RFC 4875
Extensions to Resource Reservation Protocol - Traffic Engineering (RSVP-TE) for Point-to-Multipoint TE Label Switched Paths (LSPs)
6. Resv Message
6.1. Resv Message Format
The Resv message follows the [RFC3209] and [RFC3473] format: <Resv Message> ::= <Common Header> [ <INTEGRITY> ] [ [<MESSAGE_ID_ACK> | <MESSAGE_ID_NACK>] ... ] [ <MESSAGE_ID> ] <SESSION> <RSVP_HOP> <TIME_VALUES> [ <RESV_CONFIRM> ] [ <SCOPE> ] [ <NOTIFY_REQUEST> ] [ <ADMIN_STATUS> ] [ <POLICY_DATA> ... ] <STYLE> <flow descriptor list> <flow descriptor list> ::= <FF flow descriptor list> | <SE flow descriptor> <FF flow descriptor list> ::= <FF flow descriptor> | <FF flow descriptor list> <FF flow descriptor> <SE flow descriptor> ::= <FLOWSPEC> <SE filter spec list> <SE filter spec list> ::= <SE filter spec> | <SE filter spec list> <SE filter spec> The FF flow descriptor and SE filter spec are modified as follows to identify the S2L sub-LSPs that they correspond to: <FF flow descriptor> ::= [ <FLOWSPEC> ] <FILTER_SPEC> <LABEL> [ <RECORD_ROUTE> ] [ <S2L sub-LSP flow descriptor list> ] <SE filter spec> ::= <FILTER_SPEC> <LABEL> [ <RECORD_ROUTE> ] [ <S2L sub-LSP flow descriptor list> ] <S2L sub-LSP flow descriptor list> ::= <S2L sub-LSP flow descriptor> [ <S2L sub-LSP flow descriptor list> ] <S2L sub-LSP flow descriptor> ::= <S2L_SUB_LSP> [ <P2MP_SECONDARY_RECORD_ROUTE> ] FILTER_SPEC is defined in section 19.4.
The S2L sub-LSP flow descriptor has the same format as S2L sub-LSP descriptor in section 4.1 with the difference that a P2MP_SECONDARY_RECORD_ROUTE object is used in place of a P2MP SECONDARY_EXPLICIT_ROUTE object. The P2MP_SECONDARY_RECORD_ROUTE objects follow the same compression mechanism as the P2MP SECONDARY_EXPLICIT_ROUTE objects. Note that a Resv message can signal multiple S2L sub-LSPs that may belong to the same FILTER_SPEC object or different FILTER_SPEC objects. The same label SHOULD be allocated if the <Sender Address, LSP-ID> fields of the FILTER_SPEC object are the same. However different labels MUST be allocated if the <Sender Address, LSP-ID> of the FILTER_SPEC object is different, as that implies that the FILTER_SPEC refers to a different P2MP LSP.6.2. Resv Message Processing
The egress LSR MUST follow normal RSVP procedures while originating a Resv message. The format of Resv messages is as defined in section 6.1. As usual, the Resv message carries the label allocated by the egress LSR. A node upstream of the egress node MUST allocate its own label and pass it upstream in the Resv message. The node MAY combine multiple flow descriptors, from different Resv messages received from downstream, in one Resv message sent upstream. A Resv message MUST NOT be sent upstream until at least one Resv message has been received from a downstream neighbor. When the integrity bit is set in the LSP_REQUIRED_ATTRIBUTE object, Resv message MUST NOT be sent upstream until all Resv messages have been received from the downstream neighbors. Each Fixed-Filter (FF) flow descriptor or Shared-Explicit (SE) filter spec sent upstream in a Resv message includes an S2L sub-LSP descriptor list. Each such FF flow descriptor or SE filter spec for the same P2MP LSP (whether on one or multiple Resv messages) on the same Resv MUST be allocated the same label, and FF flow descriptors or SE filter specs SHOULD use the same label across multiple Resv messages. The node that sends the Resv message, for a P2MP LSP, upstream MUST associate the label assigned by this node with all the labels received from downstream Resv messages, for that P2MP LSP. Note that a transit node may become a replication point in the future when a branch is attached to it. Hence, this results in the setup of a P2MP LSP from the ingress LSR to the egress LSRs.
The ingress LSR may need to understand when all desired egresses have been reached. This is achieved using S2L_SUB_LSP objects. Each branch node MAY forward a single Resv message upstream for each received Resv message from a downstream receiver. Note that there may be a large number of Resv messages at and close to the ingress LSR for an LSP with many receivers. A branch LSR SHOULD combine Resv state from multiple receivers into a single Resv message to be sent upstream (see section 6.2.1). However, note that this may result in overflowing the Resv message, particularly as the number of receivers downstream of any branch LSR increases as the LSR is closer to the ingress LSR. Thus, a branch LSR MAY choose to send more than one Resv message upstream and partition the Resv state between the messages. When a transit node sets the Sub-Group Originator field in a Path message, it MUST replace the Sub-Group fields received in the FILTER_SPEC objects of any associated Resv messages with the value that it originally received in the Sub-Group fields of the Path message from the upstream neighbor. ResvErr message generation is unmodified. Nodes propagating a received ResvErr message MUST use the Sub-Group field values carried in the corresponding Resv message.6.2.1. Resv Message Throttling
A branch node may have to send a revised Resv message upstream whenever there is a change in a Resv message for an S2L sub-LSP received from one of the downstream neighbors. This can result in excessive Resv messages sent upstream, particularly when the S2L sub- LSPs are first established. In order to mitigate this situation, branch nodes can limit their transmission of Resv messages. Specifically, in the case where the only change being sent in a Resv message is in one or more P2MP_SECONDARY_RECORD_ROUTE objects (SRROs), the branch node SHOULD transmit the Resv message only after a delay time has passed since the transmission of the previous Resv message for the same session. This delayed Resv message SHOULD include SRROs for all branches. A suggested value for the delay time is thirty seconds, and delay times SHOULD generally be longer than 1 second. Specific mechanisms for Resv message throttling and delay timer settings are implementation dependent and are outside the scope of this document.
6.3. Route Recording
6.3.1. RRO Processing
A Resv message for a P2P LSP contains a recorded route if the ingress LSR requested route recording by including an RRO in the original Path message. The same rule is used during signaling of P2MP LSPs. That is, inclusion of an RRO in the Path message used to signal one or more S2L sub-LSPs triggers the inclusion of a recorded route for each sub-LSP in the Resv message. The recorded route of the first S2L sub-LSP is encoded in the RRO. Additional recorded routes for the subsequent S2L sub-LSPs are encoded in P2MP_SECONDARY_RECORD_ROUTE objects (SRROs). Their format is specified in section 19.5. Each S2L_SUB_LSP object in a Resv is associated with an RRO or SRRO. The first S2L_SUB_LSP object (for the first S2L sub-LSP) is associated with the RRO. Subsequent S2L_SUB_LSP objects (for subsequent S2L sub-LSPs) are each followed by an SRRO that contains the recorded route for that S2L sub-LSP from the leaf to a branch. The ingress node can then use the RRO and SRROs to determine the end-to-end path for each S2L sub-LSP.6.4. Reservation Style
Considerations about the reservation style in a Resv message apply as described in [RFC3209]. The reservation style in the Resv messages can be either FF or SE. All P2MP LSPs that belong to the same P2MP Tunnel MUST be signaled with the same reservation style. Irrespective of whether the reservation style is FF or SE, the S2L sub-LSPs that belong to the same P2MP LSP SHOULD share labels where they share hops. If the S2L sub-LSPs that belong to the same P2MP LSP share labels then they MUST share resources. If the reservation style is FF, then S2L sub-LSPs that belong to different P2MP LSPs MUST NOT share resources or labels. If the reservation style is SE, then S2L sub-LSPs that belong to different P2MP LSPs and the same P2MP tunnel SHOULD share resources where they share hops, but they MUST not share labels in packet environments.
7. PathTear Message
7.1. PathTear Message Format
The format of the PathTear message is as follows: <PathTear Message> ::= <Common Header> [ <INTEGRITY> ] [ [ <MESSAGE_ID_ACK> | <MESSAGE_ID_NACK> ... ] [ <MESSAGE_ID> ] <SESSION> <RSVP_HOP> [ <sender descriptor> ] [ <S2L sub-LSP descriptor list> ] <S2L sub-LSP descriptor list> ::= <S2L_SUB_LSP> [ <S2L sub-LSP descriptor list> ] The definition of <sender descriptor> is not changed by this document.7.2. Pruning
The operation of removing egress LSR(s) from an existing P2MP LSP is termed as pruning. This operation allows egress nodes to be removed from a P2MP LSP at different points in time. This section describes the mechanisms to perform pruning.7.2.1. Implicit S2L Sub-LSP Teardown
Implicit teardown uses standard RSVP message processing. Per standard RSVP processing, an S2L sub-LSP may be removed from a P2MP TE LSP by sending a modified message for the Path or Resv message that previously advertised the S2L sub-LSP. This message MUST list all S2L sub-LSPs that are not being removed. When using this approach, a node processing a message that removes an S2L sub-LSP from a P2MP TE LSP MUST ensure that the S2L sub-LSP is not included in any other Path state associated with session before interrupting the data path to that egress. All other message processing remains unchanged. When implicit teardown is used to delete one or more S2L sub-LSPs, by modifying a Path message, a transit LSR may have to generate a PathTear message downstream to delete one or more of these S2L sub- LSPs. This can happen if as a result of the implicit deletion of S2L sub-LSP(s) there are no remaining S2L sub-LSPs to send in the corresponding Path message downstream.
7.2.2. Explicit S2L Sub-LSP Teardown
Explicit S2L Sub-LSP teardown relies on generating a PathTear message for the corresponding Path message. The PathTear message is signaled with the SESSION and SENDER_TEMPLATE objects corresponding to the P2MP LSP and the <Sub-Group Originator ID, Sub-Group ID> tuple corresponding to the Path message. This approach SHOULD be used when all the egresses signaled by a Path message need to be removed from the P2MP LSP. Other S2L sub-LSPs, from other sub-groups signaled using other Path messages, are not affected by the PathTear. A transit LSR that propagates the PathTear message downstream MUST ensure that it sets the <Sub-Group Originator ID, Sub-Group ID> tuple in the PathTear message to the values used in the Path message that was used to set up the S2L sub-LSPs being torn down. The transit LSR may need to generate multiple PathTear messages for an incoming PathTear message if it had performed transit fragmentation for the corresponding incoming Path message. When a P2MP LSP is removed by the ingress, a PathTear message MUST be generated for each Path message used to signal the P2MP LSP.8. Notify and ResvConf Messages
8.1. Notify Messages
The Notify Request object and Notify message are described in [RFC3473]. Both object and message SHALL be supported for delivery of upstream and downstream notification. Processing not detailed in this section MUST comply to [RFC3473]. 1. Upstream Notification If a transit LSR sets the Sub-Group Originator ID in the SENDER_TEMPLATE object of a Path message to its own address, and the incoming Path message carries a Notify Request object, then this LSR MUST change the Notify node address in the Notify Request object to its own address in the Path message that it sends. If this LSR subsequently receives a corresponding Notify message from a downstream LSR, then it MUST: - send a Notify message upstream toward the Notify node address that the LSR received in the Path message.
- process the Sub-Group fields of the SENDER_TEMPLATE object on
the received Notify message, and modify their values, in the
Notify message that is forwarded, to match the Sub-Group field
values in the original Path message received from upstream.
The receiver of an (upstream) Notify message MUST identify the state
referenced in this message based on the SESSION and SENDER_TEMPLATE.
2. Downstream Notification
A transit LSR sets the Sub-Group Originator ID in the FILTER_SPEC
object(s) of a Resv message to the value that was received in the
corresponding Path message. If the incoming Resv message carries a
Notify Request object, then:
- If there is at least another incoming Resv message that carries
a Notify Request object, and the LSR merges these Resv messages
into a single Resv message that is sent upstream, the LSR MUST
set the Notify node address in the Notify Request object to its
Router ID.
- Else if the LSR sets the Sub-Group Originator ID (in the
outgoing Path message that corresponds to the received Resv
message) to its own address, the LSR MUST set the Notify node
address in the Notify Request object to its Router ID.
- Else the LSR MUST propagate the Notify Request object unchanged,
in the Resv message that it sends upstream.
If this LSR subsequently receives a corresponding Notify message from
an upstream LSR, then it MUST:
- process the Sub-Group fields of the FILTER_SPEC object in the
received Notify message, and modify their values, in the Notify
message that is forwarded, to match the Sub-Group field values
in the original Path message sent downstream by this LSR.
- send a Notify message downstream toward the Notify node address
that the LSR received in the Resv message.
The receiver of a (downstream) Notify message MUST identify the state
referenced in the message based on the SESSION and FILTER_SPEC
objects.
The consequence of these rules for a P2MP LSP is that an upstream
Notify message generated on a branch will result in a Notify being
delivered to the upstream Notify node address. The receiver of the
Notify message MUST NOT assume that the Notify message applies to all
downstream egresses, but MUST examine the information in the message to determine to which egresses the message applies. Downstream Notify messages MUST be replicated at branch LSRs according to the Notify Request objects received on Resv messages. Some downstream branches might not request Notify messages, but all that have requested Notify messages MUST receive them.8.2. ResvConf Messages
ResvConf messages are described in [RFC2205]. ResvConf processing in [RFC3473] and [RFC3209] is taken directly from [RFC2205]. An egress LSR MAY include a RESV_CONFIRM object that contains the egress LSR's address. The object and message SHALL be supported for the confirmation of receipt of the Resv message in P2MP TE LSPs. Processing not detailed in this section MUST comply to [RFC2205]. A transit LSR sets the Sub-Group Originator ID in the FILTER_SPEC object(s) of a Resv message to the value that was received in the corresponding Path message. If any of the incoming Resv messages corresponding to a single Path message carry a RESV_CONFIRM object, then the LSR MUST include a RESV_CONFIRM object in the corresponding Resv message that it sends upstream. If the Sub-Group Originator ID is its own address, then it MUST set the receiver address in the RESV_CONFIRM object to this address, else it MUST propagate the object unchanged. A transit LSR sets the Sub-Group Originator ID in the FILTER_SPEC object(s) of a Resv message to the value that was received in the corresponding Path message. If an incoming Resv message corresponding to a single Path message carries a RESV_CONFIRM object, then the LSR MUST include a RESV_CONFIRM object in the corresponding Resv message that it sends upstream and: - If there is at least another incoming Resv message that carries a RESV_CONFIRM object, and the LSR merges these Resv messages into a single Resv message that is sent upstream, the LSR MUST set the receiver address in the RESV_CONFIRM object to its Router ID. - If the LSR sets the Sub-Group Originator ID (in the outgoing Path message that corresponds to the received Resv message) to its own address, the LSR MUST set the receiver address in the RESV_CONFIRM object to its Router ID. - Else the LSR MUST propagate the RESV_CONFIRM object unchanged, in the Resv message that it sends upstream.
If this LSR subsequently receives a corresponding ResvConf message
from an upstream LSR, then it MUST:
- process the Sub-Group fields of the FILTER_SPEC object in the
received ResvConf message, and modify their values, in the
ResvConf message that is forwarded, to match the Sub-Group field
values in the original Path message sent downstream by this LSR.
- send a ResvConf message downstream toward the receiver address
that the LSR received in the RESV_CONFIRM object in the Resv
message.
The receiver of a ResvConf message MUST identify the state referenced
in this message based on the SESSION and FILTER_SPEC objects.
The consequence of these rules for a P2MP LSP is that a ResvConf
message generated at the ingress will result in a ResvConf message
being delivered to the branch and then to the receiver address in the
original RESV_CONFIRM object. The receiver of a ResvConf message
MUST NOT assume that the ResvConf message should be sent to all
downstream egresses, but it MUST replicate the message according to
the RESV_CONFIRM objects received in Resv messages. Some downstream
branches might not request ResvConf messages, and ResvConf messages
SHOULD NOT be sent on these branches. All downstream branches that
requested ResvConf messages MUST be sent such a message.
9. Refresh Reduction
The refresh reduction procedures described in [RFC2961] are equally
applicable to P2MP LSPs described in this document. Refresh
reduction applies to individual messages and the state they
install/maintain, and that continues to be the case for P2MP LSPs.
10. State Management
State signaled by a P2MP Path message is identified by a local
implementation using the <P2MP ID, Tunnel ID, Extended Tunnel ID>
tuple as part of the SESSION object and the <Tunnel Sender Address,
LSP ID, Sub-Group Originator ID, Sub-Group ID> tuple as part of the
SENDER_TEMPLATE object.
Additional information signaled in the Path/Resv message is part of
the state created by a local implementation. This includes PHOP/NHOP
and SENDER_TSPEC/FILTER_SPEC objects.
10.1. Incremental State Update
RSVP (as defined in [RFC2205] and as extended by RSVP-TE [RFC3209] and GMPLS [RFC3473]) uses the same basic approach to state communication and synchronization -- namely, full state is sent in each state advertisement message. Per [RFC2205], Path and Resv messages are idempotent. Also, [RFC2961] categorizes RSVP messages into two types (trigger and refresh messages) and improves RSVP message handling and scaling of state refreshes, but does not modify the full state advertisement nature of Path and Resv messages. The full state advertisement nature of Path and Resv messages has many benefits, but also has some drawbacks. One notable drawback is when an incremental modification is being made to a previously advertised state. In this case, there is the message overhead of sending the full state and the cost of processing it. It is desirable to overcome this drawback and add/delete S2L sub-LSPs to/from a P2MP LSP by incrementally updating the existing state. It is possible to use the procedures described in this document to allow S2L sub-LSPs to be incrementally added to or deleted from the P2MP LSP by allowing a Path or a PathTear message to incrementally change the existing P2MP LSP Path state. As described in section 5.2, multiple Path messages can be used to signal a P2MP LSP. The Path messages are distinguished by different <Sub-Group Originator ID, Sub-Group ID> tuples in the SENDER_TEMPLATE object. In order to perform incremental S2L sub-LSP state addition, a separate Path message with a new Sub-Group ID is used to add the new S2L sub-LSPs, by the ingress LSR. The Sub-Group Originator ID MUST be set to the TE Router ID [RFC3477] of the node that sets the Sub-Group ID. This maintains the idempotent nature of RSVP Path messages, avoids keeping track of individual S2L sub-LSP state expiration, and provides the ability to perform incremental P2MP LSP state updates.10.2. Combining Multiple Path Messages
There is a tradeoff between the number of Path messages used by the ingress to maintain the P2MP LSP and the processing imposed by full state messages when adding S2L sub-LSPs to an existing Path message. It is possible to combine S2L sub-LSPs previously advertised in different Path messages in a single Path message in order to reduce the number of Path messages needed to maintain the P2MP LSP. This can also be done by a transit node that performed fragmentation and that at a later point is able to combine multiple Path messages that it generated into a single Path message. This may happen when one or more S2L sub-LSPs are pruned from the existing Path states.
The new Path message is signaled by the node that is combining multiple Path messages with all the S2L sub-LSPs that are being combined in a single Path message. This Path message MAY contain new Sub-Group ID field values. When a new Path and Resv message that is signaled for an existing S2L sub-LSP is received by a transit LSR, state including the new instance of the S2L sub-LSP is created. The S2L sub-LSP SHOULD continue to be advertised in both the old and new Path messages until a Resv message listing the S2L sub-LSP and corresponding to the new Path message is received by the combining node. Hence, until this point, state for the S2L sub-LSP SHOULD be maintained as part of the Path state for both the old and the new Path message (see section 3.1.3 of [RFC2205]). At that point the S2L sub-LSP SHOULD be deleted from the old Path state using the procedures of section 7. A Path message with a Sub-Group_ID(n) may signal a set of S2L sub- LSPs that belong partially or entirely to an already existing Sub- Group_ID(i), or a strictly non-overlapping new set of S2L sub-LSPs. A newly received Path message that matches SESSION object and Sender Tunnel Address, LSP ID, Sub-Group Originator ID> with existing Path state carrying the same or different Sub-Group_ID, referred to Sub- Group_ID(n) is processed as follows: 1) If Sub-Group_ID(i) = Sub-Group_ID(n), then S2L Sub-LSPs that are in both Sub-Group_ID(i) and Sub-Group_ID(n) are refreshed. New S2L Sub-LSPs are added to Sub-Group_ID(i) Path state and S2L Sub- LSPs that are in Sub-Group_ID(i) but not in Sub-Group_ID(n) are deleted from the Sub-Group_ID(i) Path state. 2) If Sub-Group_ID(i) != Sub-Group_ID(n), then a new Sub-Group_ID(n) Path state is created for S2L Sub-LSPs signaled by Sub- Group_ID(n). S2L Sub-LSPs in existing Sub-Group_IDs(i) Path state (that are or are not in the newly received Path message Sub- Group_ID(n)) are left unmodified (see above).11. Error Processing
PathErr and ResvErr messages are processed as per RSVP-TE procedures. Note that an LSR, on receiving a PathErr/ResvErr message for a particular S2L sub-LSP, changes the state only for that S2L sub-LSP. Hence other S2L sub-LSPs are not impacted. If the ingress node requests 'LSP integrity', an error reported on a branch of a P2MP TE LSP for a particular S2L sub-LSP may change the state of any other S2L sub-LSP of the same P2MP TE LSP. This is explained further in section 11.3.
11.1. PathErr Messages
The PathErr message will include one or more S2L_SUB_LSP objects. The resulting modified format for a PathErr message is: <PathErr Message> ::= <Common Header> [ <INTEGRITY> ] [ [<MESSAGE_ID_ACK> | <MESSAGE_ID_NACK>] ... ] [ <MESSAGE_ID> ] <SESSION> <ERROR_SPEC> [ <ACCEPTABLE_LABEL_SET> ... ] [ <POLICY_DATA> ... ] <sender descriptor> [ <S2L sub-LSP descriptor list> ] PathErr message generation is unmodified, but nodes that set the Sub-Group Originator field and propagate a received PathErr message upstream MUST replace the Sub-Group fields received in the PathErr message with the value that was received in the Sub-Group fields of the Path message from the upstream neighbor. Note the receiver of a PathErr message is able to identify the errored outgoing Path message, and outgoing interface, based on the Sub-Group fields received in the PathErr message. The S2L sub-LSP descriptor list is defined in section 5.1.11.2. ResvErr Messages
The ResvErr message will include one or more S2L_SUB_LSP objects. The resulting modified format for a ResvErr Message is: <ResvErr Message> ::= <Common Header> [ <INTEGRITY> ] [ [<MESSAGE_ID_ACK> | <MESSAGE_ID_NACK>] ... ] [ <MESSAGE_ID> ] <SESSION> <RSVP_HOP> <ERROR_SPEC> [ <SCOPE> ] [ <ACCEPTABLE_LABEL_SET> ... ] [ <POLICY_DATA> ... ] <STYLE> <flow descriptor list> ResvErr message generation is unmodified, but nodes that set the Sub-Group Originator field and propagate a received ResvErr message downstream MUST replace the Sub-Group fields received in the ResvErr message with the value that was set in the Sub-Group fields of the Path message sent to the downstream neighbor. Note the receiver of a ResvErr message is able to identify the errored outgoing Resv
message, and outgoing interface, based on the Sub-Group fields received in the ResvErr message. The flow descriptor list is defined in section 6.1.11.3. Branch Failure Handling
During setup and during normal operation, PathErr messages may be received at a branch node. In all cases, a received PathErr message is first processed per standard processing rules. That is, the PathErr message is sent hop-by-hop to the ingress/branch LSR for that Path message. Intermediate nodes until this ingress/branch LSR MAY inspect this message but take no action upon it. The behavior of a branch LSR that generates a PathErr message is under the control of the ingress LSR. The default behavior is that the PathErr message does not have the Path_State_Removed flag set. However, if the ingress LSR has set the LSP integrity flag on the Path message (see LSP_REQUIRED_ATTRIBUTEs object in section 5.2.4), and if the Path_State_Removed flag is supported, the LSR generating a PathErr to report the failure of a branch of the P2MP LSP SHOULD set the Path_State_Removed flag. A branch LSR that receives a PathErr message during LSP setup with the Path_State_Removed flag set MUST act according to the wishes of the ingress LSR. The default behavior is that the branch LSR clears the Path_State_Removed flag on the PathErr and sends it further upstream. It does not tear any other branches of the LSP. However, if the LSP integrity flag is set on the Path message, the branch LSR MUST send PathTear on all other downstream branches and send the PathErr message upstream with the Path_State_Removed flag set. A branch LSR that receives a PathErr message with the Path_State_Removed flag clear MUST act according to the wishes of the ingress LSR. The default behavior is that the branch LSR forwards the PathErr upstream and takes no further action. However, if the LSP integrity flag is set on the Path message, the branch LSR MUST send PathTear on all downstream branches and send the PathErr upstream with the Path_State_Removed flag set (per [RFC3473]). In all cases, the PathErr message forwarded by a branch LSR MUST contain the S2L sub-LSP identification and explicit routes of all branches that are reported by received PathErr messages and all branches that are explicitly torn by the branch LSR.
12. Admin Status Change
A branch node that receives an ADMIN_STATUS object processes it normally and also relays the ADMIN_STATUS object in a Path on every branch. All Path messages may be concurrently sent to the downstream neighbors. Downstream nodes process the change in the ADMIN_STATUS object per [RFC3473], including generation of Resv messages. When the last received upstream ADMIN_STATUS object had the R bit set, branch nodes wait for a Resv message with a matching ADMIN_STATUS object to be received (or a corresponding PathErr or ResvTear message) on all branches before relaying a corresponding Resv message upstream.13. Label Allocation on LANs with Multiple Downstream Nodes
A branch LSR of a P2MP LSP on an Ethernet LAN segment SHOULD send one copy of the data traffic per downstream LSR connected on that LAN for that P2MP LSP. Procedures for preventing MPLS labeled traffic replication in such a case is beyond the scope of this document.14. P2MP LSP and Sub-LSP Re-Optimization
It is possible to change the path used by P2MP LSPs to reach the destinations of the P2MP tunnel. There are two methods that can be used to accomplish this. The first is make-before-break, defined in [RFC3209], and the second uses the sub-groups defined above.14.1. Make-before-Break
In this case, all the S2L sub-LSPs are signaled with a different LSP ID by the ingress LSR and follow the make-before-break procedure defined in [RFC3209]. Thus, a new P2MP LSP is established. Each S2L sub-LSP is signaled with a different LSP ID, corresponding to the new P2MP LSP. After moving traffic to the new P2MP LSP, the ingress can tear down the old P2MP LSP. This procedure can be used to re- optimize the path of the entire P2MP LSP or the paths to a subset of the destinations of the P2MP LSP. When modifying just a portion of the P2MP LSP, this approach requires the entire P2MP LSP to be re- signaled.14.2. Sub-Group-Based Re-Optimization
Any node may initiate re-optimization of a set of S2L sub-LSPs by using incremental state update and then, optionally, combining multiple path messages.
To alter the path taken by a particular set of S2L sub-LSPs, the node initiating the path change initiates one or more separate Path messages for the same P2MP LSP, each with a new sub-Group ID. The generation of these Path messages, each with one or more S2L sub- LSPs, follows procedures in section 5.2. As is the case in section 10.2, a particular egress continues to be advertised in both the old and new Path messages until a Resv message listing the egress and corresponding to the new Path message is received by the re- optimizing node. At that point, the egress SHOULD be deleted from the old Path state using the procedures of section 7. Sub-tree re- optimization is then completed. Sub-Group-based re-optimization may result in transient data duplication as the new Path messages for a set of S2L sub-LSPs may transit one or more nodes with the old Path state for the same set of S2L sub-LSPs. As is always the case, a node may choose to combine multiple path messages as described in section 10.2.15. Fast Reroute
[RFC4090] extensions can be used to perform fast reroute for the mechanism described in this document when applied within packet networks. GMPLS introduces other protection techniques that can be applied to packet and non-packet environments [RFC4873], but which are not discussed further in this document. This section only applies to LSRs that support [RFC4090]. This section uses terminology defined in [RFC4090], and fast reroute procedures defined in [RFC4090] MUST be followed unless specified below. The head-end and transit LSRs MUST follow the SESSION_ATTRIBUTE and FAST_REROUTE object processing as specified in [RFC4090] for each Path message and S2L sub-LSP of a P2MP LSP. Each S2L sub-LSP of a P2MP LSP MUST have the same protection characteristics. The RRO processing MUST apply to SRRO as well unless modified below. The sections that follow describe how fast reroute may be applied to P2MP MPLS TE LSPs in all of the principal operational scenarios. This document does not describe the detailed processing steps for every imaginable usage case, and they may be described in future documents, as needed.
15.1. Facility Backup
Facility backup can be used for link or node protection of LSRs on the path of a P2MP LSP. The downstream labels MUST be learned by the Point of Local Repair (PLR), as specified in [RFC4090], from the label corresponding to the S2L sub-LSP in the RESV message. Processing of SEROs signaled in a backup tunnel MUST follow backup tunnel ERO processing described in [RFC4090].15.1.1. Link Protection
If link protection is desired, a bypass tunnel MUST be used to protect the link between the PLR and next-hop. Thus all S2L sub-LSPs that use the link SHOULD be protected in the event of link failure. Note that all such S2L sub-LSPs belonging to a particular instance of a P2MP tunnel SHOULD share the same outgoing label on the link between the PLR and the next-hop as per section 5.2.1. This is the P2MP LSP label on the link. Label stacking is used to send data for each P2MP LSP into the bypass tunnel. The inner label is the P2MP LSP label allocated by the next-hop. During failure, Path messages for each S2L sub-LSP that is affected, MUST be sent to the Merge Point (MP) by the PLR. It is RECOMMENDED that the PLR uses the sender template-specific method to identify these Path messages. Hence, the PLR will set the source address in the sender template to a local PLR address. The MP MUST use the LSP-ID to identify the corresponding S2L sub- LSPs. The MP MUST NOT use the <Sub-Group Originator ID, Sub-Group ID> tuple while identifying the corresponding S2L sub-LSPs. In order to further process an S2L sub-LSP the MP MUST determine the protected S2L sub-LSP using the LSP-ID and the S2L_SUB_LSP object.15.1.2. Node Protection
If node protection is desired the PLR SHOULD use one or more P2P bypass tunnels to protect the set of S2L sub-LSPs that transit the protected node. Each of these P2P bypass tunnels MUST intersect the path of the S2L sub-LSPs that they protect on an LSR that is downstream from the protected node. This constrains the set of S2L sub-LSPs being backed- up via that bypass tunnel to those S2L sub- LSPs that pass through a common downstream MP. This MP is the destination of the bypass tunnel. When the PLR forwards incoming data for a P2MP LSP into the bypass tunnel, the outer label is the bypass tunnel label and the inner label is the label allocated by the MP to the set of S2L sub-LSPs belonging to that P2MP LSP.
After detecting failure of the protected node the PLR MUST send one or more Path messages for all protected S2L sub-LSPs to the MP of the protected S2L sub-LSP. It is RECOMMENDED that the PLR use the sender template specific method to identify these Path messages. Hence the PLR will set the source address in the sender template to a local PLR address. The MP MUST use the LSP-ID to identify the corresponding S2L sub-LSPs. The MP MUST NOT use the <Sub-Group Originator ID, Sub-Group ID> tuple while identifying the corresponding S2L sub-LSPs because the Sub-Group Originator ID might be changed by some LSR that is bypassed by the bypass tunnel. In order to further process an S2L sub-LSP the MP MUST determine the protected S2L sub-LSP using the LSP-ID and the S2L_SUB_LSP object. Note that node protection MAY require the PLR to be branch capable in the data plane, as multiple bypass tunnels may be required to back up the set of S2L sub-LSPs passing through the protected node. If the PLR is not branch capable, the node protection mechanism described here is applicable to only those cases where all the S2L sub-LSPs passing through the protected node also pass through a single MP that is downstream from the protected node. A PLR MUST set the Node protection flag in the RRO/SRRO as specified in [RFC4090]. If a PLR is not branch capable, and one or more S2L sub-LSPs are added to a P2MP tree, and these S2L sub-LSPs do not transit the existing MP downstream of the protected node, then the PLR MUST reset this flag. It is to be noted that procedures in this section require P2P bypass tunnels. Procedures for using P2MP bypass tunnels are for further study.15.2. One-to-One Backup
One-to-one backup, as described in [RFC4090], can be used to protect a particular S2L sub-LSP against link and next-hop failure. Protection may be used for one or more S2L sub-LSPs between the PLR and the next-hop. All the S2L sub-LSPs corresponding to the same instance of the P2MP tunnel between the PLR and the next-hop SHOULD share the same P2MP LSP label, as per section 5.2.1. All such S2L sub-LSPs belonging to a P2MP LSP MUST be protected. The backup S2L sub-LSPs may traverse different next-hops at the PLR. Thus, the set of outgoing labels and next-hops for a P2MP LSP, at the PLR, may change once protection is triggered. Consider a P2MP LSP that is using a single next-hop and label between the PLR and the next-hop of the PLR. This may no longer be the case once protection is triggered. This MAY require a PLR to be branch capable in the data plane. If the PLR is not branch capable, the one-to-one backup mechanisms described here are only applicable to those cases where all the backup S2L sub-LSPs pass through the same next-hop downstream
of the PLR. Procedures for one-to-one backup when a PLR is not branch capable and when all the backup S2L sub-LSPs do not pass through the same downstream next-hop are for further study. It is recommended that the path-specific method be used to identify a backup S2L sub-LSP. Hence, the DETOUR object SHOULD be inserted in the backup Path message. A backup S2L sub-LSP MUST be treated as belonging to a different P2MP tunnel instance than the one specified by the LSP-ID. Furthermore multiple backup S2L sub-LSPs MUST be treated as part of the same P2MP tunnel instance if they have the same LSP-ID and the same DETOUR objects. Note that, as specified in section 4, S2L sub-LSPs between different P2MP tunnel instances use different labels. If there is only one S2L sub-LSP in the Path message, the DETOUR object applies to that sub-LSP. If there are multiple S2L sub-LSPs in the Path message, the DETOUR object applies to all the S2L sub- LSPs.16. Support for LSRs That Are Not P2MP Capable
It may be that some LSRs in a network are capable of processing the P2MP extensions described in this document, but do not support P2MP branching in the data plane. If such an LSR is requested to become a branch LSR by a received Path message, it MUST respond with a PathErr message carrying the Error Code "Routing Error" and Error Value "Unable to Branch". It is also conceivable that some LSRs, in a network deploying P2MP capability, may not support the extensions described in this document. If a Path message for the establishment of a P2MP LSP reaches such an LSR, it will reject it with a PathErr because it will not recognize the C-Type of the P2MP SESSION object. LSRs that do not support the P2MP extensions in this document may be included as transit LSRs by the use of LSP stitching [LSP-STITCH] and LSP hierarchy [RFC4206]. Note that LSRs that are required to play any other role in the network (ingress, branch or egress) MUST support the extensions defined in this document. The use of LSP stitching and LSP hierarchy [RFC4206] allows P2MP LSPs to be built in such an environment. A P2P LSP segment is signaled from the last P2MP-capable hop that is upstream of a legacy LSR to the first P2MP-capable hop that is downstream of it. This assumes that intermediate legacy LSRs are transit LSRs: they cannot act as P2MP branch points. Transit LSRs along this LSP segment do not process control plane messages associated with the P2MP LSP. Furthermore, these transit LSRs also do not need to have P2MP data
plane capabilities as they only need to process data belonging to the P2P LSP segment. Hence, these transit LSRs do not need to support P2MP MPLS. This P2P LSP segment is stitched to the incoming P2MP LSP. After the P2P LSP segment is established, the P2MP Path message is sent to the next P2MP-capable LSR as a directed Path message. The next P2MP-capable LSR stitches the P2P LSP segment to the outgoing P2MP LSP. In packet networks, the S2L sub-LSPs may be nested inside the outer P2P LSP. Hence, label stacking can be used to enable use of the same LSP segment for multiple P2MP LSPs. Stitching and nesting considerations and procedures are described further in [LSP-STITCH] and [RFC4206]. There maybe overhead for an operator to configure the P2P LSP segments in advance, when it is desired to support legacy LSRs. It may be desirable to do this dynamically. The ingress can use IGP extensions to determine P2MP-capable LSRs [TE-NODE-CAP]. It can use this information to compute S2L sub-LSP paths such that they avoid legacy non-P2MP-capable LSRs. The explicit route object of an S2L sub-LSP path may contain loose hops if there are legacy LSRs along the path. The corresponding explicit route contains a list of objects up to the P2MP-capable LSR that is adjacent to a legacy LSR followed by a loose object with the address of the next P2MP-capable LSR. The P2MP-capable LSR expands the loose hop using its Traffic Engineering Database (TED). When doing this it determines that the loose hop expansion requires a P2P LSP to tunnel through the legacy LSR. If such a P2P LSP exists, it uses that P2P LSP. Else it establishes the P2P LSP. The P2MP Path message is sent to the next P2MP-capable LSR using non-adjacent signaling. The P2MP-capable LSR that initiates the non-adjacent signaling message to the next P2MP-capable LSR may have to employ a fast detection mechanism (such as [BFD] or [BFD-MPLS]) to the next P2MP- capable LSR. This may be needed for the directed Path message head- end to use node protection fast reroute when the protected node is the directed Path message tail. Note that legacy LSRs along a P2P LSP segment cannot perform node protection of the tail of the P2P LSP segment.17. Reduction in Control Plane Processing with LSP Hierarchy
It is possible to take advantage of LSP hierarchy [RFC4206] while setting up P2MP LSP, as described in the previous section, to reduce control plane processing along transit LSRs that are P2MP capable. This is applicable only in environments where LSP hierarchy can be used. Transit LSRs along a P2P LSP segment, being used by a P2MP
LSP, do not process control plane messages associated with the P2MP LSP. In fact, they are not aware of these messages as they are tunneled over the P2P LSP segment. This reduces the amount of control plane processing required on these transit LSRs. Note that the P2P LSPs can be set up dynamically as described in the previous section or preconfigured. For example, in Figure 2 in section 24, PE1 can set up a P2P LSP to P1 and use that as a LSP segment. The Path messages for PE3 and PE4 can now be tunneled over the LSP segment. Thus, P3 is not aware of the P2MP LSP and does not process the P2MP control messages.
(page 35 continued on part 3)
