PCEP Extension for Native IP Network

Generally, Multiprotocol Label Switching Traffic Engineering (MPLS-TE) requires the corresponding network devices support Multiprotocol Label Switching (MPLS) or Resource ReSerVation Protocol (RSVP)/Label Distribution Protocol (LDP) technologies to assure the End-to-End (E2E) traffic performance. But in native IP network scenarios described in , there will be no such signaling protocol to synchronize the action among different network devices. It is necessary to use the central control mode that described in to correlate the forwarding behavior among different network devices. describes the architecture and solution philosophy for the E2E traffic assurance in Native IP network via multi Border Gateway Protocol (BGP) session based solution. This document describes the corresponding Path Computation Element Communication Protocol (PCEP) extensions to transfer the key information about BGP peer info, peer prefix advertisement and the explicit peer route on on-path routers.

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 when, and only when, they appear in all capitals, as shown here.

This document uses the following terms defined in : PCE, PCEP The following terms are defined in this document: CCDR: Central Control Dynamic Routing E2E: End to End BPI: BGP Peer Info EPR: Explicit Peer Route PPA: Peer Prefix Advertisement QoS: Quality of Service

During the PCEP Initialization Phase, PCEP Speakers (PCE or PCC) advertise their support of Native IP extensions. This document defines a new Path Setup Type (PST) for Native-IP, as follows: PST = TBD1: Path is a Native IP path as per . A PCEP speaker MUST indicate its support of the function described in this document by sending a PATH-SETUP-TYPE-CAPABILITY TLV in the OPEN object with this new PST included in the PST list. defined the PCECC-CAPABILITY sub-TLV to exchange information about their PCECC capability. A new flag is defined in PCECC-CAPABILITY sub-TLV for Native IP: N (NATIVE-IP-TE-CAPABILITY - 1 bit - TBD2): If set to 1 by a PCEP speaker, it indicates that the PCEP speaker is capable for TE in Native IP network as specified in this document. The flag MUST be set by both the PCC and PCE in order to support this extension. If a PCEP speaker receives the PATH-SETUP-TYPE-CAPABILITY TLV with the newly defined path setup type, but without the N bit set in PCECC-CAPABILITY sub-TLV, it MUST: Send a PCErr message with Error-Type=10(Reception of an invalid object) and Error-Value TBD3(PCECC NATIVE-IP-TE-CAPABILITY bit is not set). Terminate the PCEP session

PCECC Native IP TE solution utilizing the existing PCE LSP Initiate Request message(PCInitiate), and PCE Report message(PCRpt) to accomplish the multi BGP sessions establishment, E2E TE path deployment, and route prefixes advertisement among different BGP sessions. A new PST for Native-IP is used to indicate the path setup based on TE in Native IP networks. The extended PCInitiate message described in is used to download or cleanup central controller's instructions (CCIs). specifies an object called CCI for the encoding of central controller's instructions. This document specifies a new CCI object-type for Native IP. The PCEP messages are extended in this document to handle the PCECC operations for Native IP. Three new PCEP Objects (BGP Peer Info (BPI) Object, Explicit Peer Route (EPR) Object and Peer Prefix Advertisement (PPA) Object) are defined in this document. Refer to for detail object definitions.

The PCInitiate Message defined in and extended in is further extended to support Native-IP CCI. The format of the extended PCInitiate message is as follows:

::=

Where: is defined in [RFC5440] ::= [] ::= (| | ) ::= (| (||)) ::= [] Where: is as per [RFC9050] and are as per [RFC8281]. The LSP and SRP objects are defined in [RFC8231]. ]]>

When PCInitiate message is used to create Native IP instructions, the SRP, LSP and CCI objects MUST be present. The error handling for missing SRP, LSP or CCI object is as per . Further only one BPI, EPR, or PPA object MUST be present. The PLSP-ID within the LSP object should be set by PCC uniquely according to the Symbolic Path Name TLV that included in the CCI object. The Symbolic Path Name is used by the PCE/PCC to identify uniquely the E2E native IP TE path. If none of them are present, the receiving PCC MUST send a PCErr message with Error-type=6 (Mandatory Object missing) and Error-value=TBD4 (Native IP object missing). If there are more than one of BPI, EPR or PPA object are presented, the receiving PCC MUST send a PCErr message with Error-type=19(Invalid Operation) and Error-value=TBD5(Only one BPI, EPR or PPA object can be included in this message). To cleanup the existing Native IP instructions, the SRP object must set the R (remove) bit.

The PCRpt message is used to acknowledge the Native-IP instructions received from the central controller (PCE). The format of the PCRpt message is as follows:

::=

Where: ::= [] ::= (| ) ::= []

::= [] (| (||)) Where: is as per [RFC8231] and the LSP and SRP object are also defined in [RFC8231]. ]]>

The error handling for missing CCI object is as per . Further only one BPI, EPR, or PPA object MUST be present. If none of them are present, the receiving PCE MUST send a PCErr message with Error-type=6 (Mandatory Object missing) and Error-value=TBD4 (Native IP object missing). If there are more than one of BPI, EPR or PPA object are presented, the receiving PCE MUST send a PCErr message with Error-type=19(Invalid Operation) and Error-value=TBD5(Only one BPI, EPR or PPA object can be included in this message).

The detail procedures for the TE in native IP environment are described in the following sections.

The PCInitiate and PCRpt message pair is used to configure the parameters for a BGP peer session. This pair of PCEP messages is exchanged between a PCE and each BGP peer which needs to be configured. After the BGP peer session has been configured via this pair of PCEP messages, the BGP session establishes and operates in a normal fashion. The BGP peers can be configured for EBGP peers or IBGP peers. For IBGP connection topologies, the Route Reflector(RR) is required. The PCInitiate message should be sent to PCC which acts as BGP router and/or RR. The RR topology for a single AS is shown in Figure 1. The BGP routers R1, R3, and R7 are within a single AS. R1 and R7 are BGP RR clients, and R3 is a RR. The PCInitiate message should be sent only the BGP routers that need to be newly configured R1 (via M1 message), R3 (via M2 & M3 message), and R7 (via M4 message). PCInitiate message creates an auto-configuration function for these BGP peers by providing the indicated Peer AS and the Local/Peer IP Address. When PCC receives the BPI and CCI object (with the R bit set to 0 in SRP object) in PCInitiate message, the PCC should try to establish the BGP session with the indicated Peer AS and Local/Peer IP address. When PCC creates successfully the BGP session that is indicated by the associated information, it should report the result via the PCRpt messages, with BPI object and the corresponding SRP and CCI object included. When PCC receives this message with the R bit set to 1 in SRP object in PCInitiate message, the PCC should clear the BGP session that indicated by the BPI object. When PCC clears successfully the specified BGP session, it should report the result via the PCRpt message, with the BPI object included, and the corresponding SRP and CCI object.

The message number, message peers, message type and message key parameters in the above figures are shown in below table:

If the PCC cannot establish the BGP session that required by this message, it should report the error values via PCErr message with the newly defined error type: 1) Error-type=TBD6 and Error-value=TBD7, Peer AS not match when the received Peer AS is not the same the local AS, or 2) Error-type=TBD6 and Error-Value=TBD8, Peer IP can't be reached when there is no route to the Peer IP address), which is indicated in If the Local IP Address or Peer IP Address within BPI object is used in other existing BGP sessions, the PCC should report such error situation via PCErr message with: 3) Err-type=TBD6 and Error-value=TBD9, Local IP is in use, or 4) Err-type=TBD6 and Error-value=TBD10, Remote IP is in use. The detail Error-Types and Error-Values are defined in If the established BGP session is broken after some time, the PCC should report such error via PCErr message with: 5) Err-type=TBD6 and Error-value=TBD11, Existing BGP session is broken.

The explicit route establishment procedures can be used to install a route via PCE on the PCC, using PCInitiate and PCRpt message pair. Such explicit routes operate similar to static routes installed by network management protocols. The procedures of such explicit route addition and remove must be controlled by the PCE in an specific order so that the pathways are established without loops. The PCInitiate message should be sent to the on-path routers respectively. In the example, for explicit route from R1 to R7, the PCInitiate message should be sent to R1(via M1 message), R2(via M2 message) and R4(via M3 message), as shown in Figure 2. For explicit route from R7 to R1, the PCInitiate message should be sent to R7(via M1 message), R4(via M2 message) and R2(via M3 message), as shown in Figure 3. When PCC receives the EPR and the CCI object (with the R bit set to 0 in SRP object) in PCInitiate message, the PCC should install the explicit route to the peer. When PCC install successfully the explicit route to the peer, it should report the result via the PCRpt messages, with EPR object and the corresponding SRP and CCI object included. When PCC receives the EPR and the CCI object with the R bit set to 1 in SRP object in PCInitiate message, the PCC should clear the explicit route to the peer that indicated by the EPR object. When PCC clear successfully the explicit route that indicated by this object, it should report the result via the PCRpt message, with the EPR object included, and the corresponding SRP and CCI object.

The message number, message peers, message type and message key parameters in the above figures are shown in below table:

In order to avoid the transient loop during the deploy of explicit peer route, the EPR object should be sent to the PCCs in the reverse order of the E2E path. To remove the explicit peer route, the EPR object should be sent to the PCCs in the same order of E2E path. To accomplish ECMP effects, the PCE can send multiple EPR objects to the same node, with the same route priority and peer address value but different next hop addresses. The PCC should verify that the next hop address is reachable. Upon the error occurs, the PCC SHOULD send the corresponding error via PCErr message, with an error information: 1) Error-type=TBD6, Error-value=TBD12, Explicit Peer Route Error. When the peer info is not the same as the peer info that indicated in BPI object in PCC for the same path that is identified by Symbolic Path Name TLV, an PCErr message should be reported, with an error information: 2) Error-type=TBD6, Error-value=TBD17, EPR/BPI Peer Info Mismatch. When PCE receives the PCError message that indicates the BGP session to the peer address is broken(Err-type=TBD6 and Error-value=TBD11, Existing BGP session is broken), the PCE should clear the explicit peer route to the peer address.

The detail procedures for BGP prefix advertisement are shown below, using PCInitiate and PCRpt message pair. The PCInitiate message should be sent to PCC that acts as BGP peer edge router only. In the example, it should be sent to R1(M1) or R7(M2) respectively. When PCC receives the PPA and the CCI object (with the R bit set to 0 in SRP object) in PCInitiate message, the PCC should send the prefixes indicated in this object to the appointed BGP peer. When PCC sends successfully the prefixes to the appointed BGP peer, it should report the result via the PCRpt messages, with PPA object and the corresponding SRP and CCI object included. When PCC receives the PPA and the CCI object with the R bit set to 1 in SRP object in PCInitiate message, the PCC should withdraw the prefixes advertisement to the peer that indicated by this object. When PCC withdraws successfully the prefixes that indicated by this object, it should report the result via the PCRpt message, with the PPA object included, and the corresponding SRP and CCI object.

The allowed AFI/SAFI for the IPv4 BGP session should be 1/1(IPv4 prefix) and the allowed AFI/SAFI for the IPv6 BGP session should be 2/1(IPv6 prefix). If mismatch occur, an error: 1) Error-type=TBD6, Error-value=TBD18, BPI/PPR address family mismatch should be reported via PCErr message. When the peer info is not the same as the peer info that indicated in BPI object in PCC for the same path that is identified by Symbolic Path Name TLV, an error: 2) Error-type=TBD6, Error-value=TBD19, PPA/BPI peer info mismatch should be reported via the PCErr message. When PCE receives the PCError message that indicates the BGP session to the peer address is broken(Err-type=TBD6 and Error-value=TBD11, Existing BGP session is broken), the PCE should clear the prefixes advertisement to the peer.

One new CCI Object and three new PCEP objects are defined in this document. All new PCEP objects are as per

The Central Control Instructions (CCI) Object is used by the PCE to specify the forwarding instructions defined in . This document defines another object-type for Native-IP. CCI Object-Type is TBD13 for Native-IP as below

The field CC-ID is as described in . Following fields are defined for CCI Object-Type TBD13 is set to zero while sending, ignored on receipt. is used to carry any additional information pertaining to the CCI. Currently no flag bits are defined. The Symbolic Path Name TLV MUST be included in the CCI Object-Type TBD13 to identify the E2E TE path in Native IP environment and MUST be unique.

The BGP Peer Info object is used to specify the information about the peer with which the PCC should establish the BGP relationship. This object should only be included and sent to the head and end router of the E2E path in case there is no Route Reflection (RR) involved. If the RR is used between the head and end routers, then such information should be sent to head router, RR and end router respectively. By default, the Local/Peer IP address SHOULD be dedicated to the usage of native IP TE solution, and SHOULD NOT be used by other BGP sessions that established by manual or non PCE initiated configuration. BGP Peer Info Object-Class is TBD14 BGP Peer Info Object-Type is 1 for IPv4 and 2 for IPv6 The format of the BGP Peer Info object body for IPv4(Object-Type=1) is as follows:

The format of the BGP Peer Info object body for IPv6(Object-Type=2) is as follows:

Peer AS Number: 4 Bytes, to indicate the AS number of Remote Peer. ETTL: 1 Byte, to indicate the multihop count for EBGP session. It should be 0 and ignored when Local AS and Peer AS is same. Reserved: is set to zero while sending, ignored on receipt. T bit: Indicates whether the traffic that associated with the prefixes advertised via this BGP session is transported via IPinIP tunnel (when T bit is set) or not (when T bit is clear). Local IP Address(4/16 Bytes): IP address of the local router, used to peer with other end router. When Object-Type is 1, length is 4 bytes; when Object-Type is 2, length is 16 bytes. Peer IP Address(4/16 Bytes): IP address of the peer router, used to peer with the local router. When Object-Type is 1, length is 4 bytes; when Object-Type is 2, length is 16 bytes; Tunnel Source IP Address(4/16 Bytes): IP address of the tunnel source, should be owned by the local router. When Object-Type is 1, length is 4 bytes; when Object-Type is 2, length is 16 bytes. It should be different from the Local IP Address. Tunnel Destination IP Address(4/16 Bytes): IP address of the tunnel destination, should be owned by the peer router. When Object-Type is 1, length is 4 bytes; when Object-Type is 2, length is 16 bytes. It should be different from the Peer IP Address. Additional TLVs: TLVs that associated with this object, can be used to convey other necessary information for dynamic BGP session establishment. Their definitions are out of the current document. When PCC receives BPI object, with Object-Type=1, it should try to establish BGP session with the peer in AFI/SAFI=1/1; when PCC receives BPI object with Object-Type=2, it should try to establish the BGP session with the peer in AFI/SAFI=2/1.

The Explicit Peer Route object is defined to specify the explicit peer route to the corresponding peer address on each device that is on the E2E assurance path. This Object should be sent to all the devices on the E2E assurance path that calculated by PCE. The path established by this object should have higher priority than other path calculated by dynamic IGP protocol, but should have lower priority than the static route configured by manual or NETCONF or by other means. Explicit Peer Route Object-Class is TBD15. Explicit Peer Route Object-Type is 1 for IPv4 and 2 for IPv6 The format of Explicit Peer Route object body for IPv4(Object-Type=1) is as follows:

The format of Explicit Peer Route object body for IPv6(Object-Type=2) is as follows:

Route Priority: 2 Bytes, The priority of this explicit route. The higher priority should be preferred by the device. This field is used to indicate the backup path at each hop. Reserved: is set to zero while sending, ignored on receipt. Peer/Tunnel Destination Address: To indicate the peer address(4/16 Bytes). When T bit is set in the associated BPI object, use the tunnel destination address in BPI object; when T bit is clear, use the peer address in BPI object. Next Hop Address to the Peer/Tunnel Destination Address: To indicate the next hop address(4/16 Bytes) to the corresponding peer/tunnel destination address. Additional TLVs: TLVs that associated with this object, can be used to convey other necessary information for explicit peer path establishment. Their definitions are out of the current document.

The Peer Prefix Advertisement object is defined to specify the IP prefixes that should be advertised to the corresponding peer. This object should only be included and sent to the head/end router of the E2E path. The prefixes information included in this object MUST only be advertised to the indicated peer, MUST NOT be advertised to other BGP peers. Peer Prefix Advertisement Object-Class is TBD16 Peer Prefix Advertisement Object-Type is 1 for IPv4 and 2 for IPv6 The format of the Peer Prefix Advertisement object body is as follows:

Peer IPv4 Address: 4 Bytes. Identifies the peer IPv4 address that the associated prefixes will be sent to. IPv4 Prefix subojects: List of IPv4 Prefix subobjects defined in , identify the prefixes that will be sent to the peer identified by Peer IPv4 Address. Peer IPv6 Address: 16 Bytes. Identifies the peer IPv6 address that the associated prefixes will be sent to. IPv6 Prefix subojects: List of IPv6 Prefix subobjects defined in , identify the prefixes that will be sent to the peer identified by Peer IPv6 Address. Additional TLVs: TLVs that associated with this object, can be used to convey other necessary information for prefixes advertisement. Their definitions are out of the current document.

A PCEP-ERROR object is used to report a PCEP error and is characterized by an Error-Type that specifies that type of error and an Error-value that provides additional information about the error. An additional Error-Type and several Error-values are defined to represent the errors related to the newly defined objects that are related to Native IP TE procedures.

This document defines the procedures and objects to create the BGP sessions and advertise the associated prefixes dynamically. Only the key information, for example peer IP addresses, peer AS number are exchanged via the PCEP protocol. Other parameters that are needed for the BGP session setup should be derived from their default values, as described in . When the PCE sends out the PCInitiate message with BPI object embedded to establish the BGP session between the PCC peers, it should wait enough time to get the BGP session successful establishment report from the underlay PCCs. if the PCE can't get such report after the duration, then it should declare the failure of BGP session setup and try the establishment procedure again, or report the failure to the operator. Upon receiving such key information, the BGP module on the PCC should try to accomplish the task appointed by the PCEP protocol and report the successful status to the PCEP modules after the session is setup. There is no influence on current implementation of BGP Finite State Machine(FSM). The PCEP cares only the success and failure status of BGP session, and acts upon such information accordingly. The error handling procedures related to incorrect BGP parameters are specified in , , and .

The information transferred in this document is mainly used for the light weight BGP session setup, explicit route deployment and the prefix distribution. The planning, allocation and distribution of the peer addresses within IGP should be accomplished in advanced and they are out of the scope of this document. describes the state synchronization procedure between stateful PCE and PCC. The communication of PCE and PCC described in this document should also follow the same procedures, treat the three newly defined objects(BPI, EPR, PPA) associated with the same symbolic path name as the attribute of the same path in the LSP-DB. When PCE detects one or some of the PCCs are out of control, it should recompute and redeploy the traffic engineering path for native IP on the active PCCs. The PCE should assure the avoidance of possible transient loop in such node failure when it deploys the explicit peer route on the PCCs. In case of a PCE failure, a new PCE can gain control over the central controller instructions. As per the PCEP procedures in , the State Timeout Interval timer is used to ensure that a PCE failure does not result in automatic and immediate disruption for the services. Similarly, as per , the central controller instructions are not removed immediately upon PCE failure. Instead, they could be re-delegated to the new PCE before the expiration of this timer, or be cleaned up on the expiration of this timer. This allows for network clean up without manual intervention. The PCC MUST support the removal of CCI as one of the behaviors applied on expiration of the State Timeout Interval timer.

[Note to the RFC Editor - remove this section before publication, as well as remove the reference to RFC 7942.] This section records the status of known implementations of the protocol defined by this specification at the time of posting of this Internet-Draft, and is based on a proposal described in [RFC7942]. The description of implementations in this section is intended to assist the IETF in its decision processes in progressing drafts to RFCs. Please note that the listing of any individual implementation here does not imply endorsement by the IETF. Furthermore, no effort has been spent to verify the information presented here that was supplied by IETF contributors. This is not intended as, and must not be construed to be, a catalog of available implementations or their features. Readers are advised to note that other implementations may exist. According to , "this will allow reviewers and working groups to assign due consideration to documents that have the benefit of running code, which may serve as evidence of valuable experimentation and feedback that have made the implemented protocols more mature. It is up to the individual working groups to use this information as they see fit".

At the time of posting the -22 version of this document, there are no known implementations of this mechanism. A proof of concept for the overall design has been verified using another SBI protocol on the Open DayLight (ODL) controller.

ZTE is preparing an implementation of this document as the time of posting the -22 version of this document.

The setup of BGP sessions, prefix advertisement, and explicit peer route establishment are all controlled by the PCE. See and for BGP security considerations. Security consideration part in and should be considered. To prevent a bogus PCE sending harmful messages to the network nodes, the network devices should authenticate the validity of the PCE and ensure a secure communication channel between them. Mechanisms described in should be used.

created a sub-registry within the "Path Computation Element Protocol (PCEP) Numbers" registry called "PCEP Path Setup Types". IANA is requested to allocate a new code point within this sub-registry, as follows:

created a sub-registry within the "Path Computation Element Protocol (PCEP) Numbers" registry to manage the value of the PCECC-CAPABILITY sub-TLV's 32-bits Flag field. IANA is requested to allocate a new bit position within this registry, as follows:

IANA is requested to allocate new codepoints in the "PCEP Objects" sub-registry as follows:

IANA is requested to allocate new error types and error values within the "PCEP-ERROR Object Error Types and Values" sub-registry of the PCEP Numbers registry for the following errors::

Dhruv Dhody has contributed to this document.

Thanks Mike Koldychev, Susan Hares, Siva Sivabalan, Adam Simpson for his valuable suggestions and comments.