Network Working Group R. Aggarwal (Juniper) Internet Draft D. Papadimitriou (Alcatel) Expiration Date: June 2005 S. Yasukawa (NTT) Editors Extensions to RSVP-TE for Point to Multipoint TE LSPs draft-ietf-mpls-rsvp-te-p2mp-01.txt Status of this Memo By submitting this Internet-Draft, we certify that any applicable patent or IPR claims of which we are aware have been disclosed, and any of which we become aware will be disclosed, in accordance with RFC 3668. Internet-Drafts are working documents of the Internet Engineering Task Force (IETF), its areas, and its working groups. Note that other groups may also distribute working documents as Internet- Drafts. Internet-Drafts are draft documents valid for a maximum of six months and may be updated, replaced, or obsoleted by other documents at any time. It is inappropriate to use Internet-Drafts as reference material or to cite them other than as ``work in progress.'' The list of current Internet-Drafts can be accessed at http://www.ietf.org/ietf/1id-abstracts.txt The list of Internet-Draft Shadow Directories can be accessed at http://www.ietf.org/shadow.html. Abstract This document describes extensions to Resource Reservation Protocol - Traffic Engineering (RSVP-TE) for the setup of Traffic Engineered (TE) point-to-multipoint (P2MP) Label Switched Paths (LSPs) in Multi- Protocol Label Switching (MPLS) and Generalized MPLS (GMPLS) networks. The solution relies on RSVP-TE without requiring a multicast routing protocol in the Service Provider core. Protocol elements and procedures for this solution are described. There can be various applications for P2MP TE LSPs such as IP multicast. Specification of how such applications will use a P2MP TE LSP is outside the scope of this document. draft-ietf-mpls-rsvp-te-p2mp-01.txt [Page 1] Internet Draft draft-ietf-mpls-rsvp-te-p2mp-01.txt November 2004 Conventions used in this document The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in RFC-2119 [KEYWORDS]. Authors' Note Some of the text in the document needs further discussion between authors and feedback from MPLS WG. This has been pointed out when applicable. A change log and reviewed/updated text will be made available online. Table of Contents 1 Terminology............................................. 4 2 Introduction.............................................4 3 Mechanisms.............................................. 4 3.1 P2MP Tunnels............................................ 5 3.2 P2MP LSP Tunnels........................................ 5 3.3 Sub-Groups.............................................. 5 3.4 S2L Sub-LSPs............................................ 6 3.4.1 Representation of a S2L sub-LSP......................... 6 3.4.2 S2L Sub-LSPs and Path Messages.......................... 6 3.5 Explicit Routing........................................ 7 4 Path Message............................................ 9 4.1 Path Message Format..................................... 9 4.2 Path Message Processing................................. 10 4.2.1 Multiple Path Messages.................................. 11 4.2.2 Multiple S2L Sub-LSPs in One Path Message............... 12 4.2.3 Transit Fragmentation................................... 13 4.3 Grafting................................................ 14 4.3.1 Addition of S2L Sub-LSP................................. 14 5 Resv Message............................................ 14 5.1 Resv Message Format..................................... 14 5.2 Resv Message Processing................................. 15 5.2.1 Resv Message Throttling................................. 16 5.3 Record Routing.......................................... 17 5.3.1 RRO Processing.......................................... 17 6 Reservation Style....................................... 17 7 Path Tear Message....................................... 17 7.1 Path Tear Message Format................................ 17 7.2 Pruning................................................. 17 7.2.1 Explicit S2L Sub-LSP Teardown........................... 17 7.2.2 Implicit S2L Sub-LSP Teardown........................... 18 7.2.1 P2MP TE LSP Teardown.................................... 19 draft-ietf-mpls-rsvp-te-p2mp-01.txt [Page 2] Internet Draft draft-ietf-mpls-rsvp-te-p2mp-01.txt November 2004 8 Notify and ResvConf Messages............................ 20 9 Error Processing........................................ 20 9.1 PathErr Message Format.................................. 20 9.2 Handling of Failures at Branch LSRs..................... 21 10 Refresh Reduction....................................... 22 11 State Management........................................ 22 11.1 Incremental State Update................................ 22 11.2 Combining Multiple Path Messages........................ 23 12 Control of Branch Fate Sharing.......................... 24 13 Admin Status Change..................................... 24 14 Label Allocation on LANs with Multiple Downstream Nodes. 25 15 Make-Before-Break....................................... 25 15.1 P2MP Tree re-optimization............................... 25 15.2 Re-optimization of a subset of S2L sub-LSPs ............ 25 16 Fast Reroute............................................ 26 16.1 Facility Backpup........................................ 26 16.2 One to One Backup....................................... 26 17 Support for LSRs that are not P2MP Capable.............. 27 18 Reduction in Control Plane Processing with LSP Hierarchy 29 19 P2MP LSP Tunnel Remerging and Cross-Over................ 29 20 New and Updated Message Objects......................... 31 20.1 P2MP SESSION Object..................................... 31 20.2 P2MP LSP Tunnel SENDER_TEMPLATE Object.................. 32 20.2.1 P2MP LSP Tunnel IPv4 SENDER_TEMPLATE Object............. 33 20.2.2 P2MP LSP Tunnel IPv6 SENDER_TEMPLATE Object............. 33 20.3 S2L SUB-LSP Object...................................... 34 20.3.1 S2L IPv4 SUB-LSP Object................................. 34 20.3.2 S2L IPv6 SUB-LSP Object................................. 35 20.4 FILTER_SPEC Object...................................... 35 20.5 SUB EXPLICIT ROUTE Object (SERO)........................ 36 20.6 SUB RECORD ROUTE Object (SRRO).......................... 36 21 IANA Considerations..................................... 37 22 Security Considerations................................. 37 23 Acknowledgements........................................ 37 24 Example P2MP LSP Establishment ......................... 37 25 References.............................................. 39 26 Authors................................................. 40 27 Intellectual Property................................... 43 28 Full Copyright Statement................................ 43 29 Acknowledgement......................................... 44 draft-ietf-mpls-rsvp-te-p2mp-01.txt [Page 3] Internet Draft draft-ietf-mpls-rsvp-te-p2mp-01.txt November 2004 1. Terminology This document uses terminologies defined in [RFC3031], [RFC2205], [RFC3209], [RFC3473] and [P2MP-REQ]. In particular, this document uses the notation defined in [P2MP-REQ] for describing the components on a P2MP LSP between root, branches and leaves. 2. Introduction [RFC3209] defines a mechanism for setting up point-to-point (P2P) Traffic Engineered (TE) LSPs in MPLS networks. [RFC3473] defines extensions to [RFC3209] for setting up P2P TE LSPs in GMPLS networks. However these specifications do not provide a mechanism for building point-to-multipoint P2MP TE LSPs. This document defines extensions to RSVP-TE [RFC3209] and [RFC3473] protocol to support P2MP TE LSPs satisfying the set of requirements described in [P2MP-REQ]. This document relies on the semantics of RSVP that RSVP-TE inherits for building P2MP LSP Tunnels. A P2MP LSP Tunnel is comprised of multiple S2L sub-LSPs. These S2L sub-LSPs are set up between the ingress and egress LSRs and are appropriately combined by the branch LSRs using RSVP semantics to result in a P2MP TE LSP. One Path message may signal one or multiple S2L sub-LSPs. Hence the S2L sub- LSPs belonging to a P2MP LSP Tunnel can be signaled using one Path message or split across multiple Path messages. Path computation and P2MP application specific aspects are outside of the scope of this document. 3. Mechanism This document describes a solution that optimizes data replication by allowing non-ingress nodes in the network to be replication/branch nodes. A branch node is a LSR that is capable of replicating the incoming data on two or more outgoing interfaces. The solution uses RSVP-TE in the core of the network for setting up a P2MP TE LSP. The P2MP TE LSP is set up by associating multiple S2L TE sub-LSPs and relying on data replication at branch nodes. This is described further in the following sub-sections by describing P2MP tunnels and how they relate to S2L sub-LSPs. draft-ietf-mpls-rsvp-te-p2mp-01.txt [Page 4] Internet Draft draft-ietf-mpls-rsvp-te-p2mp-01.txt November 2004 3.1. P2MP Tunnels The specific aspect related to P2MP TE LSP is the action required at a branch node, where data replication occurs. Incoming labeled data is appropriately replicated to several outgoing interfaces which may have different labels. A P2MP TE tunnel comprises of one or more P2MP LSPs referred to as P2MP LSP tunnels. A P2MP TE Tunnel is identified by a P2MP SESSION object. This object contains an identifier of the P2MP session defined as a P2MP ID, a tunnel ID and an extended tunnel ID. The fields of a P2MP SESSION object are identical to those of the SESSION object defined in [RFC3209] except that the Tunnel Endpoint Address field is replaced by the P2MP Identifier (P2MP ID) field. The P2MP ID provides an identifier for the set of destinations of the P2MP TE Tunnel. The P2MP SESSION object is defined in section 20.1. 3.2. P2MP LSP Tunnel A P2MP LSP Tunnel is identified by the combination of the P2MP ID, Tunnel ID, and Extended Tunnel ID that are part of the P2MP SESSION object, and the tunnel sender address and LSP ID fields of the P2MP SENDER_TEMPLATE object. The new P2MP SENDER_TEMPLATE object is defined in section 20.2. 3.3. Sub-Groups As with all other RSVP controlled LSP Tunnels, P2MP LSP Tunnel state is managed using RSVP messages. While use of RSVP messages is the same, P2MP LSP Tunnel state differs from P2P LSP state in a number of ways. A notable difference is that a P2MP LSP Tunnel is comprised of multiple S2L Sub-LSPs As a result of this, it may not be possible to signal a P2MP LSP Tunnel in a single RSVP-TE Path/Resv message. It is also possible that such a signaling message can not fit into a single IP packet. It must also be possible to efficiently add and remove endpoints to and from P2MP TE LSPs. An additional issue is that P2MP LSP Tunnels must also handle the state "remerge" problem [P2MP-REQ]. These differences in P2MP state are addressed through the addition of a sub-group identifier (Sub-Group ID) and sub-group originator (Sub- Group Originator ID) to the SENDER_TEMPLATE and FILTER_SPEC objects. Taken together the Sub-Group ID and Sub-Group Originator ID are referred to as the Sub-Group fields. The Sub-Group fields, together with rest of the SENDER_TEMPLATE and SESSION objects, are used to represent a portion of a P2MP LSP Tunnel's state. The portion of P2MP LSP Tunnel state identified by draft-ietf-mpls-rsvp-te-p2mp-01.txt [Page 5] Internet Draft draft-ietf-mpls-rsvp-te-p2mp-01.txt November 2004 specific subgroup field values is referred to as a signaling sub- tree. It is important to note that the term "signaling sub-tree" refers only to signaling state and not data plane replication or branching. For example, it is possible for a node to "split" signaling state for a P2MP LSP Tunnel, but to not branch the data associated with the P2MP LSP Tunnel. Typical applications for generation and use of multiple subgroups are adding an egress and semantic fragmentation to ensure that a Path message remains within a single IP packet. 3.4. S2L Sub-LSPs A P2MP LSP Tunnel is constituted of one or more S2L sub-LSPs. 3.4.1. Representation of a S2L Sub-LSP A S2L sub-LSP exists within the context of a P2MP LSP Tunnel. Thus it is identified by the P2MP ID, Tunnel ID, and Extended Tunnel ID that are part of the P2MP SESSION, the tunnel sender address and LSP ID fields of the P2MP SENDER_TEMPLATE object, and the S2L sub-LSP destination address that is part of the S2L_SUB_LSP object. The S2L_SUB_LSP object is defined in section 20.3. Additionally, a sub-LSP ID contained in the S2L_SUB_LSP object may be used depending on further discussions about the make-before-break procedures described in section 14. An EXPLICIT_ROUTE Object (ERO) or SUB_EXPLICIT_ROUTE Object (SERO) is used to optionally specify the explicit route of a S2L sub-LSP. Each ERO or a SERO that is signaled corresponds to a particular S2L_SUB_LSP object. Details of explicit route encoding are specified in section 3.5. 3.4.2. S2L Sub-LSPs and Path Messages The mechanism in this document allows a P2MP LSP Tunnel to be signaled using one or more Path messages. Each Path message may signal one or more S2L sub-LSPs. Support for multiple Path messages is desirable as one Path message may not be large enough to fit all the S2L sub-LSPs; and they also allow separate manipulation of sub- trees of the P2MP LSP Tunnel. The reason for allowing a single Path message, to signal multiple S2L sub-LSPs, is to optimize the number of control messages needed to setup a P2MP LSP Tunnel. draft-ietf-mpls-rsvp-te-p2mp-01.txt [Page 6] Internet Draft draft-ietf-mpls-rsvp-te-p2mp-01.txt November 2004 3.5. Explicit Routing When a Path message signals a single S2L sub-LSP (that is, the Path message is only targeting a single leaf in the P2MP tree), the EXPLICIT_ROUTE object may encode the path to the egress LSR. The Path message also includes the S2L_SUB_LSP object for the S2L sub-LSP being signaled. The < [], > tuple represents the S2L sub-LSP. The absence of the ERO should be interpreted as requiring hop-by-hop routing for the sub-LSP based on the S2L sub-LSP destination address field of the S2L_SUB_LSP object. When a Path message signals multiple S2L sub-LSPs the path of the first S2L sub-LSP, to the egress LSR, is encoded in the ERO. The first S2L sub-LSP is the one that corresponds to the first S2L_SUB_LSP object in the Path message. The S2L sub-LSPs corresponding to the S2L_SUB_LSP objects that follow are termed as subsequent S2L sub-LSPs. One approach to encode the explicit route of a subsequent S2L sub-LSP is to include the path from the ingress to the egress of the S2L sub-LSP. However this implies potential repetition of hops that could be learned from the ERO or explicit routes of other S2L sub-LSPs. Explicit route compression using SEROs attempts to minimize such repetition and is described below. The path of each subsequent S2L sub-LSP is encoded in a SUB_EXPLICIT_ROUTE object (SERO). The format of the SERO is the same as an ERO (as defined in [RFC3209]). Each subsequent S2L sub-LSP is represented by tuples of the form [] . There is a one to one correspondence between a S2L_SUB_LSP object and a SERO. A SERO for a particular S2L sub-LSP includes only the path from a certain branch LSR to the egress LSR if the path to that branch LSR can be derived from the ERO or other SEROs. The absence of a SERO should be interpreted as requiring hop- by-hop routing for that S2L sub-LSP. Note that the destination address is carried in the S2L sub-LSP object. The encoding of the SERO and S2L sub-LSP object are described in detail in section 20. Explicit route compression is illustrated using the following figure. A | | B | | C----D----E | | | | | | draft-ietf-mpls-rsvp-te-p2mp-01.txt [Page 7] Internet Draft draft-ietf-mpls-rsvp-te-p2mp-01.txt November 2004 F G H-------I | |\ | | | \ | J K L M | | | | | | | | N O P Q--R Figure 1. Explicit Route Compression Figure 1. shows a P2MP LSP Tunnel with LSR A as the ingress LSR and six egress LSRs: (F, N, O, P, Q and R). When all the six S2L sub-LSPs are signaled in one Path message let us assume that the S2L sub-LSP to LSR F is the first S2L sub-LSP and the rest are subsequent S2L sub-LSPs. Following is one way for the ingress LSR A to encode the S2L sub-LSP explicit routes using compression: S2L sub-LSP-F: ERO = {B, E, D, C, F}, S2L_SUB_LSP Object-F S2L sub-LSP-N: SERO = {D, G, J, N}, S2L_SUB_LSP Object-N S2L sub-LSP-O: SERO = {E, H, K, O}, S2L_SUB_LSP Object-O S2L sub-LSP-P: SERO = {H, L, P}, S2L_SUB_LSP Object-P, S2L sub-LSP-Q: SERO = {H, I, M, Q}, S2L_SUB_LSP Object-Q, S2L sub-LSP-R: SERO = {Q, R}, S2L_SUB_LSP Object-R, After LSR E processes the incoming Path message from LSR B it sends a Path message to LSR D with the S2L sub-LSP explicit routes encoded as follows: S2L sub-LSP-F: ERO = {D, C, F}, S2L_SUB_LSP Object-F S2L sub-LSP-N: SERO = {D, G, J, N}, S2L_SUB_LSP Object-N LSR E also sends a Path message to LSR H and following is one way to encode the S2L sub-LSP explicit routes using compression: S2L sub-LSP-O: ERO = {H, K, O}, S2L_SUB_LSP Object-O S2L sub-LSP-P: SERO = {H, L, P}, S2L_SUB_LSP Object-P, S2L sub-LSP-Q: SERO = {H, I, M, Q}, S2L_SUB_LSP Object-Q, S2L sub-LSP-R: SERO = {Q, R}, S2L_SUB_LSP Object-R, After LSR H processes the incoming Path message from E it sends a Path message to LSR K, LSR L and LSR I. The encoding for the Path message to LSR K is as follows: S2L sub-LSP-O: ERO = {K, O}, S2L_SUB_LSP Object-O The encoding of the Path message sent by LSR H to LSR L is as follows: draft-ietf-mpls-rsvp-te-p2mp-01.txt [Page 8] Internet Draft draft-ietf-mpls-rsvp-te-p2mp-01.txt November 2004 S2L sub-LSP-P: ERO = {L, P}, S2L_SUB_LSP Object-P, Following is one way for LSR H to encode the S2L sub-LSP explicit routes in the Path message sent to LSR I: S2L sub-LSP-Q: ERO = {I, M, Q}, S2L_SUB_LSP Object-Q, S2L sub-LSP-R: SERO = {Q, R}, S2L_SUB_LSP Object-R, The explicit route encodings in the Path messages sent by LSRs D and Q are left as an exercise to the reader. This compression mechanism reduces the Path message size. It also reduces the processing that can result if explicit routes are encoded from ingress to egress for each S2L sub-LSP. No assumptions are placed on the ordering of the subsequent S2L sub-LSPs and hence on the ordering of the SEROs in the Path message. All LSRs need to process the ERO corresponding to the first S2L sub-LSP. A LSR needs to process a SERO for a subsequent S2L sub-LSP only if the first hop in the corresponding SERO is a local address of that LSR. The branch LSR that is the first hop of a SERO propagates the corresponding S2L sub-LSP downstream. 4. Path Message 4.1. Path Message Format This section describes modifications made to the Path message format as specified in [RFC3209] and [RFC3473]. The Path message is enhanced to signal one or more S2L sub-LSPs. This is done by including the S2L sub-LSP descriptor list in the Path message as shown below. ::= [ ] [ [ | ] ...] [ ] [ ] [ ] [ ... ] [ ] [ ] [ ] [ ... ] [S2L sub-LSP descriptor list] draft-ietf-mpls-rsvp-te-p2mp-01.txt [Page 9] Internet Draft draft-ietf-mpls-rsvp-te-p2mp-01.txt November 2004 Following is the format of the S2L sub-LSP descriptor list. ::= [ ] ::= [ ] Each LSR MUST use the common objects in the Path message and the S2L sub-LSP descriptors to process each S2L sub-LSP represented by the S2L sub-LSP object and the SUB-/EXPLICIT_ROUTE object combination. The first S2L_SUB_LSP object's explicit route is specified by the ERO. Explicit routes of subsequent S2L sub-LSPs are specified by the corresponding SERO. A SERO corresponds to the following S2L_SUB_LSP object. The RRO in the sender descriptor contains the hops traversed by the Path message and applies to all the S2L sub-LSPs signaled in the Path message. Path message processing is described in the next section. 4.2. Path Message Processing The ingress-LSR initiates the set up of a S2L sub-LSP to each egress- LSR that is the destination of the P2MP LSP Tunnel. Each S2L sub-LSP is associated with the same P2MP LSP Tunnel using common P2MP SESSION object and fields in the SENDER_TEMPLATE object. Hence it can be combined with other S2L sub-LSPs to form a P2MP LSP Tunnel. Another S2L sub-LSP belonging to the same instance of this S2L sub-LSP (i.e. the same P2MP LSP Tunnel) can share resources with this LSP. The session corresponding to the P2MP TE tunnel is determined based on the P2MP SESSION object. Each S2L sub- LSP is identified using the S2L_SUB_LSP object. Explicit routing for the S2L sub-LSPs is achieved using the ERO and SEROs. As mentioned earlier, it is possible to signal S2L sub-LSPs for a given P2MP LSP Tunnel in one or more Path messages. And a given Path message can contain one or more S2L sub-LSPs. 4.2.1. Multiple Path messages As described in section 3, {, } or {, } tuple is used to specify a S2L sub-LSP. Multiple Path messages can be used to signal a P2MP LSP Tunnel. Each Path message can signal one or more S2L sub-LSPs. If a Path message contains only one S2L sub-LSP, each LSR along the S2L sub-LSP follows [RFC3209] procedures for processing the Path message draft-ietf-mpls-rsvp-te-p2mp-01.txt [Page 10] Internet Draft draft-ietf-mpls-rsvp-te-p2mp-01.txt November 2004 besides the S2L SUB-LSP object processing described in this document. Processing of Path messages containing more than one S2L sub-LSP is described in Section 4.3. An ingress LSR may use multiple Path messages for signaling a P2MP LSP. This may be because a single Path message may not be large enough to signal the P2MP LSP Tunnel. Or it may be while adding leaves to the P2MP LSP Tunnel the new leaves are signaled in a new Path message. Or an ingress LSR MAY choose to break the P2MP tree into separate manageable S2L sub-trees. These trees share the same root and may share the trunk and certain branches. The scope of this management decomposition of P2MP trees is bounded by a single tree and multiple S2L sub-trees with a single leaf each. As defined in [P2MP-REQ], a P2MP LSP Tunnel must have consistent attributes across all portions of a tree. This implies that each Path message that is used to signal a P2MP LSP Tunnel is signaled using the same signaling attributes with the exception of the S2L sub-LSP information. The resulting S2L sub-LSPs from the different Path messages belonging to the same P2MP LSP Tunnel SHOULD share labels and resources where they share hops to prevent multiple copies of the data being sent. In certain cases a transit LSR may need to generate multiple Path messages to signal state corresponding to a single received Path message. For instance ERO expansion may result in an overflow of the resultant Path message. There are two cases occurring in such circumstances, either the message can be decomposed into multiple Path messages such that each of the message carries a subset of the incoming S2L sub-LSPs carried by the incoming message, or the message can not be decomposed such that each of the outgoing Path message fits its maximum size value. Multiple Path messages generated by a LSR that signal state for the same P2MP LSP are signaled with the same SESSION object and have the same in the SENDER_TEMPLATE object. In order to disambiguate these Path messages a tuple is introduced (also referred to as the Sub-Group field). Multiple Path messages generated by a LSR to signal state for the same P2MP LSP have the same Sub-Group Originator ID and have a different sub-Group ID. The Sub-Group Originator ID SHOULD be set to the TE Router ID of the LSR that originates the Path message. This is either the ingress LSR or a LSR which re-originates the Path message with its own Sub-Group Originator ID. Cases when a transit LSR may change the Sub-Group Originator ID of an incoming Path message are described below. The tuple is network-wide unique. The sub-Group ID space is specific to the Sub-Group Originator ID. Therefore the combination is network-wide unique. Also, a router that changes the Sub-Group Originator ID MUST use the same Sub-Group Originator ID on all Path messages for the same P2MP LSP Tunnel and SHOULD not vary the value during the life of the P2MP LSP Tunnel. Note: This version of the document assumes that these additional fields, i.e. , are part of the SENDER_TEMPLATE object. 4.2.2. Multiple S2L Sub-LSPs in one Path message The S2L sub-LSP descriptor list allows the signaling of one or more S2L sub-LSPs in one Path message. It is possible to signal multiple S2L sub-LSP objects and ERO/SERO combinations in a single Path message. Note that these objects are the ones that differentiate a S2L sub-LSP. Each LSR can use the common objects in the Path message and the S2L sub-LSP descriptors to process each S2L sub-LSP. All LSRs need to process the ERO corresponding to the first S2L sub- LSP when the ERO is present. If one or more SEROs are present an ERO MUST be present. The signaling information for the first S2L sub-LSP is propagated in a Path message by each LSR along the explicit route specified by the ERO. A LSR needs to process a S2L sub-LSP descriptor for a subsequent S2L sub-LSP only if the first hop in the corresponding SERO is a local address of that LSR. If this is not the case the S2L sub-LSP descriptor is included in the Path message sent to LSR that is the next hop to reach the first hop in the SERO. This next hop is determined by using the ERO or other SEROs that encode the path to the SERO's first hop. If this is the case and the LSR is also the egress the S2L sub-LSP descriptor is not propagated downstream. If this is the case and the LSR is not the egress the S2L sub-LSP descriptor is included in a Path message sent to the next-hop determined from the SERO. Hence a branch LSR only propagates the relevant S2L sub-LSP descriptors on each downstream link. A S2L sub- LSP descriptor that is propagated on a downstream link only contains those S2L sub-LSPs that are routed using that link. This processing may result in a subsequent S2L sub-LSP in an incoming Path message to become the first S2L sub-LSP in an outgoing Path message. Note that if one or more SEROs contains loose hops, expansion of such loose hops may result in overflowing the Path message size. Section 4.2.3 describes how signaling of the set of S2L sub-LSPs can be split in more than one Path message. The Record Route Object (RRO) contains the hops traversed by the Path message and applies to all the S2L sub-LSPs signaled in the Path message. A transit LSR appends its address in an incoming RRO and propagates it downstream. A branch LSR forms a new RRO for each of draft-ietf-mpls-rsvp-te-p2mp-01.txt [Page 12] Internet Draft draft-ietf-mpls-rsvp-te-p2mp-01.txt November 2004 the outgoing Path messages. Each such updated RRO is formed using the rules in [RFC3209]. If a LSR is unable to support a S2L sub-LSP setup, a PathErr message MUST be sent for the impacted S2L sub-LSP, and normal processing of the rest of the P2MP LSP Tunnel SHOULD continue. The default behavior is that the remainder of the LSP is not impacted (that is, all other branches are allowed to set up) and the failed branches are reported in PathErr messages in which the Path_State_Reomved flag MUST NOT be set. However, the ingress LSR may set a LSP Integrity flag (see section 21.3) to request that if there is a setup failure on any branch the entire LSP should fail to set up. 4.2.3. Transit Fragmentation In certain cases a transit LSR may need to generate multiple Path messages to signal state corresponding to a single received Path message. For instance ERO expansion may result in an overflow of the resultant Path message. It is desirable not to rely on IP fragmentation in this case. In order to achieve this, the multiple Path messages generated by the transit LSR, MUST be signaled with the Sub-Group Originator ID set to the TE Router ID of the transit LSR and a distinct sub-Group ID. Thus each distinct Path message that is generated by the transit LSR for the P2MP LSP Tunnel carries a distinct tuple. When multiple Path messages are used by an ingress or transit node, each Path message SHOULD be identical with the exception of the S2L sub-LSP related information (e.g., SERO), message and hop information (e.g., INTEGRITY, MESSAGE_ID and RSVP_HOP), and the SENDER_TEMPLATE objects. Except when performing a make-before-break operation, the tunnel sender address and LSP ID fields MUST be the same in each message, and for transit nodes, the same as the values in the Path message. As described above one case in which the Sub-Group Originator ID of a received Path message is changed is that of transit fragmentation. The Sub-Group Originator ID of a received Path message may also be changed in the outgoing Path message and set to that of the LSR originating the Path message based on a local policy. For instance a LSR may decide to always change the Sub-Group Originator ID while performing ERO expansion. The Sub-Group ID MUST not be changed if the Sub-Group Originator ID is not being changed. draft-ietf-mpls-rsvp-te-p2mp-01.txt [Page 13] Internet Draft draft-ietf-mpls-rsvp-te-p2mp-01.txt November 2004 4.3. Grafting The operation of adding egress LSR(s) to an existing P2MP LSP Tunnel is termed grafting. This operation allows egress nodes to join a P2MP LSP Tunnel at different points in time. 4.3.1. Addition of S2L Sub-LSPs There are two methods to add S2L sub-LSPs to a P2MP LSP Tunnel. The first is to add new S2L sub-LSPs to the P2MP LSP Tunnel by adding them to an existing Path message and refreshing the entire Path message. Path message processing described in section 4 results in adding these S2L sub-LSPs to the P2MP LSP Tunnel. Note that as a result of adding one or more S2L sub-LSPs to a Path message the ERO compression encoding may have to be recomputed. The second is to use incremental updates described in section 11.1. The egress LSRs can be added/removed by signaling only the impacted S2L sub-LSPs in a new Path message. Hence other S2L sub-LSPs do not have to be re-signaled. 5. Resv Message 5.1. Resv Message Format The Resv message follows the [RFC3209] and [RFC3473] format: ::= [ ] [ [ | ] ... ] [ ] [ ] [ ] [ ] [ ] [ ... ]