Protection Switching and Rerouting in MPLS

Transcription

1 Protection Switching and Rerouting in MPLS S. Veni Dr.G.M.Kadhar Nawaz Research Scholar, Director, Dept. of MCA Bharathiar University Sona College of Technology Coimbatore, India Salem, India Abstract Multi-protocol Label Switching (MPLS) has become an attractive technology of choice for Internet backbone service providers. MPLS recovery mechanisms are increasing in popularity because they can guarantee fast restoration and high QoS assurance. To provide dependable services MPLS networks make use of a set of procedures (detection, notification and fault recovery) which seek to ensure appropriate protection for the traffic carried in the label switched paths (LSPs). When a fault happens in the active LSP, the recovery scheme must redirect the traffic to a recovery path (the protection LSP or recovery LSP) which bypasses the fault. The two basic recovery models used to redirect traffic are rerouting and protection switching. Protection switching is faster than rerouting but cannot handle simultaneous faults in the active and the recovery path. On the other hand, rerouting is generally slow, and cannot offer QoS guaranties upon failure, but can use resources in a more efficient way. Keywords: MPLS, Label Switch Path, Quality of Service, Path. I. Introduction Multi-Protocol Label Switching (MPLS) is a multiservice internet technology based on forwarding the packets using a specific packet label switching technique. In MPLS networks, incoming packets are assigned a label at the ingress point. Packets are forwarded along a Label Switch Path (LSP) where each Label Switch Router (LSR) makes forwarding decisions. Each LSR re-labels and switches incoming packets according to its forwarding table. Label Switching speeds up the packet forwarding and offers new efficient and quick resilience mechanisms. In addition, the main advantage of Label Switching appears when the forwarding decision takes the Quality of Service (QoS) or links reservation into consideration. The reservation of network resources is also possible using MPLS, thus providing QoS guarantees. Two signaling protocols have been established to provide the reservation of bandwidth for the LSP, resource reservation protocol with traffic engineering extension (RSVP-TE) and constraint based routing label distribution protocol (CR-LDP). Multi Protocol Label Switching (MPLS) based recovery mechanisms introduce faster recovery schemes and will also be useful if IP networks are to evolve beyond best-effort services. MPLS-based recovery is becoming more significant for several reasons [1]. The first reason is that it is able to improve network reliability by enabling a faster response to faults than is possible with network layer methods alone while still providing the visibility of the network afforded by the network layer. Second, traditional IP rerouting may be too slow compared to a core MPLS network which requires recovery times smaller than times achieved by IP routing protocols. Furthermore, MPLS-based recovery schemes can be optimal to support the traffic engineering goal of optimal use of resources. schemes in MPLS can also introduce bandwidth protection to specific flow such as VoIP. The paper is organized in the following manner. In the next section some definitions pertinent to MPLS-base recovery approaches are presented. In sections 3 and 4 a review of several recovery schemes based on protection switching and on rerouting, respectively, is presented. II. MPLS Mechanism MPLS requires a set of procedures to provide protection of the traffic carried on different 216

2 paths. This requires that the label switching routers (LSRs) support fault detection, fault notification, and fault recovery mechanisms, and that MPLS signaling support the configuration of recovery. Fault detection is the time required to detect a fault; the quicker the fault is detected in the network the earlier the fault can be repaired. The current method of detecting faults on a link comes from the use of periodic messages, if the messages are absent then the link is assumed to be faulty and fault notification is initiated. Fault notification is the process of informing other LSRs of the failure that has occurred which eventually propagates to ingress LSRs to initiate recovery procedures to maintain traffic flows. The two basic recovery models used to redirect traffic are rerouting and protection switching; recovery can also be global or local, and resource or path oriented. When rerouting is used the Path (RP) (the path by which the traffic is restored after the occurrence of a fault) is signaled only upon fault detection in the Active Path (AP) (the path which carries traffic before the occurrence of a fault). Protection switching pre establishes a RP before any failure detection in the protected AP. Global recovery intents to protect against any link or node fault in a path, whereas local recovery intents to protect against a link or node fault and to minimize the time required for fault propagation. Local recovery is attempted by the node immediately upstream of the fault according to [2]. Global recovery is usually slower than local recovery because the failure notification message has to travel to the Point of Repair (POR) (a LSR that is setup for performing MPLS recovery). When a failure occurs the LSR responsible for switching or replicating the traffic between the AP and RP is the Path Switch LSR (PSL).The Path Merge LSR (PML) is the LSR responsible for receiving the RP traffic. If the PML is not the egress LSR it will merge the RP traffic back onto the AP. The POR can be a PSL or PML depending on the type of recovery scheme employed. When the node immediately upstream of the fault is unable to recover a failed path (either because that node is not a POR or because the POR was unsuccessful in its attempt), and sends a Fault Indication Signal (FIS) to the node immediately upstream, where possibly a new attempt (of local) recovery takes place, we consider this to be a mechanism of local recovery (even if the node that successfully recovers the AP coincides with the head-end node) - this type of recovery appears in [3, 4] and in [5] (where is designated as segment protection). If a recovery mechanism tries to protect a particular network element (link and/or node) we consider it to be resource oriented (depending on a particular scheme, not every AP using a protected network element is required to be under protection by this mechanism). Whenever the recovery mechanism is focused on the recovery of a particular AP, we consider it to be path oriented. III. Protection Switching Protection switching provides a mechanism for fast recovery in MPLS networks, but inefficient utilization of network resources occur, as protection paths are unused in the absence of faults. Protection switching also requires protection paths to be pre-established before failures occurs. This can lead to suboptimal protection paths as the network changes over time. Depending on the place where the reaction to failures is done, protection switching mechanisms can be distinguished into end-to-end and local protection. In case of end-to-end protection switching the reaction to a failure along a path is executed at the path ingress router. End-to-end protection switching is faster than restoration methods, but the signaling of the failure to the path ingress router takes time within which traffic is lost. Local protection schemes tackle the problem of lost traffic in case of end-to-end protection. Backup paths towards the destination are set up not only at the ingress router of the primary path but at almost every node of the path. Then, a backup path is immediately available if the path breaks at some location. Local protection switching can be implemented by MPLS fast reroute (MPLS- FRR) [6].Kang and Reed [7] presented a scheme dedicated to protection using bandwidth guaranteed bypass tunnels. The choice of p- cycles (preconfigured-cycles) was considered a good option for bypass tunnels topology. P- cycles [8] pre-configure extra network capacity in cycles, before any fault occurs. This allows the recovery not only of faults on the cycle but also of links connecting any two non-adjacent nodes in the "p-cycle" (straddling links) which contributes significantly for the p-cycles efficiency. In [7] an adaptation of the p-cycles optimal design [9] with guaranteed bandwidth (to protect against single link failures) was proposed. In [10] Kodialam and Lakshman's recovery scheme, upon a new LSP request, both 217

3 the working and recovery path are determined, on-line and simultaneously. If the requested bandwidth cannot be guaranteed (for the AP and RP), the request is rejected. To function, the scheme requires knowledge of the total bandwidth used by the APs, and of total bandwidth used by the RPs in each link (just the aggregated information, not broke down for each LSP) - and also of the free bandwidth in each link. The authors call this Partial Information (PI) model. Using this model, a dynamic routing heuristic algorithm to achieve local recovery is presented, computing both the active path and recovery paths. This scheme handles single link failures, but the authors showed it can easily be extended to handle also single node failures. In [11], Kuang et al. presented a scheme for minimizing the delay of notification messages that must be sent (upon detection of a fault) to the PSL. This scheme defines a special notification structure named Reverse Notification Tree (RNT) to distribute efficiently the fault and/or recovery information. The scheme also proposes a faster Hello protocol for fault detection and a light and scalable transport protocol for handling the notification messages In Kim's paper [l2] a numerical analysis of load distribution for global recovery, assuming equal fault probability of each link, is presented. This analysis considers two mechanisms: the Fully Shared Mechanism (FSM), where all paths may be used simultaneously as APs, doubling as RPs when needed, and the Partially Shared Mechanism (PSM), where some paths are used as APs and others are set aside to use as RPs. This study enforces a constraint (not a very realistic one) of homogeneity on the network traffic and reserved bandwidth. On Yoon et al.'s [13], for each working path, the ingress LSR and each intermediate LSRs with recovery capability pre-calculate a recovery path to the nearest downstream LSR. These paths are updated whenever network state changes (link usage information) arrive at the path computing LSRs. This scheme assumes that all the possible faults are either failure or degraded usage of a single link in a single MPLS protection domain. Facing such a fault, the PSL will be the nearest upstream LSR with recovery capability. If the direct upstream LSR does not have such capabilities, a FIS will be sent through a RNT. Since the intermediate LSRs can only handle single link failures, a FIS (if sent) will always be propagated to the ingress LSR. To update the recovery path, the LSRs use network state information, which requires an extension to the used Interior Gateway Protocol (IGP) (e.g OSPF or IS-IS) [1]. Park et al. [14] proposed a pre-qualified mechanism for MPLS networks named dynamic path protection, in order to quickly recover node or link failures. Upon the fault detection by an LSR, a recovery path is selected among the existing (non faulty) working paths that start or transit there and with the same destination (if there are several paths to choose from, the paper proposes some ranking criteria). If later a fault occurs in the previously selected recovery path, the same procedure is used again. It is a fast scheme because it avoids signaling a new path, and requires mere routing table changes. However, if no such paths can be found, the LSR must create a new path (from itself up to the original destination). It is claimed in [14] that this scheme does not require signaling protocol extensions. Table I gives a comparison of the recovery models. IV Rerouting The rerouting model establishes the recovery path only after a fault has occurred. Optimization of recovery paths can be achieved as the current network state is considered before recovery path selection. The rerouting model can be further divided into two types, establish-on-demand and pre-qualified. Establish-on-demand calculates and establishes a recovery path only after a fault is detected. Pre-qualified implies that a recovery path is already calculated but is only established when a fault occurs. It has the advantages of efficient resource utilization as bandwidth is not reserved and speed up recovery as path selection is completed before faults occur. 218

4 Model Rerouting Protection Switching Table I Comparison of the recovery models. Path Type Time Pre- Qualified Establish- On- Demand 1+1 1:1, 1:n, m:n Path Setup Point After fault Before fault Resource Utilization Optimization * * * * V. Conclusion There are two main methods of recovery, protection switching and rerouting. Protection switching provides fast restoration of service but lacks efficient use of network resources in MPLS networks. Rerouting has the advantage of optimizing the recovery paths, allowing for more working or recovery LSP requests. The disadvantage is slower restoration of service. Other approaches to improving service restoration include using concepts such as reverse notification tree (RNT) that aims to improve the reliability and efficiency of fault notification. Protection switching and rerouting both have their complementary advantages and disadvantages; the choice of recovery method should be based on the type of protection required. References [1] Paul Meyers, Natalie Degrande, Sven Van den Bosc. (2009) Alcatel- Lucent Telecom Review. [Online]. HYPERLINK " Com [2] V. Sharma, F. Hellstrand, B. Mack-Crane, S. Makam, K. Owens, C. Huang, J. Weil, B. Cain, L. Anderson, B. Jamoussi, A. Chiu, and S. Civanlar. framework for multi-protocol label switching (MPLS)-based recovery. IETF RFC 3469, February [3,4] D.-K. Hong, C. S. Hong, and Dongsik-Yun. A hierarchical restoration scheme with dynamic adjustment of restoration scope in an MPLS network. In Network Operations and Management Symposium, pages , April [5] D. Xu, Y. Xiong,, and C. Qiao. Novel algorithms for shared segment protection. IEEE Journal on Selected Areas in Communication, 21(8): , [6] P. Pan, G. Swallow, and A. Atlas, RFC4090: Fast Reroute Extensions to RSVP-TE for LSP Tunnels, May [7]. Kang and M. J. Reed. Bandwith protection in MPLS networks using p-cycle structure. In Design of Reliable Communication Networks (DRCN) 2003, pages , Banff, Alberta, Canada, October [8] W. D. Grover and D. Stamatelakis. Cycle-oriented distributed preconfiguration: Ring-like speed with mesh-like capacity for self-planning network restoration. In Proceedings of IEEE ICC'98, pages , Atlanta, Georgia, June [9] D. Stamatelakis and W. D. Grover. IP layer restoration and network planning based on virtual protection cycles. IEEE Journal on Selected Areas in Communications, 18(10): , October [10] M. Kodialam and T. V. Lakshman. Dynamic routing of locally restorable bandwidth guaranteed tunnels using aggregated link usage information. In Proceedings of IEEE INFOCOM 2001, pages , April [11]C. Huang, V. Sharma, K. Owens, and S. Makam. Building reliable MPLS networks using a path protection mechanism. IEEE Communications Magazine, pages , March

5 [12] S.-Y. Kim. Effect of load distribution in path protection of MPLS. International Journal of Communication Systems, 16(4): , February [13] S. Yoon, H. Lee, D. Choi, Y. Kim, G. Lee, and M. Lee. An efficient recovery mechanism for MPLSbased protection LSP. In Joint 4th IEEE International Conference on ATM (ICATM 2001), pages 75-79, Seoul, Korea, April [14]. Park, H.-S. Yoon, S. C. Kim, J. Park, and S. Yang. Design of a dynamic path protection mechanism in MPLS networks. In The 6th Ihternataonal Conference on Advanced Communication Technology, volume 2, pages , [15] T. Kodialam, M.; Lakshman, Minimum interference routing with applications to mpls traffic engineering, in INFOCOM Nineteenth Annual Joint Conference of the IEEE Computer and Communications Societies. Proceedings. IEEE, vol. 2, 2000, pp vol.2. First S.Veni I have presented eight papers in National conferences and one in international conferences. I have completed M.Sc., MPhil in Computer Science and currently working as a Asst Professor in Karpagam University. I have seven years of experience in teaching. I am Pursuing Doctorate degree in Computer Science under the guidance of Dr.G.M.Kadhar Nawaz, who is working as Director in Department of Computer Applications in Sona College of Technology,Salem. My area of research is Traffic Engineering in Multiprotocol Label Switching environment. Second Dr.G.M.Kadhar Nawaz I have presented and published papers in various national and international Conferences and journals. I have also organized national conferences. I completed my Ph.D in Computer Science from Periyar University and my area of research includes network security and network communications. Currently I am working as Director in the Department of Computer Applications, Sona College of Technology, Salem. 220