IETF Internet Draft Arthi Ayyangar(Editor) Proposed Status: Standards Track Juniper Networks Expires:
AprilSeptember 2006 Jean-Philippe Vasseur(Editor) Cisco Systems, Inc. October 2005March 2006 Inter domain GMPLS Traffic Engineering - RSVP-TE extensions draft-ietf-ccamp-inter-domain-rsvp-te-02.txtdraft-ietf-ccamp-inter-domain-rsvp-te-03.txt Status of this Memo By submitting this Internet-Draft, each author represents that any applicable patent or other IPR claims of which he or she is aware have been or will be disclosed, and any of which he or she becomes aware will be disclosed, in accordance with Section 6 of BCP 79. 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. This Internet-Draft will expire on April 3,September 6, 2006. Copyright Notice Copyright (C) The Internet Society (2005).(2006). All Rights Reserved. Abstract This document describes extensions to Generalized Multi-Protocol Label Switching (GMPLS) Resource ReserVation Protocol - Traffic Engineering (RSVP-TE) signaling required to support mechanisms for the establishment and maintenance of GMPLS Traffic Engineering (TE) Label Switched Paths (LSPs), both packet and non-packet, that traverse multiple domains. For the purpose of this document, a domain is considered to be any collection of network elements within a common realm of address space or path computation responsibility. Examples of such domains include Autonomous Systems, IGP areas and GMPLS overlay networks. Table of Contents 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 3 1.1 Conventions used in this document . . . . . . . . . . . . 3 1.2 Terminology . . . . . . . . . . . . . . . . . . . . . . . 3 2. Signaling overview . . . . . . . . . . . . . . . . . . . . . 4 2.1 Signaling options . . . . . . . . . . . . . . . . . . . . 4 3. Procedures on the domain boundary node . . . . . . . . . . . . 5 3.1 Rules on ERO processing . . . . . . . . . . . . . . . . . 6 3.2 LSP setup failure and crankback . . . . . . . . . . . . . 8 3.3 RRO processing across domains . . . . . . . . . . . . . . 9 4. RSVP-TE signaling extensions . . . . . . . . . . . . . . . . . 9 4.1 Control of downstream choice of signaling method . . . . . 9 5. Protection and recovery of inter-domain TE LSPs . . . . . . . 11 5.1 Fast Recovery support using MPLS TE Fast Reroute . . . . . 11 5.1.1 Failure within a domain (link or node failure) . . . . 11 5.1.2 Failure of link at domain boundaries . . . . . . . . . 11 5.1.3 Failure of a boundary node . . . . . . . . . . . . . . 12 5.2 Protection and recovery of GMPLS LSPs . . . . . . . . . . 13 6. Re-optimization of inter-domain TE LSPs . . . . . . . . . . . 13 7. Security Considerations . . . . . . . . . . . . . . . . . . . 14 8. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 15 8.1 Attribute Flags for LSP_ATTRIBUTES object . . . . . . . . 15 8.2 New Error Codes . . . . . . . . . . . . . . . . . . . . . 16 9. Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . 16 10. References . . . . . . . . . . . . . . . . . . . . . . . . . 16 10.1 Normative References . . . . . . . . . . . . . . . . . . 16 10.2 Informative References . . . . . . . . . . . . . . . . . 17 Appendix 1: Examples . . . . . . . . . . . . . . . . . . . . . . 18 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . 22 Intellectual Property and Copyright Statements . . . . . . . . 23 1. Introduction The requirements for inter-area and inter-AS MPLS Traffic Engineering have been developed by the Traffic Engineering Working Group and have been stated in [INTER-AREA-TE-REQS] and [INTER-AS-TE-REQS] respectively. Many of these requirements also apply to GMPLS networks. The framework for inter-domain GMPLS Traffic Engineering has been provided in [INTER-DOMAIN-FRAMEWORK]. This document presents the RSVP-TE signaling extensions for the setup and maintenance of TE LSPs that span multiple domains. The signaling procedures described in this document are applicable to both MPLS packet LSPs ([RSVP-TE]) and all LSPs that use RSVP-TE GMPLS extensions as described in [RSVP-GMPLS]. Three different signaling methods along with the corresponding RSVP-TE extensions and procedures are proposed in this document. For the purpose of this document, a domain is considered to be any collection of network elements within a common realm of address space or path computation responsibility. Examples of such domains include Autonomous Systems, IGP areas and GMPLS overlay networks ([GMPLS- OVERLAY]). 1.1. 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 [RFC2119]. 1.2. Terminology ASBR: routers used to connect together ASes of a different or the same Service Provider via one or more Inter-AS links. Bypass Tunnel: an LSP that is used to protect a set of LSPs passing over a common facility. ERO: Explicit Route Object H-LSP: Hierarchical LSP FA-LSP: Forwarding Adjacency LSP LSP: MPLS Label Switched Path MP: Merge Point. The node where bypass tunnels meet the protected LSP. NHOP bypass tunnel: Next-Hop Bypass Tunnel. A backup tunnel, which bypasses a single link of the protected LSP. NNHOP bypass tunnel: Next-Next-Hop Bypass Tunnel. A backup tunnel, which bypasses a single node of the protected LSP. PLR: Point of Local Repair. The head-end of a bypass tunnel. Protected LSP: an LSP is said to be protected at a given hop if it has one or multiple associated backup tunnels originating at that hop. RRO - Record Route Object TE: Traffic Engineering TE LSP: Traffic Engineering Label Switched Path TE link: Traffic Engineering link TED: MPLS Traffic Engineering Database 2. Signaling overview The RSVP-TE signaling of a TE LSP within a single domain is described in [RSVP-TE] and [RSVP-GMPLS]. This document focuses on the RSVP-TE signaling extensions required for inter-domain TE LSP setup and maintenance. Any other extensions that may be needed for routing or path computation are outside the scope of this document. 2.1. Signaling options There are three ways in which an RSVP-TE LSP could be signaled across multiple domains: Contiguous - A contiguous TE LSP is a single end-to-end TE LSP that is setup across multiple domains using RSVP-TE signaling procedures described in [RSVP-TE] and [RSVP-GMPLS]. No additional TE LSPs are required to signal a contiguous TE LSP and the same RSVP-TE information for the TE LSP is maintained along the entire LSP path. Nesting - Nesting one or more TE LSPs into another TE LSP is described in [LSP-HIERARCHY]. This technique can be used to nest one or more inter-domain TE LSPs into an intra-domain hierarchical LSP (H-LSP). Label stacking construct may be used to achieve nesting, when appropriate, say in packet networks. In the rest of this document, the term H-LSP is used to refer to an LSP that allows nesting of other LSPs into it and that may or may not be advertised as a TE link. Furthermore, an H-LSP may or may not be an FA-LSP [LSP- HIERARCHY]. Stitching - The concept of LSP stitching as well as the required signaling procedures is described in [LSP-STITCHING]. This technique can be used to stitch an inter-domain TE LSP to an intra-domain LSP segment. A inter-domain stitched TE LSP is a TE LSP made up of different TE LSP segments within each domain which are "stitched" together in the data plane so that an end-to-end LSP is achieved in the data plane. In the control plane, however, the different LSP segments are signaled as distinct RSVP sessions which are independent from the RSVP session for the inter-domain LSP. While stitching is similar to nesting in the control plane, in the data plane, stitching allows for only one inter-domain LSP to be associated with any one intra-domain LSP, and does not require the use of label stacks. On receipt of an LSP setup request for an inter-domain TE LSP, the decision of whether to signal the LSP contiguously or whether to nest or stitch it to another TE LSP, depends on the signaled TE LSP characteristics from the head-end node or the local node configuration, when not explicitly signaled. Also, the TE LSP segment or H-LSP within the domain may either be pre-configured or signaled dynamically based on the arrival of the inter-domain TE LSP setup request. 3. Procedures on the domain boundary node Whether an inter-domain TE LSP is contiguous, nested or stitched is determined mostly by the signaling method supported by or configured on the intermediate nodes, usually the domain boundary nodes that the inter-domain TE LSP traverses. It also depends on certain parameters that may be signaled by the head-end node for the inter-domain TE LSP. When a domain boundary node receives the RSVP Path message for an inter-domain TE LSP setup, it MUST carry out the following procedures before it can forward the Path message to the next node along the path: 1. Apply any locally configured policies. In case of a policy failure, the node SHOULD fail the setup of the LSP and originate a PathErr with error code=2 ("Policy control failure") and error sub-code="Inter-domain policy failure" (TBD). 2. Determine the signaling method to be used. The head-end node of the inter-domain TE LSP may have explicitly specified a signaling method or if the signaling method is not explicitly signaled, then the node MAY determine the signaling method based on local configuration and policies. If the desired signaling method signaled by the head-end cannot be supported at this node for some reason, then a PathErr message as described in Section 4.1 MUST be generated. 3. Carry out ERO procedures as described in the next section in addition to the procedures in [RSVP-TE] and [RSVP-GMPLS]. 4. Perform any path computations as required (say for ERO expansion). The path computation procedure itself is outside the scope of this document. One such path computation option is addressed in [INTER-DOMAIN-PD-PATH-COMP]. Another option is to use a Path Computation Element (PCE) ([PCE]) for path computation. Path computation may itself involve determining the exit point from the TE domain ([INTER-DOMAIN-PD-PATH-COMP]). 4a. In case of nesting or stitching, either find an existing intra-domain TE LSP to carry the inter-domain TE LSP or signal a new one, depending on local policy. 4b. In case of a path computation failure, a PathErr SHOULD be generated as described in [INTER-DOMAIN-PD-PATH-COMP]. 5. In case of any other RSVP signaling failures, procedures as described in [RSVP-TE] and [RSVP-GMPLS] SHOULD be followed. Follow procedures related to LSP setup failure and crankback as described in Section 3.2, where applicable. 6. Carry out RRO procedures as described in Section 3.3, if applicable. 3.1. Rules on ERO processing The ERO that a domain boundary node receives in the Path message for an inter-domain TE LSP will be dependent on several factors such as the level of visibility that the head-end node of the inter-domain TE LSP has into other domains, the path computation techniques applied at the head-end node, policy agreements between two domains; etc. Eventually, when the ERO reaches a domain boundary node, the following rules SHOULD be used for ERO processing and signaling. Within a domain, there may be no H-LSPs or LSP segments. If they are present, then they may originate and terminate on domain boundary nodes. There could also be H-LSPs and LSP segments that may originate and terminate at other nodes in the domain. In general, these ERO processing rules are also applicable to non-boundary nodes that may participate in signaling the inter-domain TE LSP. 1. If there are any policies related to ERO processing for the LSP, they SHOULD be applied and corresponding actions should be taken. E.g. there could exist a policy to reject inter-domain LSP setup request containing an ERO with subobjects identifying nodes within the domain. In case of inter-domain LSP setup failures due to policy failures related to ERO processing, the node SHOULD issue a PathErr with error code=2 ("Policy control failure") and error sub-code="Inter-domain explicit route rejected" (TBD). 2. Section 8.2 of [LSP-HIERARCHY] describes how a node at the edge of a region (domain) processes the ERO in the incoming Path message and uses this ERO, to either find an existing H-LSP or signal a new H-LSP using the ERO hops. This also includes adjusting the ERO before sending the Path message to the next hop node. These procedures SHOULD also be followed for nesting or stitching of inter-domain TE LSPs to H-LSPs or LSP segments respectively. While the domain boundaries are tied to link switching capabilities in [LSP-HIERARCHY], these procedures are also applicable to other domain boundary nodes in the context of this document. 3. If the ERO subobject identifies a TE link formed by a H-LSP or LSP segment within the domain, either numbered or unnumbered, then a contiguous LSP MUST NOT be signaled. The node MUST either nest or stitch the inter-domain TE LSP to the local H-LSP or LSP segment. If, however, the head-end node for the inter-domain LSP has requested that the inter-domain TE LSP be contiguous, then this is a conflict and a PathErr with error code=24 ("Routing Problem") and error sub-code="ERO conflicts with inter-domain signaling method" (TBD) SHOULD be issued. 4. In the absence of any ERO subobjects, the LSP destination in the SESSION object SHOULD be considered as the next loose hop. A node may also receive an ERO with explicit IPv4, IPv6 or AS number loose hops. In such cases, a path computation to expand this loose hop SHOULD be carried out. Path computation as described in [INTER-DOMAIN-PD-PATH-COMP] or using a PCE should be used to expand the ERO in these cases and to determine the intermediate hops. 5. In case of any other failures in processing the ERO hop(s), a PathErr message with appropriate error codes as described in [RSVP-TE] or [RSVP-GMPLS] SHOULD be generated. 3.2. LSP setup failure and crankback In case of any setup failures along the path due to policy or admission control or other reasons, a corresponding PathErr/ResvErr/Notify is generated and sent upstream and/or downstream. An ERROR_SPEC comprises of information regarding the point of failure (network element) and details about the failure itself. When an LSP traverses multiple domains, a failure could be generated in any domain along the path and an error notification may need to be propagated across multiple domains. So, an error notification message itself may be subjected to policies as it traverses domain boundaries and a boundary node MAY modify domain specific information carried in the error message. E.g. the error node address in the RSVP ERROR_SPEC or IF_ID ERROR_SPEC or the Interface Identifier in IF_ID ERROR_SPEC may be modified at the boundary node for confidentiality reasons. Nodes other than domain boundary nodes SHOULD NOT modify ERROR_SPEC contents. It is also RECOMMENDED that such a policy be implemented only on domain boundary nodes for inter-domain LSPs where preserving confidentiality is a requirement. Also, in case of an inter-domain LSP setup failure, there may be cases when the error is not propagated all the way upstream to the head-end node. A PathErr may be intercepted by a boundary node in the domain in which the error is generated (or any other node along the path) and this node may attempt to find an alternate path. Finding an alternate path means finding a new path downstream to the node performing re-routing and avoiding the failed network elements. This may involve finding a path within the domain experiencing the failure or it may mean finding new boundary node(s) or new downstream domain. Crankback re-routing depends not only on local configuration and ability of a boundary node to do local crankback re-routing, but also on any specific parameters requested by the head-end node itself for that LSP. In certain cases, it may be desirable for the head-end node to exert some control on whether crankback re-routing at intermediate nodes is desirable or not. Procedures and extensions described in [CRANKBACK] should be followed for crankback re-routing. When crankback re-routing is allowed, a node along the TE LSP path may either decide to forward the PathErr message upstream towards the head-end node of the inter-domain TE LSP or try to determine an alternate path around the failure. When crankback re-routing is not allowed or if the node cannot perform crankback re-routing, then, on receiving a PathErr message, the node should propagate the PathErr message upstream. [RSVP-GMPLS] allows nodes to generate a targeted Notify message in addition to a PathErr/ResvErr message, to expedite error notification, if a Notify Request has been received in the corresponding Path/Resv message. This is also applicable to inter- domain TE LSPs which implement [RSVP-GMPLS]. However, since PathErr and Notify need not follow the same path, their recepient nodes could be different. This has certain implications on crankback re-routing. Procedures and recommendations in [CRANKBACK] should be followed for Notify message processing for crankback re-routing. 3.3. RRO processing across domains [RSVP-TE] defines the RECORD_ROUTE object (RRO) as an optional object, which is primarily used for loop detection and for providing information about the hops traversed by LSPs. [FAST-REROUTE] also uses the RRO to record labels and determine MP. The address or ID recorded in the RRO usually represents a link/interface. This information by itself may not be useful enough when LSPs traverse domains. [NODE-ID] defines extensions which allows a node to record its node ID in the RRO, to provide an additional context for the link address/ID. Note that there may also be cases while traversing administrative domain boundaries, where a network may not wish to expose its internal addresses/IDs to preserve confidentiality. In such cases RRO MAY be subjected to policies, filtering and modifications at domain boundaries. Internal network element identities may be masked off and replaced with boundary information or AS information, by domain boundary entities. This is not expected to hamper the working of the signaling protocol. This does, however, result in information loss, thereby leading to inefficient paths or procedures that depend on RRO information. 4. RSVP-TE signaling extensions The following RSVP-TE signaling extensions are introduced in this document. 4.1. Control of downstream choice of signaling method In certain mixed environments with different techniques (contiguous, stitched or nested TE LSPs), a head-end node of the inter-domain TE LSP may wish to signal its requirement regarding the signaling method used at an intermediate node along the path. [LSP-ATTRIBUTES] defines the format of the Attributes Flags TLV included in the LSP_ATTRIBUTES object carried in an RSVP Path message. The following bit in the Flags TLV is used by the head-end node of the inter-domain TE LSP to restrict the signaling method used by the intermediate nodes to be contiguous. Bit Number 4 (TBD): Contiguous LSP bit - this flag is set by the head-end node that originates the inter-domain TE LSP if it desires a contiguous end-to-end TE LSP (in the control & data plane). When set, this indicates that an intermediate node MUST NOT perform any stitching or nesting on the TE LSP and the TE LSP MUST be routed as any other TE LSP (it must be contiguous end to end). When this bit is cleared, an intermediate node MAY decide to perform stitching or nesting. This bit MUST NOT be modified by any downstream node. An intermdediate node that supports the LSP_ATTRIBUTES object and the Attributes Flags TLV, and also recognizes the "Contiguous LSP" bit, but cannot support contiguous TE LSP MUST send a Path Error message upstream with an error code=24 ("Routing Problem") and error sub- code="Contiguous LSP type not supported" (TBD). If an intermediate node receiving a Path message with the "Contiguous LSP" bit set in the Flags field of the LSP_ATTRIBUTES, recognizes the object, the TLV and the bit and also supports the desired contiguous LSP behavior, then it MUST signal a contiguous LSP and MUST set the "Contiguous LSP signaled" bit in the Flags field of the RRO Attributes subobject in the corresponding Resv message. However, if the intermediate node supports the LSP_ATTRIBUTES object but does not recognize the Attributes Flags TLV, or supports the TLV as well, but does not recognize this particular bit, then it SHOULD simply ignore the above request. It MAY signal any type of LSP in this case. The "Contiguous LSP signaled" bit in the Flags field of the RRO Attributes SHOULD NOT be set. An ingress node for an inter-domain LSP requesting a contiguous LSP SHOULD examine the RRO Attributes subobject Flags to determine if the LSP was indeed signaled contiguous end to end. If the "Contiguous LSP signaled" bit is cleared, the head end may take appropriate actions such as restricting the type of traffic that gets mapped to this LSP, tearing down the LSP, or rerouting the LSP around the nodes that do not support the contiguous signaling; etc. Choice of action to be taken is upto the implementation on the ingress node and it is out of the scope of this document to recommend any particular action. 5. Protection and recovery of inter-domain TE LSPs The procedures described in Section 3 MUST be applied to all inter- domain TE LSPs, including bypass tunnels, detour LSPs [FAST-REROUTE] or segment recovery LSPs [SEGMENT-PROTECTION] that may cross domains. This means that these LSPs will also be subjected to ERO processing, policies, path computation; etc. Also, note that the explicit route for these backup LSPs needs to be either configured or computed at the PLR. Just like any inter-domain TE LSP, depending on the visibility into the subsequent domain, the ERO may comprise of strict and/or loose hops. So, if there are loose hops, backup LSPs would also need to undergo loose hop expansion at nodes other than the PLR. So, the PLR in this case needs to signal the node or link that needs to be excluded for backup computation to other downstream nodes along the backup path. It is also possible that some protection schemes already signal this information in the DETOUR object([FAST-REROUTE]). However, the mechanisms for signaling this are out of scope of this document. [EXCLUDE-ROUTE] discusses one such solution to achieve this. 5.1. Fast Recovery support using MPLS TE Fast Reroute (FRR) [FAST-REROUTE] describes two methods for local protection for a packet TE LSP in case of link, SRLG or node failure. This section describes how these mechanisms work with the proposed signaling solutions for inter-domain TE LSP setup. 5.1.1. Failure within a domain (link or node failure) The mode of operation of MPLS TE Fast Reroute to protect a contiguous, stitched or nested TE LSP within a domain is identical to the existing procedures described in [FAST-REROUTE]. In case of nested or stitched inter-domain TE LSPs, protecting the intra-domain TE H-LSP or LSP segment will automatically protect the traffic on the inter-domain TE LSP. No new extensions are required for any of the signaling methods. 5.1.2. Failure of link at domain boundaries The procedures for doing link protection of the link at domain boundaries is the same for contiguous, nested and stitched TE LSPs. To protect an inter-domain link with MPLS TE Fast Reroute, a set of backup tunnels must be configured or dynamically computed between the two domain boundary nodes diversely routed from the protected inter- domain link. The region connecting two domains may not be TE enabled. In this case, an implementation will have to support the set up of TE LSP over a non-TE enabled region. For each protected inter-domain TE LSP traversing the protected link, a NHOP backup must be selected by a PLR (i.e domain exit boundary router), when the TE LSP is first set up. This requires for the PLR to select a bypass tunnel terminating at the NHOP. Finding the NHOP bypass tunnel of an inter-AS LSP can be achieved by analyzing the content of the RRO object received in the RSVP Resv message of both the bypass tunnel and the protected TE LSP(s). As defined in [RSVP- TE], the addresses specified in the RRO IPv4 subobjects can be node- ids and/or interface addresses (with specific recommendation to use the interface address of the outgoing Path messages). The PLR may or may not have sufficient topology information to find where the backup tunnel intersects the protected TE LSP based on the RRO. [NODE-ID] proposes a solution to this issue, defining an additional RRO IPv4 suboject that specifies a node-id address. 5.1.3. Failure of a boundary node For each protected inter-domain TE LSP traversing the boundary node to be protected, a NNHOP backup must be selected by the PLR. This requires the PLR to setup a bypass tunnel terminating at the NNHOP. Finding the NNHOP bypass tunnel of an inter-domain TE LSP can be achieved by analyzing the content of the RRO object received in the RSVP Resv message of both the bypass tunnel and the protected TE LSP(s) (see [NODE-ID]). The main difference with node protection, between a protected contiguous inter-domain TE LSP and a protected nested or stitched inter-domain TE LSP is that the PLR and NNHOP (MP) in case of a contiguous TE-LSP could be any node within the domain. However, in case of a nested or stitched TE-LSP the PLR and MP can only be the end-points of the H-LSP or LSP segment. The consequence is that the backup path is likely to be longer and if bandwidth protection is desired, for instance, ([FAST-REROUTE]) more resources may be reserved in the domain than necessary. Note, however, that even for a contiguous LSP, there may be cases where the addresses within the domain could have been masked in the RRO for confidentiality reasons, in which case, the RRO for the contiguous LSP may only contain boundary nodes, and so the MP can only be a boundary node. Also, while a contiguous LSP does allow backup LSPs to terminate inside the domain, there could be policies which may reject an LSP that originates in another domain from carrying addresses in ERO that are local to this domain. In these cases, the backup LSP cannot terminate inside the domain and must terminate only at the boundary node. In case of stitching or nesting, when the node to be protected is the H-LSP/S-LSP tail-end node, the PLR is not immediately upstream of this node. Hence, the failure detection time on failure of H-LSP/S- LSP tail-end node is bound to be longer than that in the case where PLR is immediately upstream of the node to be protected. In such cases, it is RECOMMENDED that the PLR rely on methods proposed in [BFD-MPLS] to rapidly detect H-LSP/S-LSP tail-end node failure. This would help in fast recovery. 5.2. Protection and recovery of GMPLS LSPs [SEGMENT-PROTECTION] describes GMPLS based segment recovery. This allows protection against a span failure, a node failure, or failure over any particular portion of a network used by an LSP. The scenarios described above for MPLS Fast reroute also apply to segment protection. No new extensions are needed for segment protection of LSPs that span multiple domains. However, in the inter-domain LSP case, it may not always be possible for the upstream node (outside a domain) to identify end-points of segment recovery LSP in another domain. Even if this was somehow determined, SERO and SRRO in the recovery LSP MUST be subjected to ERO and RRO processing rules as described above, so policy could disallow explicit control of LSP segment recovery inside the domain by a node outside the domain. This is treated as a Segment Protection failure and error handling as described in Section 4.2.1 of [SEGMENT-PROTECTION]. 6. Re-optimization of inter-domain TE LSPs Re-optimization of a TE LSP is the process of moving the LSP from the current path to a more prefered path. This usually involves computation of the new prefered path and make-before-break signaling procedures [RSVP-TE], to minimize traffic disruption. The path computation procedures involved in re-optimization of an inter-domain TE LSP are covered in [INTER-DOMAIN-PD-PATH-COMP]. Another option is to use PCE-based mechanisms ([PCE]) for re-optimization. In the context of an inter-domain TE LSP, since the LSP traverses multiple domains, re-optimization may be required in one or more domains at a time. Again, depending on the nature of the LSP and/or policies and configuration at domain boundaries (or other nodes), one may either always want the head-end node of the inter-domain TE LSP to be notified of any local need for re-optimizations and let the head-end initiate the make-before-break process or one may want to restrict local re-optimizations with the domain. [LOOSE-REOPT] describes mechanisms that allow, o The head-end node to trigger on every node whose next hop is a loose hop the re-evaluation of the current path in order to detect a potentially more optimal path. This is done via explicit signaling request: the head-end node sets the "ERO Expansion request" bit of the SESSION-ATTRIBUTE object carried in the RSVP Path message. o A node whose next hop is a loose-hop to signal to the head-end node that a better path exists. This is performed by sending an RSVP PathErr Notify message (error-code = 25), sub-code=6 (Better path exists). This indication may either be sent in response to a query sent by the head-end node or spontaneously by any node having detected a more optimal path. The above mechanisms SHOULD be used for a contiguous inter-domain TE LSP to allow the head-end node of the inter-domain TE LSP to initiate make-before-break procedures. For nested or stitched TE LSPs, it is possible to re-optimize the local H-LSP or LSP segment without involving the head-end node of the inter-domain TE LSP. This will automatically re-route the traffic for the inter-domain TE LSP along the new path, within the domain. Such local re-optimizations, including parameters for re-optimization can be controlled by local policy or configuration in that domain. 7. Security Considerations When signaling an inter-domain RSVP-TE LSP, an operator may make use of the already defined security features related to RSVP-TE (authentication). This may require some coordination between the domains to share the keys (see RFC 2747 and RFC 3097). Note that this may involve additional synchronization, should the domain boundary LSR be protected with MPLS TE Fast Reroute, since the merge point should also share the key. For an inter-domain TE LSP, especially when it traverses different administrative or trust domains, the following mechanisms (also see [INTER-AS-TE-REQS]) SHOULD be provided to an operator :- 1) a way to enforce policies and filters at the domain boundaries to process the incoming inter-domain TE LSP setup requests (Path messages) based on certain agreed trust and service levels/contracts between domains. Various LSP attributes such as bandwidth, priority; etc could be part of such a contract. 2) a way for the operator to rate limit LSP setup requests or error notifications from a particular domain. 3) a mechanism to allow policy-based outbound RSVP message processing at the domain boundary LSR, which may involve filtering or modification of certain addresses in RSVP objects and messages. Some examples of the policies described above are:- 1) An operator may choose to implement some kind of ERO filtering policy on the domain boundary LSR to disallow or ignore hops within the domain being identified in the ERO of an incoming Path message. 2) In order to preserve confidentiality, an operator may choose to disallow recording of hops within the domain in the RRO or may choose to filter out certain recorded RRO addresses at the domain boundary LSR. 3) An operator may require the boundary LSR to modify the addresses of certain messages like PathErr or Notify originated from hops within the domain. Note that the detailed specification of such mechanisms (local implementation) is outside the scope of this document. 8. IANA Considerations The following values have to be defined by IANA for this document. The registry is, http://www.iana.org/assignments/rsvp-parameters. 8.1. Attribute Flags for LSP_ATTRIBUTES object The following new flag bit is being defined for the Attributes Flags TLV in the LSP_ATTRIBUTES object. The numeric value should be assigned by IANA. Contiguous LSP bit - Bit Number 4 (Suggested value) This flag bit is only to be used in the Attributes Flags TLV on a Path message. The "Contiguous LSP" bit has a corresponding "Contiguous LSP signaled" bit (Bit Number 4) to be used in the RRO Attributes sub- object. 8.2. New Error Codes The following new error sub-codes are being defined under the RSVP error-code "Routing Problem" (24). The numeric error sub-code value should be assigned by IANA. Suggested values are, 1. Contiguous LSP type not supported - sub-code 21 2. ERO conflicts with inter-domain signaling method - sub-code 22 These error codes are to be used only in a RSVP PathErr. The following new error sub-codes are being defined under the RSVP error-code "Policy control failure" (2). The numeric error sub-code value should be assigned by IANA. Suggested values are, 1. Inter-domain policy failure - sub-code 102 2. Inter-domain explicit route rejected - sub-code 103 9. Acknowledgements The authors would like to acknowledge the input and helpful comments from Adrian Farrel on various aspects discussed in the document. 10. References 10.1. Normative References [OSPF-TE] Katz, D., Yeung, D., Kompella, K., "Traffic Engineering Extensions to OSPF", RFC 3630 (Updates RFC 2370), September 2003. [ISIS-TE] Smit, H., Li, T., "IS-IS extensions for Traffic Engineering", RFC 3784 [RSVP-TE] Awduche, et al, "Extensions to RSVP for LSP Tunnels", RFC 3209, December 2001. [RSVP-GMPLS] L. Berger, et al, "Generalized Multi-Protocol Label Switching (GMPLS) Signaling Resource ReserVation Protocol-Traffic Engineering (RSVP-TE) Extensions", RFC 3473, January 2003. [LSP-HIERARCHY] Kompella K., Rekhter Y., "LSP Hierarchy with Generalized MPLS TE", (work in progress).RFC 4206, October 2005. [LSP-STITCHING] Ayyangar A., Vasseur JP., "LSP Stitching with Generalized MPLS TE", (work in progress). [CRANKBACK] Farrel A. et al, "Crankback Signaling Extensions for MPLS Signaling", (work in progress). [LSP-ATTRIBUTES] Farrel A. et al, "Encoding of Attributes for Multiprotocol Label Switching (MPLS) Label Switched Path (LSP) Establishment Using RSVP-TE", (work in progress).RFC 4420, February 2006. 10.2. Informative References [INTER-AS-TE-REQS] Zhang et al, "MPLS Inter-AS Traffic Engineering requirements", (work in progress).RFC 4216, November 2005. [INTER-AREA-TE-REQS] LeRoux JL, Vasseur JP, Boyle J. et al, "Requirements for support of Inter-Area MPLS Traffic Engineering", (work in progress).RFC 4105, June 2005. [INTER-DOMAIN-FRAMEWORK] Farrel A. et al, "A Framework for Inter- Domain MPLS Traffic Engineering", (work in progress). [INTER-DOMAIN-PD-PATH-COMP] Vasseur JP., Ayyangar A., Zhang R., "A Per-domain path computation method for computing Inter-domain Traffic Engineering Label Switched Path", (work in progress). [PCE] Ash, G., Farrel, A., and Vasseur, JP., "Path Computation Element (PCE) Architecture", draft-ietf-pce-architecture, work in progress. [GMPLS-OVERLAY] G. Swallow et al, "GMPLS RSVPUNI: RSVP-TE Support for the Overlay Model", (work in progress).RFC 4208, October 2005. [FAST-REROUTE] Ping Pan, et al, "Fast Reroute Extensions to RSVP-TE for LSP Tunnels", RFC 4090, May 2005. [NODE-ID] Vasseur, Ali and Sivabalan, "Definition of an RRO node-id subobject", (work in progress). [SEGMENT-PROTECTION] L. Berger et al, "GMPLS Based Segment Recovery", (work in progress). [BFD-MPLS] R. Aggarwal et al, "BFD For MPLS LSPs", (work in progress). [LOOSE-REOPT] Vasseur JP. et al, "Reoptimization of an explicit loosely routed MPLS TE paths", (work in progress). Appendix 1: Examples This Appendix provides some examples to illustrate the inter-domain signaling procedures described in this document. We consider one example topology which covers inter-domain TE LSP signaling across Autonomous systems. Inter-domain TE LSP signaling across other domains covered by this document are also meant to follow similar signaling procedures, but are not covered here. <-- AS-1 ---> <--- AS-2 ---> <-- AS-3 --> <---BGP---> <---BGP--> CE1---R0----X1-ASBR1-----ASBR4--R3---ASBR7----ASBR9----R6 | | | | / | / | / | | | | | +-ASBR2----/ ASBR5 | / | | | | | | | | / | | | R1--R2----ASBR3------ASBR6--R4---ASBR8----ASBR10----R7---CE2 <================ Inter-AS TE LSP ================> 1.1 Assumptions - Three interconnected ASes, respectively AS-1, AS-2, and AS-3. Note that AS3 might be AS1 in some scenarios described in [INTER-AS-TE-REQS]. - The various ASBRs are BGP peers, without any IGP running on the single hop link interconnecting the ASBRs - Each AS runs an IGP (IS-IS or OSPF) with the required IGP TE extensions (see [OSPF-TE] and [ISIS-TE]). In other words, the ASes are TE enabled. Note that each AS can run a different IGP. - Each AS can be made of several areas. In this case, the TE LSP will rely on the inter-area TE techniques to compute and set up a TE LSP traversing multiple IGP areas. For the sake of simplicity, each routing domain will be considered as single area in this document, but the solutions described in this document does not prevent the use of multi-area techniques. In fact, these inter-domain solutions are equally applicable to inter-area TE. - An inter-AS TE LSP T1 originated at R0 in AS1 and terminating at R7 in AS3 with following possible explicit paths: o p1 - path defined by ERO with a set of loose node hops crossing AS-2, ASBR1(loose)-ASBR4(loose)-ASBR7(loose)-ASBR9(loose)-R7(loose) o p2 - path defined by ERO containing a set of strict interface hops crossing AS-2, ASBR1(loose)-link[ASBR1-ASBR4](strict)-link[ASBR4-R3](strict) -link[R3-ASBR7](strict)-link[ASBR7-ASBR9](strict)-R7(loose) o p3 - path defined by ERO containing a set of AS number hops crossing AS-2, ASBR1(loose)-link[ASBR1-ASBR4](strict)-AS3(loose)-R7(loose) - The explicit paths (EROs) may have been configured and/or computed at the head-end node using any of the path computation schemes. 1.2 ERO processing Let us consider an inter-AS TE LSP setup from R0 to R7, with example paths p1, p2. In this example, we will examine the behavior on node ASBR4 which is the boundary node for AS-2, for the different signaling methods. Contiguous:- The head-end node, R0, that desires to setup an end-to-end contiguous TE LSP, MAY originate a Path message with LSP_ATTRIBUTES object with the "Contiguous LSP" bit set in the Attributes Flags TLV. For path p1, additional computation to expand the loose hops may be required at various hops along the LSP path. When the Path message arrives at ASBR4, it may carry out a path computation or use some other means to find the intermediate hops to reach ASBR7. It may then adjust the outgoing ERO and forward the Path message through the intermediate hops in AS-2 to ASBR7. For path p2, the ERO next hop points to a node within the domain. ASBR4 will then directly forward the Path message to the next hop in the ERO. For path p3, the ERO nexthop is a loose hop pointing to the subsequent AS number. In this case, ASBR4 will perform path computation to determine the intermediate hops to reach AS-3. It may adjust the ERO based on the computation results. In this case either ASBR7 or ASBR8 may be chosen as the exit points from AS-2. Similarly, either ASBR9 or ASBR10 may be chosen as entry points into AS-3. Nesting and Stitching:- When the Path message for the inter-AS TE LSP from R0 to R7, reaches ASBR4, ASBR4 SHOULD first determine from the ERO hops, the boundary node to the domain along the path. In this example, the domain boundary node for all paths is ASBR7. It SHOULD then use the ERO hops up to ASBR7 to find an existing H-LSP in case of nesting or LSP segment in case of stitching, that satisfies the TE constraints. If there are no existing H-LSPs or LSP segments and ASBR4 is capable of setting up the H-LSP or LSP segment on demand, it SHOULD do so using the ERO hops in the Path message of the inter-domain TE LSP. In either case, ASBR4 will adjust the ERO in the inter-domain TE LSP and will forward the Path message directly to the end-point of the H-LSP or LSP segment using the procedures described in [LSP-HIERARCHY]. In case of path p1, since there are no ERO hops between ASBR4 and ASBR7, and ASBR7 hop is loose, ASBR4 may select any existing H-LSP (nesting) or LSP segment (stitching) that satisfies the constraints or it may compute a path for the H-LSP or LSP segment up to ASBR7 or some other intermediate node in AS-2. In case of path p2, ASBR4 may either select an existing H-LSP or LSP segment with ERO hops link[ASBR4-R3](strict)-link[R3-ASBR7](strict) or it may compute a new path for the H-LSP or LSP segment using the above hops. In either case, the ERO hops for the H-LSP or LSP segment MUST be the same as the signaled strict hops in that domain. Processing of path p3 is similar to p1 except that exit points from AS-2 and entry points into AS-3 are determined as part of path computation. Now, suppose, we have a path p4, defined by an ERO comprising a set of strict node hops crossing AS-2 as shown below, ASBR1(loose)-ASBR4(loose)-ASBR7(strict)-ASBR9(loose)-R7(loose) In this case, the ERO nexthop at ASBR4 is ASBR7(strict). In this case, ASBR4 will try to find or compute a H-LSP or LSP segment directly to ASBR7. The main difference between processing of p1 and p4 for nesting or stitching is that in case of p1, since the ERO nexthop is a loose hop, ASBR4 need not find a H-LSP or LSP segment directly from ASBR4 to ASBR7. So, there could be multiple H-LSPs or LSP segments between ASBR4 and ASBR7 terminating and originating on other nodes. On the other hand, for path p4, since the ERO hop to ASBR7 is a strict hop, ASBR4 MUST find or signal a H-LSP or LSP segment that directly connects ASBR4 and ASBR7. 1.3 Examples of local protection with MPLS FRR - Let us again consider the example topology above and assume that the TE LSP is now protected. Let us consider the following backup tunnels in the above example, o B1 from ASBR1 to ASBR4 following the path link[ASBR1-ASBR2](strict)-link[ASBR2-ASBR4](strict) and protecting against a failure of the ASBR1-ASBR4 link o B2 from ASBR1 to R3 following the path defined by ERO, link[ASBR1-ASBR2](strict)-link[ASBR2-ASBR3](strict)- link[ASBR3-ASBR6](strict)-link[ASBR6-ASBR5](strict)-R3(strict) and protecting against a failure of the ASBR4 node. o B3 from ASBR1 to ASBR7 following the path defined by ERO, link[ASBR1-ASBR2](strict)-link[ASBR2-ASBR3](strict)- link[ASBR3-ASBR6](strict)-ASBR7(loose) and protecting against a failure of the ASBR4 node. In this case B3 will need additional path computation (loose hop expansion) on ASBR6. o B4 from R3 to ASBR9 following the path defined by ERO, link[R3-R4](strict)-link[R4-ASBR8](strict)-ASBR9(loose) and protecting against a failure of the ASBR7 node. B4 may involve loose hop expansion on ASBR8. o B5 from ASBR4 to ASBR9 following the path defined by ERO, ASBR4-ASBR8(loose)-ASBR9(loose) and protecting against a failure of the ASBR7 node. B5 may involve loose hop expansion on ASBR8. Note that in addition to the ERO for backup tunnels, additional information regarding node/link to exclude may need to be signaled as well if backup tunnel setup involves path computation at nodes other than PLR (say for loose hop expansion). The protected inter-domain TE LSP is an inter-AS TE LSP from R0 to R7 with path p1. Also, for nesting or stitching, let us assume that the end-points of the H-LSP or LSP segment in AS-2 are ASBR4 and ASBR7. This gives rise to the following two scenarios for node protection: Protecting the boundary node at the entry to a domain :- Example: protecting against the failure of ASBR4 If the inter-AS TE LSP in this example, is a contiguous LSP, then the PLR is ASBR1 and the NNHOP (MP) could be R3 or any other intermediate node along the LSP path, if this information can be gleaned from the RRO. A backup tunnel B2 may be used to protect the inter-AS TE LSP against failure of ASBR4. However, as explained in Section 5.1.3 if RRO information related to R3 has been masked off or if there are restrictions on terminating the backup tunnel inside AS-2, then B2 cannot be used. In this case B3 may be used to protect the LSP against failure of ASBR4. If the inter-AS TE LSP in this example, is nested or stitched at ASBR4 into an intra-domain TE H-LSP or LSP segment between ASBR4 and ASBR7, then the PLR is ASBR1 and the NNHOP (MP) is ASBR7. A backup tunnel B3 may be used to protect the inter-AS TE LSP against failure of ASBR4. Protecting the boundary node at the exit of a domain :- Example: protecting against failure of ASBR7. If the inter-AS TE LSP in this example, is a contiguous LSP, then the PLR could be R3 and the NNHOP (MP) is ASBR9. A backup tunnel B4 may be used to protect the inter-AS TE LSP against failure of ASBR7. If the inter-AS TE LSP in this example, is nested or stitched at ASBR4 into an intra-domain TE H-LSP or LSP segment between ASBR4 and ASBR7, then the PLR is ASBR4 and the NNHOP (MP) is ASBR9. A backup tunnel B5 may be used to protect the inter-AS TE LSP against failure of ASBR7. Author's addresses Arthi Ayyangar Juniper Networks, Inc. 1194 N.Mathilda Ave Sunnyvale, CA 94089 USA e-mail: email@example.com Jean Philippe Vasseur Cisco Systems, Inc. 300 Beaver Brook Road Boxborough , MA - 01719 USA e-mail: firstname.lastname@example.org Intellectual Property Statement The IETF takes no position regarding the validity or scope of any Intellectual Property Rights or other rights that might be claimed to pertain to the implementation or use of the technology described in this document or the extent to which any license under such rights might or might not be available; nor does it represent that it has made any independent effort to identify any such rights. Information on the procedures with respect to rights in RFC documents can be found in BCP 78 and BCP 79. Copies of IPR disclosures made to the IETF Secretariat and any assurances of licenses to be made available, or the result of an attempt made to obtain a general license or permission for the use of such proprietary rights by implementers or users of this specification can be obtained from the IETF on-line IPR repository at http://www.ietf.org/ipr. The IETF invites any interested party to bring to its attention any copyrights, patents or patent applications, or other proprietary rights that may cover technology that may be required to implement this standard. Please address the information to the IETF at ietf- email@example.com. Disclaimer of Validity This document and the information contained herein are provided on an "AS IS" basis and THE CONTRIBUTOR, THE ORGANIZATION HE/SHE REPRESENTS OR IS SPONSORED BY (IF ANY), THE INTERNET SOCIETY AND THE INTERNET ENGINEERING TASK FORCE DISCLAIM ALL WARRANTIES, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO ANY WARRANTY THAT THE USE OF THE INFORMATION HEREIN WILL NOT INFRINGE ANY RIGHTS OR ANY IMPLIED WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. Copyright Statement Copyright (C) The Internet Society (2005).(2006). This document is subject to the rights, licenses and restrictions contained in BCP 78, and except as set forth therein, the authors retain all their rights. Acknowledgment Funding for the RFC Editor function is currently provided by the Internet Society.