| PCE Working Group | Y. Tanaka |
| Internet-Draft | Y. Kamite |
| Intended status: Standards Track | NTT Communications |
| Expires: September 8, 2019 | R. Palleti |
| C. Li | |
| H. Zheng | |
| Huawei Technologies | |
| March 7, 2019 |
Make-Before-Break (MBB) MPLS-TE LSP restoration and reoptimization procedure using Stateful Path Computation Element (PCE).
draft-tanaka-pce-stateful-pce-mbb-08
Stateful Path Computation Element (PCE) and its corresponding protocol extensions provide a mechanism that enables PCE to do stateful control of Multiprotocol Label Switching (MPLS) Traffic Engineering Label Switched Paths (TE LSP). Stateful PCE supports manipulating of the existing LSP's state and attributes (e.g., bandwidth and path) via delegation and also instantiation of new LSPs in the network via PCE Initiation procedures.
In the current MPLS TE network using Resource ReSerVation Protocol (RSVP-TE), LSPs are often controlled by Make-before-break (M-B-B) signaling by the headend for the purpose of LSP restoration and reoptimization. In most cases, it is an essential operation to reroute LSP traffic without any data disruption.
This document specifies the procedure of applying stateful PCE's control to make-before-break RSVP-TE signaling. In this document, two types of restoration/reoptimization procedures are defined, implicit mode and explicit mode. This document also specifies the usage and handling of stateful PCEP (PCE Communication Protocol) messages, expected behavior of PCC as RSVP-TE headend and necessary extensions of additional PCEP objects.
This Internet-Draft is submitted in full conformance with the provisions of BCP 78 and BCP 79.
Internet-Drafts are working documents of the Internet Engineering Task Force (IETF). Note that other groups may also distribute working documents as Internet-Drafts. The list of current Internet-Drafts is at https://datatracker.ietf.org/drafts/current/.
Internet-Drafts are draft documents valid for a maximum of six months and may be updated, replaced, or obsoleted by other documents at any time. It is inappropriate to use Internet-Drafts as reference material or to cite them other than as "work in progress."
This Internet-Draft will expire on September 8, 2019.
Copyright (c) 2019 IETF Trust and the persons identified as the document authors. All rights reserved.
This document is subject to BCP 78 and the IETF Trust's Legal Provisions Relating to IETF Documents (https://trustee.ietf.org/license-info) in effect on the date of publication of this document. Please review these documents carefully, as they describe your rights and restrictions with respect to this document. Code Components extracted from this document must include Simplified BSD License text as described in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the Simplified BSD License.
[RFC5440] describes the Path Computation Element Protocol PCEP. PCEP enables the communication between a Path Computation Client (PCC) and a Path Control Element (PCE), or between PCE and PCE, for the purpose of computation of Multiprotocol Label Switching (MPLS) as well as Generalzied MPLS (GMPLS) for Traffic Engineering Label Switched Path (TE LSP) characteristics.
[RFC8231] describes the stateful Path Computation Elements (PCE) and defines the extensions to PCEP to enable stateful control of LSPs within and across PCEP sessions, further it also describes mechanisms to effect LSP state synchronization between PCCs and PCEs, and PCE control of timing and sequence of path computations within and across PCEP sessions.
Today, however, there is no detailed procedure specified for restoration and reoptimization of MPLS-TE LSP using stateful PCE. In today's MPLS RSVP-TE mechanism, make-before-break (M-B-B) is a widely common scheme supported by headend Label Edge Router (LER) in order to assure no traffic disruption during restoration and reoptimization. Hence it is naturally desirable for stateful PCE to control M-B-B based signaling and forwarding process.
This document specifies the definite procedures of applying stateful PCE's control of the M-B-B procedures. In this document, two types of restoration/reoptimization procedures are defined, Implicit mode and Explicit mode. This document also specifies the usage and handling of stateful PCEP (PCE Communication Protocol) messages, expected behavior of PCC as RSVP-TE headend and several extensions of additional objects.
The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in BCP 14 [RFC2119] [RFC8174] when, and only when, they appear in all capitals, as shown here.
This document uses the following terms defined in [RFC5440]: PCC, PCE, PCEP Peer.
This document uses the following terms defined in [RFC3209]: make-before-break (M-B-B), Path State Block (PSB).
This document uses the following terms defined in [RFC4426] and [RFC4427]: recovery, protection, restoration.
According to their definition the term "recovery" is generically used to denote both protection and restoration; the specific terms "protection" and "restoration" are used only when differentiation is required. The subtle distinction between protection and restoration is made based on the resource allocation done during the recovery period. Hence the protection allocates LSP resource in advance of a failure, while the restoration allocates LSP resource after a failure occur.
As for current MPLS mechanism, make-before-break(M-B-B) concept is outlined in [RFC3209], which allows adaptive and smooth RSVP-TE LSP rerouting that does not disrupt traffic or adversely impact network operations while rerouting is in progress. M-B-B is applicable for reoptimizing LSP's route and resources for several use cases, for example, to adopt better path for reversion after failure, to change traversing node/links for planned maintenance, to change bandwidth of LSPs etc. M-B-B is also used for global restoration scenario in case of failure, which is effective if operators do not want to reserve both working and standby LSP's bandwidth in advance. Once failure occur, LSP becomes down, however PSB (Path State Block) of a headend node remains and keep resources intact. Using M-B-B, the headend node is able to resignals working LSP while the PSB remains until new restoration LSP is successfully established. In real deployment, it can also be operated with local protection scheme FRR (Fast ReRoute).
Since M-B-B operational scheme is universally common in MPLS network today, it is naturally much desirable to utilize it under the architecture of stateful PCE.
The basic procedure of the Make-Before-Break method is outlined as follows:
In M-B-B, it is an important behavior that headend node handles the sequence of data traffic switchover. The headend is able to Make one or more new LSPs for a particular Tunnel (i.e., it is allowed to signal multiple LSPs with different LSP-IDs that share a common Tunnel IDs), and the headend will switch the traffic to only one (or some) of those LSPs. In some use cases about stateful PCE, it is expected that controller/operators can watch and control when the data is switched over and which LSPs are used. Therefore, this document covers such a procedure and related message extensions.
There are possibly two modes introduced for Make-Before-Break procedure under stateful PCE. The first one is "implicit M-B-B mode", where the operation is triggered by a Update Request(PCUpd) message from a PCE, and a PCC handles whole Make-Before-Break steps (signaling, transferring data traffic and teardown) by itself. This mode utilizes the existing messages and procedures as defined in [RFC8231] .
The second one is "explicit M-B-B mode", where the operation is triggered by a PCUpd message with a new TRIAL LSP TLV (defined in Section 6.2). A PCE also controls timing and sequence of the M-B-B steps that a PCC takes. This procedure uses ASSOCIATION Object that is defined in [I-D.ietf-pce-association-group].
Both types of procedure require at least two LSPs residing in a single MPLS-TE tunnel, working LSP and trial LSPs. An ingress node is currently transporting data traffic on the working LSP, and then it establishes one or more trial LSPs. As per [RFC3209] Section 2.5. "LSP ID" of a restoration LSP, which is newly signaled, differs from that of a working LSP in RSVP-TE. Note that it is also used for LSP-ID in LSP Identifiers TLVs in PCEP messages, and it differs from PLSP-ID ([RFC8231]). In this document, LSP ID of a working LSP describes "old" and that of a trial LSP describes "new" as a simple example.
Implicit mode has high affinity with most existing MPLS edge node implementations which perform entire steps of M-B-B automatically at once. This mode is particularly applicable for migration scenario for the existing deployment where service providers want their recovery/reoptimization operation be delegated to a centralized PCE.
Explicit mode is much more flexible than Implicit mode since it allows PCEs to manage each step of the M-B-B. Explicit mode is applicable to several new use cases that require split control of signaling and data switchover. For example, if end-to-end data path is created by connecting multiple individual LSPs across different segments (e.g., LSP stitching), in reoptimization scenario, data flowing cannot be started unless signaling of all LSPs is completed. Similarly, there is a case under Software Defined Networking (SDN) applications, where MPLS domain is connected to other non-MPLS domains, and the end-to-end data switchover timing should be carefully coordinated with various different methods of path/flow setup in each domain.
PCC and PCE can distinguish which mode, implicit mode or explicit mode, is to be performed by checking the presence of ASSOCIATION and certain TLV in the PCEP messages. The implementation MAY support both modes, but for each restoration/reoptimization operation, either one of them SHOULD be exclusively applied.
This specifies the detailed procedure of M-B-B LSP restoration and reoptimization using existing messages which are defined in [RFC8231] . This procedure is based on the existing messages/TLVs and no extensions are required. Once a PCC receives PCUpd message from a PCE, the PCC automatically executes the implicit M-B-B procedure as described in [RFC8231] Section 6.2.
Following Figure 1 illustrates the example of implicit M-B-B procedure, in following conditions. Tunnel ID and LSP ID are included in an LSP Identifiers TLV in a LSP Object.
__c__
/ \
PCE PCC(Ingress)--a-------b---Egress
| | |
| Data on old LSP =>)))))))))))))))))))))))|
| | : |
|--PCUpd(PLSP-ID=X,->| : |
| SRP-ID=Y, | |
| ERO=a-c-b) |---Path(ERO=a-c-b-, --> |
| | LSP ID new) |
| | |
| | <-----Resv-------------|
| <- PCRpt(PLSP-ID=X,| |
| O=Up, | |
| SRP-ID=Y, | |
| Tunnel ID=T1, | |
| LSP ID=new) | |
| | |
| | |
| Transfer data |))))))))))))))))))))))))|
| from old to new =>}}}}}}}}}}}}}}}}}}}}}}}}|
| | : |
| | : |
| |---PathTear(ERO=a-b, -> |
| | LSP ID old) |
| <- PCRpt(PLSP-ID=X,| |
| O=Dn,R=1, | |
| SRP-ID=0, | |
| Tunnel ID=T1, | |
| LSP ID=old) | |
O flag = Operational flag in LSP object.
R flag = Remove flag in LSP object.
Figure 1: Implicit Make-Before-Break Procedure
Comparing to the implicit M-B-B mode, explicit M-B-B mode allows a PCE to control timing and sequence of subsequent make-before-break steps.
As per [I-D.ietf-pce-association-group], LSPs are associated with other LSPs with which they interact by adding them to a common association group. In this draft, this grouping is used to define associations between a set of LSPs. This document define one new association type called "Explicit MBB Association Type" of value TBD1.
The following subsections specify each Explicit Make-Before-Break step in detail.
As a pre-requisite before starting explicit M-B-B, PCE makes an association group for working LSP by sending PCUpd message that contains ASSOCIATION object with TRIAL-LSP TLV with both D and T flags set to zero. TRIAL-LSP TLV is optional in the ASSOCIATION object at this step.
Figure 2 illustrates an example of working LSP (PLSP-ID P1, Tunnel ID T1, LSP-ID old, Association Group ID G1 and ERO Ingress-a-b-Egress).
__c__
/ \
PCE PCC(Ingress)--a-------b---Egress
| data traffic on old LSP |
| |))))))))))))))))))))))))|
|--PCUpd ------>| : |
| LSP Object | : |
| PLSP-ID=P1 | : |
| SRP-ID=S1 | : |
| LSP ID=old | |
| ASSOC Object | |
| Assoc-Type=MBB | |
| Assoc-ID=G1 | |
| +TRIAL-LSP TLV | |
| D-Flag=0 | |
| T-Flag=0 | |
| | |
Figure 2: Initiate Associate Group for old LSP
As a first step of M-B-B procedure, a PCC establishes a new LSP for restoration once PCC receives a PCInitiate/PCUpd message with T flag (in TRIAL-LSP TLV) set to 1, in a ASSOCIATION Object from a PCE. We call this newly established LSPs for restoration "trial LSP". A trial LSP is signaled the same RSVP-TE Tunnel ID but different LSP ID from active working LSP, and both the active working LSP and new trial LSPs MUST be signaled with Shared Explicit style as describes in [RFC3209].
When a new trial LSP was signaled successfully, the PCC sends a PCRpt message toward the PCE to notify the result. The PCRpt message from the PCC MUST have the LSP object with LSP-IDENTIFIERS TLV that indicates RSVP-TE Tunnel ID and LSP ID the PCC has just established.
If a new trial LSP failed to be established by some reason of RSVP-TE signaling, the PCC MUST send to the PCE a PCRpt message carrying LSP-IDENTIFIERS TLV and RSVP-ERROR-SPEC TLV as defined in [RFC8231] Section 7.3.4.
A PCC SHOULD accept multiple PCInitiate/PCUpd messages with TRIAL-LSP TLV in a ASSOCIATION Object. And a PCC SHOULD establish as many trial lsps as the number of PCInitiate/PCUpd messages it receives. A PCC may also choose to implement a limit on the number of such PCInitiate/PCUpd message.
Figure 3 illustrates a example, working LSP(PLSP-ID P1, Tunnel ID T1, LSP-ID old, ERO Ingress-a-b-Egress), trial LSP(PLSP-ID P1, Tunnel ID T1, LSP-ID new, ERO Ingress-a-c-b-Egress).
__c__
/ \
PCE PCC(Ingress)--a-------b---Egress
| data traffic on old LSP |
| | |
| PCInitiate/ |))))))))))))))))))))))))|
|--PCUpd ------>| : |
| LSP Object | : |
| PLSP-ID=P1 | : |
| SRP-ID=S2 | : |
| Tunnel ID=T1 | |
| LSP ID=0 | |
| ASSOC Object | |
| Assoc-Type=MBB | |
| Assoc-ID=G1 | |
| +TRIAL-LSP TLV | |
| D-Flag=0 | |
| T-Flag=1 | |
| ERO Obj=a-c-b | |
| | |
| |---Path(LSP ID=new, --> |
| | ERO=a-c-b) |
| | |
| | <----- Resv------------|
|<--PCRpt ---------| |
| LSP Object | : |
| PLSP-ID=P1 |))))))))))))))))))))))))|
| SRP-ID=S2 | : |
| Tunnel ID=T1 | : |
| LSP ID=new | : |
| ASSOC Object | : |
| Assoc-Type=MBB | : |
| Assoc-ID=G1 | : |
| +TRIAL-LSP TLV | : |
| D-Flag=0 | : |
| T-Flag=1 | : |
| RRO Obj=a-c-b | : |
| | |
Figure 3: Establish new LSP
As a second step, the PCC(Ingress) transfers data traffic from a working LSP to a trial LSP. To specify desired LSP for transferring data traffic, a PCUpd message from a PCE MUST have a TRIAL-LSP TLV set D flag to 1, in a ASSOCIATION Object.
Data switchover happens from old LSP to new trial LSP, once PCC receives a PCUpd message with D flag in TRIAL-LSP TLV set to 1 in the ASSOCIATION object from a PCE.
The PCC SHOULD tear down the old working LSP and other trial LSPs which the data traffic is no longer used immediately once the data traffic successfully switched over (See Figure 4).
[Editor's Note - Another option would be, a PCC tears down old lsp separately using mechanism in [RFC8281] for PCE-Initiated LSPs.]
The PCC sends to the PCE a PCRpt message to notify the removal of both old LSP and other trial LSPs, which SRP-ID is set to 0x00000000.
__c__
/ \
PCE PCC(Ingress)--a-------b---Egress
| | |
| |))))))))))))))))))))))))| data on old LSP
|--PCUpd ------> |))))))))))))))))))))))))|
| LSP Object |}}}}}}}}}}}}}}}}}}}}}}}}| data on new LSP
| PLSP ID=P1 |}}}}}}}}}}}}}}}}}}}}}}}}|
| SRP ID=S3 |}}}}}}}}}}}}}}}}}}}}}}}}|
| Tunnel ID=T1 | : |
| LSP ID=new | : |
| ASSOC Object | : |
| Assoc-Type=MBB | |
| Assoc-ID=G1 | |
| +TRIAL-LSP TLV | |
| D-Flag=1 | |
| T-Flag=0 | |
| | |
| <-- PCRpt --------| |
| LSP Object | |
| PLSP ID=P1 | |
| SRP ID=S3 | |
| Tunnel ID=T1 | |
| LSP ID=new | |
| |--PathTear(ERO a-b, -->| Tear down old
| | Tunnel=T1,LSP ID=old) | automatically
| | |
| <-- PCRpt(O=Dn,R=1,| |
| PLSP ID=P1 | |
| SRP ID=0 | |
| Tunnel ID=T1 | |
| LSP-ID=old) | |
| | |
| | |
O flag = Operational flag in LSP object.
R flag = Remove flag in LSP object.
Figure 4: Transfer data traffic from old LSP to new LSP
As per [I-D.ietf-pce-association-group], LSPs are associated with other LSPs with which they interact by adding them to a common association group. The Association ID will be used to identify the MBB group a set of LSPs belongs to. This document defines a new Association type, based on the generic Association object -
[I-D.ietf-pce-association-group] specify the mechanism for the capability advertisement of the association types supported by a PCEP speaker by defining a ASSOC-Type-List TLV to be carried within an OPEN object. This capability exchange for the association type described in this document (i.e. Explicit MBB Association Type) MUST be done before using the policy association, i.e., the PCEP speaker MUST include the Explicit MBB Association Type (TBD1) in the ASSOC-Type-List TLV before using this association type in the PCEP messages.
This Association-Type is dynamic in nature and created by the PCC or PCE for the LSPs belonging to the same TE tunnel (as described in [RFC3209]) originating at the same head node and terminating at the same destination. These associations are conveyed via PCEP messages to the PCEP peer. Operator-configured Association Range MUST NOT be set for this association-type and MUST be ignored.
This document defines a new TLV named TRIAL-LSP TLV which can be optionally carried in the ASSOCIATION object.
0 1 2 3
0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Type=TBD2 | Length |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Flags |D|T|
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
Figure 5: TRIAL-LSP TLV format
TRIAL-LSP TLV is an optional TLV of the ASSOCIATION Object and is used in a PCInitiate/PCUpd message especially to perform explicit mode M-B-B. A PCC signals a trial LSP once it receives a PCUpd in which ASSOCIATION object has a TRIAL-LSP TLV.
The MBB association group MAY carry some optional TLVs including but not limited to:
o VENDOR-INFORMATION-TLV: Used to communicate arbitrary vendor specific behavioral information,, described in [RFC7470].
As per the processing rules specified in section 5.4 of [I-D.ietf-pce-association-group], if a PCEP speaker does not support this association-type, it would return a PCErr message with Error-Type 26 (Early allocation by IANA) "Association Error" and Error-Value 1 "Association-type is not supported".
All LSPs (new or old) within this association MUST belong to the same TE Tunnel (as described in [RFC3209]) and have the same source and destination. If a PCEP speaker attempts to add an LSP to this association and the Tunnel ID (as carried in LSP-IDENTIFIERS TLV [RFC8231], with description as per [RFC3209]) or source or destination of the LSP is different from the LSP(s) in the PPAG, the PCC MUST send PCErr with Error-Type= 29 (Early allocation by IANA) (Association Error) [I-D.ietf-pce-association-group] and Error-Value = TBD (Tunnel ID or End points mismatch).
All processing as per section 5.4 of [I-D.ietf-pce-association-group] continue to apply.
This document defines one new type for association, which do not add any new security concerns beyond those discussed in [RFC5440], [RFC8231] and [I-D.ietf-pce-association-group] in itself.
This document defines the following new PCEP TLVs:
Value Meaning Reference
TBD2 TRIAL-LSP TLV This document
This document defines the following new association type originally defined in [I-D.ietf-pce-association-group].
Value Name Reference
TBD1 MBB Association Type This document
In addition to basic operations under multiple PCEs as described in [RFC8231], a PCC supports both types of M-B-B operations.
Implicit mode M-B-B requires only one PCUpd message to trigger M-B-B process, therefore a PCC accepts a message from a primary PCE whom the PCC delegates the LSPs to. An attempt to update parameters of a non-delegated LSP results in the PCC sending a PCErr message as defined in [RFC8231].
Explicit mode M-B-B requires at least three PCUpd messages(1. for new Association-Group creation, 2. for trial-LSP signaling, 3. for traffic switchover) to trigger each subsequent step. All steps MUST be taken by one primary PCE because state synchronization of trial-LSPs between the primary and backup PCE may be complex. If the PCC revokes LSP delegations after a Redelegation Timeout Interval, the PCC MUST tear down all trial-LSPs and redelegate a working LSP to alternate PCE. An attempt to trigger either step of explicit mode M-B-B of a non-delegated LSP results in the PCC sending the same PCErr as implicit mode M-B-B.
Many thanks to Ina Minei, Adrian Farrel, Yimin Shen, and Xian Zhang for their ideas and feedback in documentation.
| [RFC3209] | Awduche, D., Berger, L., Gan, D., Li, T., Srinivasan, V. and G. Swallow, "RSVP-TE: Extensions to RSVP for LSP Tunnels", RFC 3209, DOI 10.17487/RFC3209, December 2001. |
| [RFC4426] | Lang, J., Rajagopalan, B. and D. Papadimitriou, "Generalized Multi-Protocol Label Switching (GMPLS) Recovery Functional Specification", RFC 4426, DOI 10.17487/RFC4426, March 2006. |
| [RFC4427] | Mannie, E. and D. Papadimitriou, "Recovery (Protection and Restoration) Terminology for Generalized Multi-Protocol Label Switching (GMPLS)", RFC 4427, DOI 10.17487/RFC4427, March 2006. |
| [RFC7470] | Zhang, F. and A. Farrel, "Conveying Vendor-Specific Constraints in the Path Computation Element Communication Protocol", RFC 7470, DOI 10.17487/RFC7470, March 2015. |
Dhruv Dhody
Huawei Technologies
Divyashree Techno Park, Whitefield
Bangalore, Karnataka 560066
India
Email: dhruv.ietf@gmail.com