Internet Engineering Task Force Y. Shen Internet-Draft Juniper Networks Intended status: Standards Track Y. Kamite Expires: September 6, 2012 NTT Communications Corporation March 5, 2012 RSVP Setup Protection draft-shen-mpls-rsvp-setup-protection-00 Abstract RFC 4090 specifies an RSVP facility-backup fast reroute mechanism that can protect LSPs against link and node failures. This document extends the mechanism to provide "setup protection" for LSPs during initial Path message signaling time. In particular, it enables a router to reroute an LSP via a bypass LSP, when there is a link or node failure along the desired path. Status of this Memo 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 http://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 6, 2012. Copyright Notice Copyright (c) 2012 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 (http://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 Shen & Kamite Expires September 6, 2012 [Page 1] Internet-Draft RSVP Setup Protection March 2012 the Trust Legal Provisions and are provided without warranty as described in the Simplified BSD License. Table of Contents 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 3 2. Specification of Requirements . . . . . . . . . . . . . . . . 4 3. Theory of Operation . . . . . . . . . . . . . . . . . . . . . 4 3.1. New RSVP Attributes TLVs . . . . . . . . . . . . . . . . . 5 3.1.1. Protected LSP Sender IPv4 Address TLV . . . . . . . . 5 3.1.2. Protected LSP Sender IPv6 Address TLV . . . . . . . . 6 3.2. PLR behavior . . . . . . . . . . . . . . . . . . . . . . . 6 3.3. MP behavior . . . . . . . . . . . . . . . . . . . . . . . 8 3.4. Local Revertive Mode . . . . . . . . . . . . . . . . . . . 9 4. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 9 5. Security Considerations . . . . . . . . . . . . . . . . . . . 9 6. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . 9 7. References . . . . . . . . . . . . . . . . . . . . . . . . . . 9 7.1. Normative References . . . . . . . . . . . . . . . . . . . 9 7.2. Informative References . . . . . . . . . . . . . . . . . . 10 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . . 10 Shen & Kamite Expires September 6, 2012 [Page 2] Internet-Draft RSVP Setup Protection March 2012 1. Introduction In RSVP facility-backup fast reroute (FRR) [RFC 4090], the router at a point of local repair (PLR) of an LSP can redirect traffic onto a bypass LSP upon a failure of the immediate downstream link or node. The establishment of such protection is normally triggered by receiving a Resv message. In link protection, the PLR must learn the label and address of the nexthop router, before it can set up or select a bypass LSP to protect the LSP. Likewise, in node protection, the PLR must learn the label and address of the next- nexthop router. The information is carried in Resv message. Imagine a scenario where an LSP is being signaled, but its Path message carries an EXPLICIT_ROUTE object (ERO) that is pre-computed, statically configured, or computed based on a topology that may not reflect the current state of every link or node of the network. If a link or node on this path has already failed, the signaling will halt at the router immediate upstream of the failure. This will still be the case even if there is an existing bypass LSP protecting the link or node for some other P2P or P2MP LSP. In other words, the LSP is not protected during initial Path message signaling time. In this situation, the network would rely on IGP to flood the up-to- date traffic engineering (TE) information, and the router immediate upstream of the failure to originate a PathErr message, so that the ingress router of the LSP can compute and signal a new path to avoid the failed link or node. However, this approach may not always be possible or desirable, as can be seen in the scenarios described below. 1. Pre-computed or explicitly defined paths. If the path is pre- computed or fixed, or the ingress router is incapable of re- computing the path, an alternative path will not be set up. 2. Requirements for LSP setup time. Control protocol convergence and path computation may introduce a significant delay, which may impact signaling performance for services that have a specific requirement for LSP setup time. 3. Sibling sub-LSPs of a P2MP LSP sharing the failed link. In this case, the existing bypass LSP will not be used, even if it is the preferred alternative path. For example, the LSP being signaled is a sub-LSP of a P2MP LSP, and it is expected to share the same downstream link with an existing sibling sub-LSP (sub-LSP of the same P2MP LSP). If the new sub-LSP is rerouted through another path, unnecessary traffic will be generated on the network. This document extends the RSVP facility-backup fast reroute mechanism Shen & Kamite Expires September 6, 2012 [Page 3] Internet-Draft RSVP Setup Protection March 2012 to provide so-called "setup protection" for LSPs. During the initial Path message signaling of an LSP, if there is a link or node failure on the desired path, and there is a bypass LSP protecting the link or node, the LSP will be signaled through the bypass LSP. The LSP will be established as if it was originally set up along its primary path and then failed over to the bypass LSP after the link or node failure. When the link or node is restored, the LSP MAY be reverted to the primary path. The mechanism supports both P2P and P2MP LSPs. 2. Specification of Requirements 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. 3. Theory of Operation When an LSP is being signaled by RSVP, a Path message is sent hop by hop from the ingress router to the egress router, following the path defined by an ERO. The setup protection mechanism in this document allows an ingress or transit router to reroute the LSP via a bypass LSP, if the router detects a failure of the immediate downstream link or node represented by the next hop in the ERO (i.e. next ERO hop). This router is referred to as a PLR. The mechanism is relevant when the Path message carries the "local protection desired" flag, and optionally the "node protection desired" and "bandwidth protection desired" flags in the SESSION_ATTRIBUTE object. On a given PLR, the mechanism is only applicable when the next ERO hop is a strict hop, and in case of node protection, the next-next ERO hop is also a strict hop. A strict next ERO hop allows the PLR to unambiguously decide the intended downstream link or node, and hence reliably detect its status. For link protection, the strict nexthop also indicates the merge point (MP), i.e. the egress router of the bypass LSP to be used for rerouting the LSP. For node protection, the strict next-next ERO hop indicates the MP. During setup protection operation, the PLR signals a backup LSP by tunneling a Path message through the bypass LSP. Unlike the normal facility-backup FRR, this Path message carries additional information of the protected LSP (Section 3.1). When the MP receives the Path message, it terminates the backup LSP, and also re-creates the protected LSP. If the MP is a transit router of the protected LSP, it signals the LSP further downstream. Shen & Kamite Expires September 6, 2012 [Page 4] Internet-Draft RSVP Setup Protection March 2012 Eventually, the LSP is established end to end, with the backup LSP being tunneled through the bypass LSP from the PLR to the MP. The RSVP states on the PLR and the MP and the RSVP messages generated by these routers are the same as those in a normal facility-backup FRR scenario. After the failed link or node is restored, the PLR MAY revert the LSP to the primary path. This is referred to as local revertive mode, as described in [RFC 4090]. The setup protection mode MAY be enabled and disabled on a router based on configuration. For an LSP to be setup-protected, the mode MUST be enabled on both PLR and MP. If it is enabled on a PLR but disabled on an MP, the MP SHOULD reject the Path message of the backup LSP and send a PathErr message, as described Section 3.3. 3.1. New RSVP Attributes TLVs This document defines two new RSVP Attributes TLVs [RFC 5420]. They are used by a PLR to convey to an MP the original sender address of a protected LSP. o Protected LSP Sender IPv4 Address TLV o Protected LSP Sender IPv6 Address TLV Both TLVs are carried by the LSP_REQUIRED_ATTRIBUTES object in the Path message of a backup LSP. 3.1.1. Protected LSP Sender IPv4 Address TLV The Protected LSP Sender IPv4 Address TLV is defined with type 2. It is allowed on LSP_REQUIRED_ATTRIBUTES object, and not allowed on LSP_ATTRIBUTES object. It is encoded as the following. 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 (TBD) | Length (8) | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Value | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ Figure 1 Type Shen & Kamite Expires September 6, 2012 [Page 5] Internet-Draft RSVP Setup Protection March 2012 TBD Length 8 Value Original sender address in the IPv4 SENDER_TEMPLATE object of the protected LSP. 3.1.2. Protected LSP Sender IPv6 Address TLV The Protected LSP Sender IPv6 Address TLV is defined with type 3. It is allowed on LSP_REQUIRED_ATTRIBUTES object, and not allowed on LSP_ATTRIBUTES object. It is encoded as the following. 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 (TBD) | Length (20) | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | | // Value // | | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ Figure 2 Type TBD Length 20 Value Original sender address in the IPv6 SENDER_TEMPLATE object of the protected LSP. 3.2. PLR behavior When a router has a Path message to send out and the next ERO hop is a strict IPv4 or IPv6 prefix, the router validates the hop against the routing table, traffic engineering (TE) database, and/or a Shen & Kamite Expires September 6, 2012 [Page 6] Internet-Draft RSVP Setup Protection March 2012 topology database. If the hop is reachable and one hop away from the router, the Path message is sent as it is. Otherwise, there is a possibility that the hop has experienced a link or node failure. The router determines this by searching for an existing bypass LSP that is protecting the hop. If the LSP being signaled desires link protection, the egress router of the bypass LSP (i.e. MP) must be the router that owns the IP prefix of the hop. If the LSP desires node protection, the next-next ERO hop of the LSP must also be a strict IP prefix, and the MP must be the router that owns this IP prefix. If a bypass LSP is not found, the router MUST originate a PathErr with code = 24 (routing problem) and sub-code = 2 (bad strict node). If a bypass LSP is found, the router MUST act as a PLR of setup protection, and reroute the LSP via the bypass LSP. If multiple such bypass LSPs exist, the PLR MAY select one based on bandwidth constraints or policy. If the protected LSP is a sub-LSP of a P2MP LSP, a bypass LSP that is protecting an existing sibling sub-LSP MUST be preferred. This helps against generating duplicate traffic on two separate bypass LSPs. The PLR SHOULD NOT send the Path message of the protected LSP. Instead, it MUST create a backup LSP, and send a Path message for the backup LSP via the bypass LSP. The Path message is constructed by using the sender template specific method [RFC 4090]. In particular, it has the sender address in SENDER_TEMPLATE object set to an address of the PLR. It MUST also carried a LSP_REQUIRED_ATTRIBUTES object containing a Protected LSP Sender IPv4 Address TLV or Protected LSP Sender IPv6 Address TLV. Upon receiving a Resv message from the MP, the PLR brings up both of the backup LSP and the protected LSP. If the PLR is the ingress router of the protected LSP, the LSP has been set up successfully. If the PLR is a transit router, it MUST send a Resv message upstream, with the "local protection available", "local protection in use", and optionally "node protection" and "bandwidth protection" flags set to 1 in the RRO hop corresponding to the PLR [RFC 4090]. The PLR MUST originate a PathErr message with code = 25 (notify error) and sub- code = 3 (tunnel locally repaired). The PLR also installs a forwarding entry for the LSP. The nexthop of this entry MAY indicate zero, one or two outgoing labels, depending on whether any of the backup LSP's label and the bypass LSP's label is Implicit NULL. In the case of two labels, the inner label is the backup LSP's label, and the outer label is the bypass LSP's label. Shen & Kamite Expires September 6, 2012 [Page 7] Internet-Draft RSVP Setup Protection March 2012 If the PLR receives a PathErr message when signaling the backup LSP, the PLR MUST NOT bring up the backup LSP or the protected LSP. If the PLR is a transit router of the protected LSP, it MUST propagate the PathErr message upstream. Likewise, if the PLR receives a PathErr message after the backup LSP and the primary LSP have been set up, and the PLR is a transit router of the protected LSP, it MUST also propagate the PathErr message upstream. When the PLR receives a ResvTear message of the backup LSP, the PLR MUST bring down both the backup LSP and the protected LSP. If the PLR is a transit router of the protected LSP, it MUST send a ResvTear message upstream. In any case where the PLR tears down the protected LSP due to receipt of a PathTear message, state time-out, configuration, etc, the PLR MUST also tear down the backup LSP by sending a PathTear message through the bypass LSP. 3.3. MP behavior When a MP receives the Path message of a backup LSP, it detects the setup protection condition based on the presence of Protected LSP Sender IPv4 Address TLV or Protected LSP Sender IPv6 Address TLV in LSP_REQUIRED_ATTRIBUTES object. If the setup protection mode is disabled on the router, it MUST reject the Path message. In this case, the router recognizes the TLV, but does not support it. Therefore, it MUST originate a PathErr with code = 2 (policy control failure). The MP then terminates the backup LSP, and re-creates the protected LSP. If the MP is the egress router of the LSP, it MUST also terminate the protected LSP. Otherwise, it MUST send a Path message of the protected LSP downstream. The Path message has the sender address in SENDER_TEMPLATE object set to the original address of the ingress router, based on the above received TLV. The Path message MUST NOT carry a Protected LSP Sender IPv4 Address TLV or Protected LSP Sender IPv6 Address TLV. The MP MUST allocate a label for the LSP, and distribute it to the PLR via the Resv message of the backup LSP. If the protected LSP is a sub-LSP of a P2MP LSP, the MP MAY allocate the same label as an existing sibling sub-LSP, in order to avoid traffic duplication. When the MP receives a PathTear message of the backup LSP, it MUST tear down both the backup LSP and the protected LSP. If the MP is a transit router of the protected LSP, it MUST send a PathTear message downstream. Shen & Kamite Expires September 6, 2012 [Page 8] Internet-Draft RSVP Setup Protection March 2012 In any case where the MP receives or originates a PathErr or ResvTear message of the protected LSP, it SHOULD translate it into a message of the backup LSP and send it to the PLR. 3.4. Local Revertive Mode When the failed link or node is restored, the PLR MAY revert the protected LSP to its primary path, following the procedure of local revertive mode described in [RFC 4090]. 4. IANA Considerations This document defines two new RSVP Attributes TLVs. New type values need to assigned to them by IANA. Protected LSP Sender IPv4 Address TLV Protected LSP Sender IPv6 Address TLV 5. Security Considerations The security considerations discussed in RFC 3209, RFC 4090 and RFC 4875 apply to this document. 6. Acknowledgements Thanks to Rahul Aggarwal, Disha Chopra, and Nischal Sheth for their contribution. 7. References 7.1. Normative References [RFC2205] Braden, B., Zhang, L., Berson, S., Herzog, S., and S. Jamin, "Resource ReSerVation Protocol (RSVP) -- Version 1 Functional Specification", RFC 2205, September 1997. [RFC3209] Awduche, D., Berger, L., Gan, D., Li, T., Srinivasan, V., and G. Swallow, "RSVP-TE: Extensions to RSVP for LSP Tunnels", RFC 3209, December 2001. [RFC4090] Pan, P., Swallow, G., and A. Atlas, "Fast Reroute Extensions to RSVP-TE for LSP Tunnels", RFC 4090, May 2005. Shen & Kamite Expires September 6, 2012 [Page 9] Internet-Draft RSVP Setup Protection March 2012 [RFC5420] Farrel, A., Papadimitriou, D., Vasseur, JP., and A. Ayyangarps, "Encoding of Attributes for MPLS LSP Establishment Using Resource Reservation Protocol Traffic Engineering (RSVP-TE)", RFC 5420, February 2009. [RFC4875] Aggarwal, R., Papadimitriou, D., and S. Yasukawa, "Extensions to Resource Reservation Protocol - Traffic Engineering (RSVP-TE) for Point-to-Multipoint TE Label Switched Paths (LSPs)", RFC 4875, May 2007. [RFC3471] Berger, L., "Generalized Multi-Protocol Label Switching (GMPLS) Signaling Functional Description", RFC 3471, January 2003. [RFC3472] Ashwood-Smith, P. and L. Berger, "Generalized Multi- Protocol Label Switching (GMPLS) Signaling Constraint- based Routed Label Distribution Protocol (CR-LDP) Extensions", RFC 3472, January 2003. [RFC3031] Rosen, E., Viswanathan, A., and R. Callon, "Multiprotocol Label Switching Architecture", RFC 3031, January 2001. 7.2. Informative References [RFC5920] Fang, L., "Security Framework for MPLS and GMPLS Networks", RFC 5920, July 2010. Authors' Addresses Yimin Shen Juniper Networks 10 Technology Park Drive Westford, MA 01886 USA Phone: +1 9785890722 Email: yshen@juniper.net Yuji Kamite NTT Communications Corporation Granpark Tower 3-4-1 Shibaura, Minato-ku Tokyo 108-8118 Japan Email: y.kamite@ntt.com Shen & Kamite Expires September 6, 2012 [Page 10]