-TREE NLRI encodes replication information for a <tree, node > that is identified by the NRLI. Each tunnel in the TEA identifies a branch - either an upstream branch towards the tree root (Section 3.1.5) or a downstream branch towards some leaves. A tunnel in the TEA could have an outer encapsulation (e.g. MPLS label stack) or it could just be a one-hop direct connection for native IP multicast forwarding without any outer encapsulation.¶
This document specifies three new Tunnel Types and four new sub-TLVs. The type codes will be assigned by IANA from the "BGP Tunnel Encapsulation Attribute Tunnel Types".¶
When a multicast packet needs to be sent from an upstream node to a downstream node, it may not matter how it is sent - natively when the two nodes are directly connected or tunneled otherwise. In case of tunneling, it may not matter what kind of tunnel is used - MPLS, GRE, IPinIP, or whatever.¶
To support this, an "Any-Encapsulation" tunnel type of value 20 is defined. This tunnel MAY have a Tunnel Egress Endpoint and other Sub-TLVs. The Tunnel Egress Endpoint Sub-TLV specifies an IP address, which could be any of the following:¶
Consider that a multicast packet needs to be sent to a downstream node, which could be reached via four paths P1~P4. If it does not matter which of path is taken, an "Any-Encapsulation" tunnel with the Tunnel Egress Endpoint Sub-TLV specifying the downstream node's loopback address works well. If the controller wants to specify that only P1~P2 should be used, then a "Load-balancing" tunnel needs to be used, listing P1 and P2 as member tunnels of the "Load-balancing" tunnel.¶
A load-balancing tunnel has one "Member Tunnels" Sub-TLV defined in this document. The Sub-TLV is a list of tunnels, each specifying a way to reach the downstream. A packet will be sent out of one of the tunnels listed in the Member Tunnels Sub-TLV of the load-balancing tunnel.¶
A Segment List tunnel has a Segment List sub-TLV. The encoding of the sub-TLV is as specified in Section 2.4.4 of [I-D.ietf-idr-segment-routing-te-policy].¶
While [I-D.ietf-bess-bgp-multicast] uses S-PMSI A-D routes to signal forwarding information for MP2MP upstream traffic, when controller signaling is used, a single Replication State route is used for both upstream and downstream traffic. Since different upstream and downstream labels need to be used, a new "Receiving MPLS Label Stack" of type TBD is added as a tunnel sub-TLV in addition to the existing MPLS Label Stack sub-TLV. Other than type difference, the two are the encoded the same way.¶
The Receiving MPLS Label Stack sub-TLV is added to each downstream tunnel in the TEA of Replication State route for an MP2MP tunnel to specify the forwarding information for upstream traffic from the corresponding downstream node. A label stack instead of a single label is used because of the need for neighbor based RPF check, as further explained in the following section.¶
The Receiving MPLS Label Stack sub-TLV is also used for downstream traffic from the upstream for both P2MP and MP2MP, as specified below.¶
The RPF sub-TLV is of type 124 allocated by IANA and has a one-octet length. The length is 0 currently, but if necessary in the future, sub-sub-TLVs could be placed in its value part. If the RPF sub-TLV appears in a tunnel, it indicates that the "tunnel" is for the upstream node instead of a downstream node.¶
In case of MPLS, the tunnel contains an Receiving MPLS Label Stack sub-TLV for downstream traffic from the upstream node, and in case of MP2MP it also contains a regular MPLS Label Stack sub-TLV for upstream traffic to the upstream node.¶
The inner most label in the Receiving MPLS Label Stack is the incoming label identifying the tree (for comparison the inner most label for a regular MPLS Label Stack is the outgoing label). If the Receiving MPLS Label Stack sub-TLVe has more than one labels, the second inner most label in the stack identifies the expected upstream neighbor and explicit RPF checking needs to be set up for the tree label accordingly.¶
The MPLS Label Stack sub-TLV can be used to specify the complete label stack used to send traffic, with the stack including both a transport label (stack) and label(s) that identify the (tree, neighbor) to the downstream node. There are cases where the controller only wants to specify the tree-identifying labels but leave the transport details to the router itself. For example, the router could locally determine a transport label (stack) and combine with the tree-identifying labels signaled from the controller to get the complete outgoing label stack.¶
For that purpose, a new Tree Label Stack sub-TLV of type 125 is defined, with a one-octet length field. It MAY appear in an Any-Encapsulation tunnel. The value field contains a label stack with the same encoding as value part of the MPLS Label Stack sub-TLV, but with a different type. A stack is specified because it may take up to three labels (see Section 1.4):¶
For a router to signal the context label that it assigns for a controller (or any label allocator that assigns labels - from its local label space - that will be received by this router), a new BGP-LS Node Attribute TLV is defined:¶
0 1 2 3
0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Type | Length |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Context Label |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| IPv4/v6 Address of Label Space Owner |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
¶
The Length field implies the type of the address. Multiple Context Label TLVs may be included in a Node Attribute, one for each label space owner.¶
An as example, a controller with address 11.11.11.11 allocates label 200 from its own label space, and router A assigns label 100 to identify this controller's label space. The router includes the Context Label TLV (100, 11.11.11.11) in its BGP-LS Node Attribute and the controller instructs router B to send traffic to router A with a label stack (100, 200), and router A uses label 100 to determine the Label FIB in which to look up label 200.¶
A tree node needs to acknowledge to the controller the success/failure in installing forwarding state for a tree. In case of failure, an MCAST NACK Extended Community is attached. The value field is set to 0. In the future, flag bits may be defined to signal specific failure.¶
The MCAST NACK Extended Community is an MCAST Extended Community with a sub-type to be assigned by IANA. The MCAST Extended Community is a new Extended Community with a type to be assigned by IANA.¶
0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Type=TBD1 | Sub-Type=TBD2 | Reserved=0 |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Reserved=0 |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
¶
The NLRI route type for signaling from controllers to tree nodes is "Replication State". The NLRI has the following format:¶
+-----------------------------------+
| Route Type - Replication State |
+-----------------------------------+
| Length (1 octet) |
+-----------------------------------+
| Tree Type (1 octet) |
+-----------------------------------+
|Tree Type Specific Length (1 octet)|
+-----------------------------------+
| RD (8 octets) |
+-----------------------------------+
~ Tree Identification (variable) ~
+-----------------------------------+
| Tree Node's IP Address |
+-----------------------------------+
| Originator's IP Address |
+-----------------------------------+
Replication State NLRI
¶
Notice that Replication State is just a new route type with the same format of Leaf A-D route except some fields are renamed:¶
With this arrangement, IP multicast tree and mLDP tunnel can be signaled via Replication State routes from controllers, or via Leaf A-D routes either hop by hop or from controllers with maximum code reuse, while newer types of trees can be signaled via Replication State routes with maximum code reuse as well.¶
As described in Section 1.3, tree label instead of tree identification could be encoded in the NLRI to identify the tree in the control plane as well as in the forwarding plane. For that a new Tree Type of 2 is used and the Replication State route has the following format:¶
+-------------------------------------+
| Route Type - Replication State |
+-------------------------------------+
| Length (1 octet) |
+-------------------------------------+
| Tree Type 2 (Label as Tree ID) |
+-------------------------------------+
|Tree Type specific Length (1 octet) |
+-------------------------------------+
| RD (8 octets) |
+-------------------------------------+
~ Label Stack (variable) ~
+-------------------------------------+
| Tree Node's IP Address |
+-------------------------------------+
| Originating Router's IP Address |
+-------------------------------------+
Replication State route for tree identification by label stack
¶
As discussed in Section 1.4.2, a label stack may have to be used to identify a tree in the data plane so a label stack is encoded here. The number of labels is derived from the Tree Type Specific Length field. Each label stack entry is encoded as following:¶
0 1 2 3
0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Label |0 0 0 0 0 0 0 0 0 0 0 0|
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
¶
This section applies to MPLS data plane. While the concept of BGP signaling applies to SRv6 data plane as well, SRv6 related specification is outside the scope of this document. Note that, this document does not assume Segment Routing is used, even though the SRGB/SRLB terminologies are used to describe label blocks, and some scenarios of Segment routing are considered.¶
In case of labeled trees for either (x, g) IP multicast or mLDP tunnels, an operator first determines which of the following methods is used to allocate tree-identifying labels, as explained in Section 1.4:¶
For option 2 and 3, how the controller learns the common SRGB or each node's SRLB is outside the scope of this document.¶
For option 1, each tree node MUST advertise a label from its default label space to identify the controller's label space, via the Context Label TLV in BGP-LS Node Attribute (Section 3.2). The tree-identifying label in TEA and packets MUST be preceded by the label-space-identifying label.¶
For option 1 and 2, the operator also determines if the controller allocates a new label for each tree or <tree, direction> and resignal to all tree nodes even when only some tree nodes need to be changed. If not, then another neighbor-identifying label needs to precede the tree-identifying label (and follow the label-space-identifying label in case of option 1). The neighbor-identifying label MUST be allocated from the same label space or SRGB from which the tree-identifying label is allocated.¶
To generalize, a label stack of one label (for option 3), two labels (for option 2 and 1 if neighbor-identifying label is not needed), or three labels (for option 2 and 1 if neighbor-identifying label is needed). In the rest of the document, tree-identifying label(-stack) term is used generically.¶
After the controller calculates a tree, it constructs one or more Replication State Route for each tree node as following:¶
The TEA includes one or more tunnels. If the route is for the root node, one and only one tunnel of type Any-Encapsulation MAY be included with a RPF sub-TLV and a Receiving MPLS Label Stack sub-TLV to encode a binding SID if it is assigned for the tree. If the route is for a leaf or bud node (which is both a leaf node and transit node at the same time), one and only one tunnel of type Any-Encapsulation MUST be included with a Tunnel Egress Endpoint sub-TLV whose address is set to a loopback address on the node.¶
Additionally, for any node, a tunnel is included for each upstream/downstream node. Each tunnel MUST include a Tunnel Egress Endpoint sub-TLV if it is needed to derive forwarding information. Otherwise, the sub-TLV MAY be included for informational purposes.¶
A tunnel MAY be one of the following types:¶
Other tunnel types and sub-TLVs may also be used but not specified here. This specification is intentionally not stringent to favor flexibility - as long as the route can be syntactically parsed, no BGP protocol error will be reported. However, missing/conflicting/inappropriate sub-TLVs may cause multicast forwarding state not to be created, and that is a problem for multicast service only. In such case, appropriate (dampened) logs MUST be generated on the targeted node, and an negative acknowledgement MUST be sent back to the controller.¶
Each potential tree node MUST be (auto-)configured with an IP Address Specific Route Target to import Replication State Routes targeted at itself. The Route Target's Global Administration Field is set to a local address known to be used by the controller to identify the node, and the Local Administration Field is set to 0.¶
When a BGP speaker receives Replication State Route and the attached Route Target matches its (auto-)configured Route Target to import the route, it MUST stops re-advertising the route further. Otherwise, normal BGP route propagation rules apply.¶
If an imported Replication State Route carries an Extented Community derived from a Route Target for a local VRF, the route is imported into that VRF otherwise it is imported into the default routing instance.¶
For the same <tree, node>, there may be multiple routes imported, from the same or different controllers. BGP best route selection process selects one of them as the active path. Without considering the RD field, all routes with the same NLRI as the active path MUST be considered together to create forwarding state on the node for the tree. Recall that multiple such routes may be advertised when it is desired to signal a large set of replication branches via multiple routes.¶
The forwarding state includes two parts - the "nexthop" part that has the replication branches and the "route" part (with the key being a label or (x,g) tuple), just like how a unicast IP route points to a nexthop in a RIB/FIB.¶
The error checking is intentionally lenient: as long as the route can be syntactically parsed, no BGP protocol error will be reported. However, missing/conflicting/inappropriate sub-TLVs may cause multicast forwarding state not to be created, and that is a problem for multicast service only. In such case, appropriate (dampened) logs MUST be generated, and an negative acknowledgement MUST be sent back to the controller (Section 4.3.3).¶
The receiving node goes through the tunnels in the TEAs in all the relevant routes as described above, and build the nexthop as a collection of replication branches.¶
If a tunnel is of type Any-Encapsulation, it must have a Tunnel Egress Endpoint sub-TLV.¶
If the egress endpoint address is a local interface address, the interface is the replication branch. The interface could be a loopback, indicating that traffic needs to be delivered locally off the tree, e.g.:¶
The above procedures build a nexthop to be pointed to by some label or (x,g) routes. The routes are determined by checking the tree identification in the NLRI and tunnels in the TEA.¶
If the tree is a bidirectional (*,g) IP multicast, a (*,g) route is installed, pointing to the nexthop built as above,¶
If the tree is a unidirectional (x,g) IP multicast, one of the tunnels MUST have the RPF sub-TLV (referred to as the RPF tunnel) with a Tunnel Egress Endpoint sub-TLV with a local interface address. If it has no Receiving MPLS Label Stack sub-TLV, an (x,g) route MUST be installed with the corresponding interface as the expected incoming interface and the route points to the nexthop built as above. The route MAY be installed even if there is a receiving MPLS Label Stack sub-TLV in the tunnel - this is to allow native IP multicast packets to be put onto the tree at this node.¶
If the tree is unidirectional, only one of the tunnels MAY have a Receiving MPLS Label Stack sub-TLV. If it is bidirectional, multiple tunnels MAY have the Receiving MPLS Label Stack sub-TLV. For each tunnel with the Receiving MPLS Label Stack sub-TLV:¶
After processing a received Replication State Route, the node MUST send an acknowledgement back to the controller. It originates a route with similar NLRI except that the Originating Router's IP Address is set to the same Tree Node's IP Address. It attaches a IP Address Specific Route Target with the Global Administration Field set to the same as the Originating Router's IP Address in the receive route and the Local Administration Field to 0.¶
If the processing is not successful (e.g., due to missing/conflicting/inappropriate sub-TLVs in the TEA), an MCAST NACK Extended Community MUST be attached.¶
This document does not introduce new security implications beyond what a typical BGP-based controller-to-node signaling of forwarding state.¶
IANA has assigned the following code points:¶
This document makes the following additional IANA requests:¶
Create a "Tree Type Registry for Replication State Route", with the following initial assignments:¶
Create an MCAST Extended Community Sub-Type registry with the following initial assignments:¶
The registration procedure is First Come First Served.¶
The authors Eric Rosen for his questions, suggestions, and help finding solutions to some issues like the neighbor based explicit RPF checking. The authors also thank Lenny Giuliano, Sanoj Vivekanandan and IJsbrand Wijnands for their review and comments.¶