Y. Suemura Internet Draft A. Kolarov Document: draft-suemura-protection-hierarchy-00.txt T. Shiragaki Expires: June 2002 NEC December 2001 Protection of Hierarchical LSPs Status of this Memo This document is an Internet-Draft and is in full conformance with all provisions of Section 10 of RFC2026. 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. Abstract In this document, we propose two different mechanisms for protection of hierarchical LSPs. We assume that an hierarchical LSP traverses a network that is partitioned into multiple smaller, non-overlapping subnetworks. Protection of the hierarchical LSP can be realized through two mechanisms: 1)Subnetwork Protection and 2)End-to-end Protection. In the subnetwork protection, a backup LSP is pre- established within each subnetwork. In the end-to-end protection, only one end-to-end backup LSP traversing the subnetworks is pre- established. These LSP protection mechanisms are applicable to both vertical and horizontal network hierarchy. In this document, we also propose procedures for applying the LSP protection mechanisms to vertical hierarchy in a single routing domain. A simple coordination mechanism for avoiding contentions between protection mechanisms at different layers is addressed as well. Suemura et al. Expires June 2002 [Page 1] Protection of Hierarchical LSPs December 2001 1. Introduction Recovery from a failure is an essential function of a transport network. There are two techniques for recovery: protection and restoration [1]. Both are implemented in either a path-based mechanism or a span-based mechanism. This document focuses on path- based mechanisms. In protection, disjoint working and backup LSPs are pre-established. When the working LSP fails, the protected traffic is switched to the backup LSP. Protection techniques can be implemented by several architectures: 1+1, 1:1, and 1:N. In case of protection of a circuit switched LSP, the 1:1 and 1:N protection architectures require signaling between two ends of the protected domain after the occurrence of a fault, while the 1+1 protection architecture does not [1]. In restoration, on the other hand, a new LSP is established for restoring affected traffic after a failure has occurred. In most cases, restoration is slower in recovering from a failure than protection since it involves path selection and rerouting for the affected traffic. In this document, we propose two protection mechanisms for hierarchical LSPs. Here, network hierarchy is considered from two perspectives [1]: (1) Vertically oriented: between two network technology layers. (2) Horizontally oriented: between two areas or administrative subdivisions within the same network technology layer. The document is organized as follows. The LSP protection mechanisms are described in Section 2. Although they are applicable to both network hierarchies, in Section 3, we describe their implementation only for a network with the vertical hierarchy and a single routing domain. Protection implementations at the same network technology layer across multiple domains are discussed in [2]. 2. LSP Protection Mechanisms We propose two LSP protection mechanisms for hierarchical LSPs: 1) Subnetwork protection, and 2) End-to-end protection. In this section, we only describe protection of a bi-directional LSP, but note that protection of a uni-directional LSP is also possible using the same mechanisms. 2.1 Subnetwork Protection In the subnetwork protection, backup LSPs are pre-established within all subnetworks that the working LSP traverses. Each backup LSP is routed from an ingress node (a node at which the working LSP enters a subnetwork) to an egress node (a node at which the working LSP exits Suemura et al. Expires June 2002 [Page 2] Protection of Hierarchical LSPs December 2001 a subnetwork). After a failure is detected at the working LSP, the protected traffic is switched from the working LSP to the backup LSP only within the subnetwork where the failure has occurred. Switching is not performed in the other subnetworks. An example of the hierarchical LSP with the subnetwork protection mechanism is shown in Figure 1. From the signaling perspective, nodes N1 and N10 are initiator and terminator nodes of the working LSP, respectively. The working LSP is routed along N1-N2-N5-N6-N9- N10 route. In subnetwork 1, a protection segment for the working LSP begins at the ingress node N1, and ends at the egress node N2. A backup LSP for this segment, namely P1B, is routed along N1-N3-N4-N2 route. Likewise, backup LSPs, P2B and P3B, for protection segments N5-N6 and N9-N10 are routed along N5-N7-N8-N6 and N9-N11-N12-N10 routes, respectively. When a node on the working LSP detects a failure, it sends failure indication alarms to the end-nodes of that protection segment, i.e. the ingress and egress nodes of the corresponding subnetwork. In case of the 1+1 protection architecture, the end nodes immediately perform a switchover from the working LSP to the backup LSP. If protection architecture requires signaling (as in the case of 1:1 and 1:N protection architectures of circuit switched LSPs), either (or both) of the end-nodes of the protection segment must initiate signaling for the switchover. Note that in the subnetwork protection mechanism, the protection segment is shorter than the working LSP, and the alarm transmission time and the signaling time are relatively short. In other words, recovery time is short. However, neither the border links of the subnetworks, such as N2-N5 and N6-N9, nor the ingress and egress nodes are protected against failure. Some extra protection mechanisms, e.g. SONET APS, are necessary to provide recovery from a fault on these links and nodes. Subnetwork 1 Subnetwork 2 Subnetwork 3 +--------------------+---------------------+--------------------+ | N1 N2 | N5 N6 | N9 N10 | | _____ _____ | _____ _____ | _____ _____ | | | | | | | | | | | | | | | | | | | |___|_____|__|__|_____|___|_____|__|__|_____|___| | | | | |P0W| | | | | | | | | | | | | | |_____| |_____| | |_____| |_____| | |_____| |_____| | | |P1B | | |P2B | | |P3B | | | __|__ __|__ | __|__ __|__ | __|__ __|__ | | | | | | | | | | | | | | | | | | | | | | | | | |__|___|__| | | | |__|___|__| | | | |__|___|__| | | | | | | | | | | | | | | | | | | | |_____| |_____| | |_____| |_____| | |_____| |_____| | | | | | | N3 N4 | N7 N8 | N11 N12 | +--------------------+---------------------+--------------------+ Figure 1 Subnetwork Protection. Suemura et al. Expires June 2002 [Page 3] Protection of Hierarchical LSPs December 2001 2.2 End-to-end Protection In the end-to-end protection, a backup LSP is pre-established between initiator and terminator nodes of a working LSP. After a failure has occurred on the working LSP, the end-to-end traffic is switched from the working LSP to the backup LSP. An example of the hierarchical LSP with the end-to-end protection mechanism is shown in Figure 2. Disjoint working and backup LSPs are routed along N1-N2-N5-N6-N9-N10 and N1-N3-N4-N7-N8-N11-N12-N10 routes, respectively. In this example, there is only one protection segment between N1 and N10 nodes. When a node on the working LSP detects a failure, it sends failure indication alarms to the initiator node N1, and the terminator node N10. If necessary, signaling between N1 and N10 nodes is done along the backup LSP. A protection segment in the end-to-end protection case is longer than the one in the subnetwork protection case. This causes longer recovery time, but on the other side, such LSP setup protects all links and nodes, excluding nodes N1 and N10, on the working LSP. Subnetwork 1 Subnetwork 2 Subnetwork 3 +--------------------+---------------------+--------------------+ | N1 N2 | N5 N6 | N9 N10 | | _____ _____ | _____ _____ | _____ _____ | | | | | | | | | | | | | | | | | | | |___|_____|__|__|_____|___|_____|__|__|_____|___| | | | | |P0W| | | | | | | | | | | | | | |_____| |_____| | |_____| |_____| | |_____| |_____| | | |P0B | | | | | __|__ _____ | _____ _____ | _____ __|__ | | | | | | | | | | | | | | | | | | | | | |__|___|_____|__|__|_____|___|_____|__|__|_____|___|__| | | | | | | | | | | | | | | | | | | | |_____| |_____| | |_____| |_____| | |_____| |_____| | | | | | | N3 N4 | N7 N8 | N11 N12 | +--------------------+---------------------+--------------------+ Figure 2 End-to-end Protection. 3. Implementation of LSP Protection Mechanisms in Case of Vertical Hierarchy In this section, we propose procedures for applying the LSP protection mechanisms shown in Section 2 to a network with vertical hierarchy and a single routing domain. A GMPLS network provides a connection from one client to another through an LSP. We call this LSP a "client LSP". If vertical hierarchy exists in a network, the client LSP may be accommodated by forwarding adjacency LSPs (FA-LSPs) [3]. Our proposal includes configuration of the client LSP that provides a connection with the desired level of survivability. Suemura et al. Expires June 2002 [Page 4] Protection of Hierarchical LSPs December 2001 In order to avoid contentions between recovery mechanisms at different layers of a vertically hierarchical network, a coordination mechanism is necessary [1]. The simplest coordination mechanism sets up LSPs in such a way that recovery is performed solely at one layer. To define this coordination mechanism, we propose the following rules for setting the Link Protection Type of an FA. 3.1 Link Protection Type The Link Protection Type represents the protection capability that exists for a link [4]. The following values are defined: Extra Traffic Unprotected Shared Dedicated 1:1 Dedicated 1+1 Enhanced We propose the following rules for setting the Link Protection Type of an FA: i. When working and backup LSPs are established, and the working LSP is advertised as an FA, the Link Protection Type of the FA is determined based on the protection mechanism used when the working and backup LSPs are established. For example, if the working and backup LSPs are established in the 1+1 protection architecture, and the working LSP is advertised as the FA, then the Link Protection Type of the FA is "Dedicated 1+1". ii. When working and backup LSPs are established, and the backup LSP is advertised as an FA, the Link Protection Type of the FA is "Extra Traffic". iii. When a single FA-LSP is established, and the FA-LSP is advertised as an FA, the Link Protection Type of the FA is found as the smallest value among the Link Protection Types of the traffic engineering (TE) links that compose the route of the FA- LSP. Here, we assume the smallest value of the Link Protection Type is "Extra Traffic" and the largest one is "Enhanced". For example, if a single FA-LSP is routed along one physical link and another FA with Link Protection Types "Dedicated 1+1" and "Dedicated 1:1" respectively, and the FA-LSP is advertised as a new FA, the Link Protection Type of the new FA is "Dedicated 1:1". 3.2 Subnetwork Protection An example of the application of the subnetwork protection in case of a network with vertical hierarchy is shown in Figure 3. The network is partitioned into three subnetworks. Subnetwork 1 and subnetwork 3 are SONET layer subnetworks with SONET crossconnect (SXC) nodes. On the other hand, subnetwork 2 is a lambda layer subnetwork with lambda Suemura et al. Expires June 2002 [Page 5] Protection of Hierarchical LSPs December 2001 crossconnect (LXC) nodes. 1+1 span protection mechanism (SONET APS) is implemented on physical links SXC2-LXC1 and LXC2-SXC5, which means that these physical links are of type "Dedicated 1+1". The other physical links are of type "Unprotected". In this network, a SONET STS-1 SPE connection, which requires "Dedicated 1+1" protection, should be established between clients C1 and C2. Signaling for setting up a client LSP, P0, between nodes C1 and C2 is initiated by node SXC1 (initiator) and terminated by node SXC6 (terminator). Before setting up P0, working and backup LSPs are established within each subnetwork, as required by the 1+1 protection mechanism. In subnetwork 1, the working LSP, P1W, and the backup LSP, P1B, are routed along SXC1-SXC2 and SXC1-SXC3-SXC4-SXC2 routes, respectively. In subnetwork 2, P2W and P2B are routed along LXC1-LXC2 and LXC1- LXC3-LXC4-LXC2 routes, respectively. Similarly, in subnetwork 3, P3W and P3B are routed along SXC5-SXC6 and SXC5-SXC7-SXC8-SXC6 routes, respectively. These LSPs are the FA-LSPs and are later used to accommodate P0. The bandwidth of the FA-LSPs must be at least as big as P0, but note that this is a little different concept from the conventional FA-LSP concept. In general, an FA-LSP is created for aggregating some LSPs, and its bandwidth must be at least as big as the sum of the LSPs that induced it. However, P1W/B, P2W/B, and P3W/B are established for protection purposes, and they do not necessarily aggregate multiple LSPs. Introduction of such FA-LSPs (FA-LSPs for protection) brings some benefits: 1) a protection segment can be placed anywhere on the protected LSP and is clearly indicated, 2) the protection granularity can be arbitrary selected, and 3) these features are realized by supporting only one protection mechanism (path protection between the end-nodes of the FA-LSPs). In this example, P1W/B and P3W/B are SONET STS-1 LSPs, and P2W/B are lambda OC-48 LSPs. P1W, P2W, and P3W form FA1, FA2, and FA3 respectively. The Link Protection Types of FA1, FA2, and FA3 are set to "Dedicated 1+1" according to the rule i defined in Section 3.1. It is important to emphasize that nodes LXC1 and LXC2 cannot aggregate/disaggregate P0 (STS-1) into/from P2W (OC-48) because their switching granularities are lambda. Therefore, it is necessary to establish a SONET STS-48 FA-LSP, P4, and form an FA, FA4, between nodes SXC2 and SXC5. FA4 is used for aggregation purpose only and does not provide protection capability by itself. However, the route of P4 is composed of TE links SXC2-LXC1, FA2, and LXC2-SXC5, all of which have the protection type of "Dedicated 1+1", and therefore, the Link Protection Type of FA4 is set to "Dedicated 1+1", according to the rule iii defined in Section 3.1. Finally, the client LSP, P0, is established. P0 is a SONET STS-1 SPE LSP and provides a connection between nodes C1 and C2. The route of P0 is composed of TE links FA1, FA4, and FA3. All of these FAs are configured with "Dedicated 1+1" protection, which is carried out at one layer only at any portion of the route. This avoids contention between recovery mechanisms at different layers. Suemura et al. Expires June 2002 [Page 6] Protection of Hierarchical LSPs December 2001 Creation of P1W/B, P2W/B, P3W/B, and P4 can be triggered administratively or by a signaling message for setting up P0. A signaling procedure in the latter case is for further study. Subnetwork 1 Subnetwork 2 Subnetwork 3 +--------------------+---------------------+--------------------+ | SXC1 SXC2 | LXC1 LXC2 | SXC5 SXC6 | | _____ _____ | _____ P2W _____ | _____ _____ | | | |P1W| | P4 | |___| | P4 | |P3W| | | C1_|_|_____+---+_____+-----+-----+ +-----+-----+_____+---+_____|_|_C2 / | | +---+ +--+--+-----+___+-----+-----+ +---+ | | P0 | |_____| |_____| | |_____| |_____| | |_____| |_____| | | | | | | | | | | | | | | | | | | _|_|_ _|_|_ | |___| P2B |___| | _|_|_ _|_|_ | | | | | |P1B| | | | | || ||___|| || | | | | |P3B| | | | | | | | +-+---+-+ | | | || | | || | | | +-+---+-+ | | | | | +---+---+---+ | | ||____|___|____|| | | +---+---+---+ | | | |_____| |_____| | |_____| |_____| | |_____| |_____| | | | | | | SXC3 SXC4 | LXC3 LXC4 | SXC7 SXC8 | +--------------------+---------------------+--------------------+ Figure 3 Example of Subnetwork Protection. 3.3 End-to-end Protection We illustrate two examples, A and B, for the case of end-to-end protection. They are shown in Figure 4 and 5. In the end-to-end protection example A, working and backup client LSPs are created to carry out protection by themselves. On the other hand, in the end- to-end protection example B, working and backup FA-LSPs are created to carry out protection. The working FA-LSP accommodates a single client LSP. The network topology is the same as in Section 3.2, except that all links have the Link Protection Type of "Unprotected". Again, we consider the case where a SONET STS-1 SPE connection, which requires "Dedicated 1+1" protection, should be established between nodes C1 and C2. In the end-to-end protection scheme A, SONET STS-48 FA-LSPs, namely P1 and P2, are created first. P1 is routed along SXC2-LXC1-LXC2-SXC5 route and it forms FA1. P2 is routed along SXC4-LXC3-LXC4-SXC7 route and it forms FA2. These FA-LSPs are not for protection but for aggregation/disaggregation of an STS-1 client LSP into/from an OC-48 lambda TE link. The routes of P1 and P2 are composed of physical links with "Unprotected" Link Protection Type. Therefore, the Link Protection Types of FA1 and FA2 are "Unprotected", according to the rule iii defined in Section 3.1. Then, SONET STS-1 SPE working and backup client LSPs, P0W and P0B, Suemura et al. Expires June 2002 [Page 7] Protection of Hierarchical LSPs December 2001 are established. P0W is routed along C1-SXC1-SXC2-FA1-SXC5-SXC6-C2 route, and P0B is routed along C1-SXC1-SXC3-SXC4-FA2-SXC7-SXC8-SXC6- C2 routed. These routes are composed of TE links with "Unprotected" Link Protection Type. The client LSPs, P0W and P0B, provide 1+1 path protection between nodes SXC1 and SXC6. When there is no fault on P0W, the connection between nodes C1 and C2 is provided by P0W. Otherwise, in case of a fault on P0W, the connection is provided by P0B. Note that recovery contention between layers does not occur because protection is carried out at one layer only. Creation of P1 and P2 can be triggered by signaling messages for setting up P0W and P0B [3]. Subnetwork 1 Subnetwork 2 Subnetwork 3 +--------------------+---------------------+--------------------+ | SXC1 SXC2 | LXC1 LXC2 | SXC5 SXC6 | | _____ _____ | _____ _____ | _____ _____ | | | | | | | | |P1 | | | | | | | | C1_|_|_____|___|_____+-----+-----+---+-----+-----+_____|___|_____|_|_C2 | | |P0W| +-----+-----+---+-----+-----+ | | | | | |_____| |_____| | |_____| |_____| | |_____| |_____| | | | | | | | | __|__ _____ | _____ _____ | _____ __|__ | | | | | | | | | |P2 | | | | | | | | | | | |__|___|_____+-----+-----+---+-----+-----+_____|___|__| | | | | |P0B| +-----+-----+---+-----+-----+ | | | | | |_____| |_____| | |_____| |_____| | |_____| |_____| | | | | | | SXC3 SXC4 | LXC3 LXC4 | SXC7 SXC8 | +--------------------+---------------------+--------------------+ Figure 4 Example of End-to-end Protection Scheme A. In the end-to-end protection example B, FA-LSPs P1 and P2 are created first. They form FA1 and FA2 respectively. Next, SONET STS-1 working and backup FA-LSPs, namely P3W and P3B, are established. P3W and P3B are FA-LSPs for protection purpose only, not for aggregation. P3W forms FA3 and is routed along SXC1-SXC2-FA1-SXC5-SXC6 route, and P3B is routed along SXC1-SXC3-SXC4-FA2-SXC7-SXC8-SXC6 route. These routes are composed of TE links with "Unprotected" Link Protection Type. Since FA-LSPs P3W and P3B provide 1+1 path protection mechanism, the Link Protection Type of FA3 is "Dedicated 1+1", according to the rule i in Section 3.1. Finally, a SONET STS-1 SPE client LSP, P0, is setup. P0 is routed along C1-FA3-C2 route, and provides a connection between nodes C1 and C2. This connection is protected between nodes SXC1 and SXC6 by 1+1 path protection mechanism, provided by P3W and P3B. Recovery contention between layers does not occur because protection is carried out at one layer only. Creation of FA-LSPs P1 and P2 can be triggered by signaling messages Suemura et al. Expires June 2002 [Page 8] Protection of Hierarchical LSPs December 2001 for establishing FA-LSPs P3W and P3B. These signaling messages can be initiated by node SXC1 before it initiates a signaling message for setting up the client LSP P0. The end-to-end protection scheme B has the following advantages over scheme A: o The protection segment can be placed at any portion of the connection and at any switching granularity. o It only requires support of a path protection mechanism between the end-nodes. The disadvantage is: o The number of required LSPs is larger than in scheme A. Subnetwork 1 Subnetwork 2 Subnetwork 3 +--------------------+---------------------+--------------------+ | SXC1 SXC2 | LXC1 LXC2 | SXC5 SXC6 | | _____ _____ | _____ P1 _____ | _____ _____ | | | |P3W| |__|__|_____|___|_____|__|__| |P3W| | | C1_|_|_____+---+-----+ | | | | | | +-----+---+_____|_|_C2 / | | +---+-----+__|__|_____|___|_____|__|__+-----+---+ | | P0 | |_____| |_____| | |_____| |_____| | |_____| |_____| | | | | | | | | | | _|_|_ _____ | _____ P2 _____ | _____ _|_|_ | | | | | |P3B| |__|__|_____|___|_____|__|__| |P3B| | | | | | | | +-+---+-----+ | | | | | | +-----+---+-+ | | | | | +---+---+-----+__|__|_____|___|_____|__|__+-----+---+---+ | | | |_____| |_____| | |_____| |_____| | |_____| |_____| | | | | | | SXC3 SXC4 | LXC3 LXC4 | SXC7 SXC8 | +--------------------+---------------------+--------------------+ Figure 5 Example of End-to-end Protection Scheme B. 4. Security Considerations No security issues are considered in this document. Suemura et al. Expires June 2002 [Page 9] Protection of Hierarchical LSPs December 2001 References [1] W. S. Lai et al., "Network Hierarchy and Multilayer Survivability," draft-ietf-tewg-restore-hierarchy-00.txt, work in progress. [2] Y. Maeno et al., "Restoration for Multi-domain Network Applications," OIF2001.574. [3] K. Kompella et al., "LSP Hierarchy with MPLS TE," draft-ietf- mpls-lsp-hierarchy-03.txt, work in progress. [4] K. Kompella et al., "Routing Extensions in Support of Generalized MPLS," draft-ietf-ccamp-gmpls-routing-01.txt, work in progress. Author's Addresses Yoshihiko Suemura NEC Corporation 4-1-1, Miyazaki, Miyamae-ku, Kawasaki, 216-8555, Japan Phone: +81-44-856-8109 Email: y-suemura@bp.jp.nec.com Aleksandar Kolarov NEC USA, Inc. 4 Independence Way, Princeton, NJ 08540, USA Phone: +1-609-951-2985 Email: kolarov@nec-lab.com Tatsuya Shiragaki NEC Corporation 1753 Shimonumabe, Nakahara-ku Kawasaki, 211-8666, Japan Phone: +81-44-396-2773 Email: t-shiragaki@cq.jp.nec.com