Internet Draft D. Papadimitriou Document: draft-pbh-packet-optical-escalation-01.txt J. Jones Category: Internet Draft Alcatel Expires: November 2001 Riad Hartani Caspian Debashis Basak Accelight May 2001 Packet-Optical Escalation Strategies draft-pbh-packet-optical-escalation-01.txt Status of this Memo This document is an Internet-Draft and is in full conformance with all provisions of Section 10 of RFC2026 [1]. The following verbatim statement must follow the optional opening sentence: 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. 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 [2]. Abstract This memo focused on hybrid packet and optical layer restoration architectures. We start with an initial description of general restoration related aspects, analyze various restoration schemes and focus on efficient inter-working between packet and optical layer Papadimitriou D.(Editor) - Expires November 2001 1 draft-pbh-packet-optical-escalation-01.txt May 2001 restoration mechanisms. This analysis results in potential control protocols enhancements required to achieve the fast-restoration targets. 1. Introduction - Scope In the context of packet-over-optical networks, also known Packetû over-Wavelength (POW) networks, the restoration approach depends on the service being offered. Restoration can happen either at the packet layer, the optical layer or both. In the latter case, coordination between the various restoration schemes is required to maximize efficiency and to avoid conflicting actions occurring at different layers. The scope of this memo is to define and extend the proposed mechanisms for a packet over all-optical network where packet capable interfaces are only available on O/E/O capable devices. 2. Restoration Strategies Prior to restoration, fault detection and fault location must be performed. This can happen at the packet and/or optical layer. Nevertheless, fault detection and location can be performed simultaneously (within the same layer). Various mechanisms can be used to achieve this (some of them may be extended for the hybrid packet/optical case). Following detection and notification of the failure to the nodes responsible for restoration, restoration is performed. In all possible cases, we assume that both the packet and optical layers have their own control entities (in a general case, these are instantiations of various signaling, routing and management protocols). No assumption is made on the physical location of these controlling entities. They could be centralized or distributed, with dynamic or static behavior. For efficient Packet-Optical interaction, a well-defined message exchange between the packet and optical control entities is required. The so-called integrated Packet-Optical control entity, similar to the concept of a Packet Optical Network Controller (PONC) should be informed as soon as possible of the failures and their location. The PONC may or may not be involved in the restoration based on the selected scheme, but can be responsible for overall network re-optimization once the restoration is performed. Optical network re-optimization implies to define what a Lambda- Packet LSP restoration mechanism requires in terms of restoration time, spare resources, mechanism robustness, capabilities, etc. Consequently, we need to first define what re-optimization means by keeping the topological information prior to the various restoration processes performed. This in turn requires to define which are the constraints to keep the same topology that the one prior to the failure or take into account the updated "availability" values into Papadimitriou D. (Editor) - Expires November 2001 2 draft-pbh-packet-optical-escalation-01.txt May 2001 account in order to have a new topology (in fact this could be mapped to a dampening mechanism). Our analysis covers different scenarios, where there may be various intermediate layers between the Packet and Optical Layers. Since a common goal in optical network development is to simplify the layering (to have the Packet Layer directly over the Optical Layer), we examine a strategy where only the Packet and Optical Layers participate in the restoration process. This method can still apply in cases where some intermediate layers exist. All of these scenarios consider cases where failures occur at the Optical Channel Layer (i.e. Lambda Layer) or the Physical Layer (i.e. Fiber Layer). They can be extended to include Packet Layer LSP failures, which is the ultimate goal of the top-down strategy (see Section 3). Also, these scenarios consider failures within the optical network and not packet to optical ôaccessö failures (physical connectivity between packet and optical device) or intermediate layers such as ATM (Layer2) and Sonet/SDH (Layer1). In this case, alternative strategies may be used at both the packet and optical layer. Most of these intermediate layers will provide M:N, 1:N and/or 1:1 restoration mechanisms or have an access path protection while relying on Sonet/SDH overhead for failure detection. If neither the Sonet/SDH nor the Optical Channel Layer (i.e. Lambda Layer) provide a restoration mechanism then Layer3 should handle the restoration. The approach should be Sonet/SDH or even G.709 independent since the target is to address packet over lambda switching networks (i.e. Sonet/SDH is only used as a framing procedure). It should also define the appropriate mechanisms to allow the packet and optical layers to efficiently cooperate while providing the required restoration. This implies that the control plane mechanisms currently defined in [GMPLS-SIG] need to be extended. 3. Escalation Strategies 3.1 Scenarios The various scenarios analyzed here are described as follows: 1. The optical layer handles the restoration: in this case, the restoration will be transparent to the packet layer. This is generally achieved by having pre-provisioned (or not) optical paths in a 1+1, 1:1 or 1:N scheme within the optical network or establishing them when required. Refer for instance to [IPO-REST] where optical-layer restoration mechanisms are fully detailed. 2. The packet layer handles the restoration. In this case, we assume that the packet layer is made aware of the optical channel or fiber layer failure within the optical network through GMPLS-based Notification like the one proposed in [GMPLS-RSVP] or [GMPLS- CRLDP] for instance or through PONC for example. Under this assumption, the packet layer handles the restoration by putting Papadimitriou D. (Editor) - Expires November 2001 3 draft-pbh-packet-optical-escalation-01.txt May 2001 traffic over alternate paths that have not been affected by the failure. The restoration path can be either pre-established or established on request by the packet layer. The PONC can in turn perform a longer term overall network optimization. 3. The packet and optical layer coordinate the overall restoration: This can happen in various scenarios. For example, the optical is responsible for restoration but is not able to find sufficient optical bandwidth within the optical network or in a similar way, the packet layer is responsible for restoration but can not fit the restored traffic onto alternate unaffected Layer-3 paths. In both of these cases, the optical and packet layers should cooperate to achieve the right recovery. In this case, it is important to make sure that restoration at both levels does interact positively. For example, hold-off timers may be used to inform one of the layers to wait until the results of the action performed by the other layer is known. However, by combining inter-layer synchronization a better restoration time should be achieved. 3.2 Packet/Optical Layer Interaction and Escalation Strategy This section focuses on the inter-working aspects of packet and optical restoration aspects. The escalation strategy is defined by the set of detection functions describing: - the originating failure - the protection functions applied within the recovery process - and the interaction between the upper and/or lower layers protection functions applied during the recovery process. The escalation strategies are governed either by arbitrarily setting failure detection and recovery completion time or by explicit message exchange between the layers. Two main escalation strategies (also referred to as inter-working strategy) are currently defined and briefly summarized here: - Bottom-up strategy starts at closest layer to the failure (generally the bottom layer) and escalates toward the upper layer upon expiration of the recovery timer (hold-off timer). This timer is defined in such a way that is allocates ôenough timeö for the lower layer(s) to detect the failure, execute the recovery process and recovery completion time before triggering recovery process defined at the higher layer(s). - Top-down strategy starts at the upper(most) layer and escalates downward to the lower layer(s); depending on the working level, the top-down strategy is suitable when each of the layers define Class- of-Services (CoS) so that the escalation strategy might take this CoS into account when executing the recovery process. Papadimitriou D. (Editor) - Expires November 2001 4 draft-pbh-packet-optical-escalation-01.txt May 2001 A combination of these strategies known as the Hybrid Strategy refers to an approach where both upper (in this case the packet/MPLS) layer and the lower layer collaborate in order to achieve the fastest restoration. 3.3 Bottom-Up Strategy In packet LSP over lambda layer (i.e. optical channel layer), the following escalation strategy can be defined, where the physical layer (optical transmission) triggers the OCh layer protection mechanism. Subsequently, if this mechanism does not restore the connection within a given maximum restoration time the upper layer (LSP restoration mechanism is performed). For that purpose, the thresholds (hold-off timers) at which the upper-layer protection needs to be ôexecutedö must be initially configured and subsequently dynamically re-evaluated each time the network experience a failure. A mechanism like the ones defined in [IPO-RFR] and [IPO-RMR] can be used for that purpose. The Bottom-up strategy is generally considered as having the shortest recovery time (with respect to the Top-Down escalation strategy), since recovery takes place as close as possible to the failure. Another advantage to this strategy is that the coarser switching granularity of the lower layers can simplify recovery, since fewer individual flows may need to be re-routed. A drawback, however is the coordination required between layers, which are capable of initiating recovery. Therefore, the following escalation strategy can be defined (here we assume FSC capable interfaces, so that GMPLS signalling is also considered at the lowest fiber-port layer): Packet Layer <==> PSC interface <==> GMPLS Signalling ^ ^ ^ | + | | + | | (1) + Message | (1) | + | | + | Lambda Layer <==> LSC interface <==> GMPLS Signalling ^ ^ ^ | + | | + | | (0) + Message | (0) | + | | + | Fiber Layer <==> FSC Interface <==> GMPLS Signalling Compared to the ôclassicalö packet LSP layer over SDH/Sonet (i.e. TDM-LSP) as described in the GMPLS architecture [GMPLS-ARCH]: Papadimitriou D. (Editor) - Expires November 2001 5 draft-pbh-packet-optical-escalation-01.txt May 2001 Packet Layer <==> PSC interface <==> GMPLS Signalling ^ ^ ^ | + | | (2) + Message | (2) | + | TDM Layer <==> TDM interface <==> GMPLS Signalling ^ ^ ^ | + | | (1) + Message | (1) | + | Lambda Layer <==> LSC interface <==> GMPLS Signalling ^ ^ ^ | + | | (0) + Message | (0) | + | Fiber Layer <==> FSC Interface <==> GMPLS Signalling This is basically the classical approach applied today in most of transmission networks, where the TDM Layer has in most cases an embedded protection mechanism which by definition decouples the circuit-level from the packet-level restoration mechanisms. At the packet layer, IP re-routing is the mechanism used to switch from the primary to the secondary path. The major drawback of this approach comes from the slow restoration time. In this case, the Network Management System (NMS) of the TDM Layer is not efficient and robust enough to provide fast restoration mechanism for a large number of TDM-LSP to be restored within a small period of time (i.e. for instance, within a second). This occurs when any backup LSP is provisioned and when any pre- computation of the backup LSP routes is performed prior to the failure occurrence. This approach is based on the common coordination achieved by resorting the escalation strategies. At each level, the resiliency scheme is activated sequentially starting from the lower to the upper layer. This implies that at each level the detection of a failure condition (and restoration time) triggers either a protection/restoration mechanism and if the protection/restoration completion time is reached then triggers an upper-layer protection/restoration mechanism. 3.4 Top-Down Strategy As described in [RES-MLN], in a top-down escalation strategy, recovery is always initiated at the higher layer (for instance, the packet or MPLS layer) and escalates downward. For instance, a fast protection/restoration scheme starts at the higher layer and is combined with a slowest protection/restoration scheme at the lower layer. Papadimitriou D. (Editor) - Expires November 2001 6 draft-pbh-packet-optical-escalation-01.txt May 2001 If a given layer (for instance, the packet layer) cannot find the resources that are needed for recovery, it can still escalate a recovery request to a lower layer (for instance, the optical channel or lambda layer) asking for the establishment of a new optical channels (i.e. lambda-switched LSP) to carry the packet traffic that could not be recovered at the packet layer. The rationale for a top-down approach is that it allows a fine differentiation of the availability levels of individual packet traffic streams. This allows different recovery actions to take place for service classes with different reliability requirements. A fiber cut may affect many flows from different service classes, which should not all be recovered in the same manner (or recovered at all). From a capacity efficiency point-of-view, top-down escalation is much better than bottom-up escalation if one wants a finer control over the availability levels of individual packet traffic streams. Top-down escalation does not require the partitioning of optical layer resources into separate sinks for each of the packet traffic availability classes. A disadvantage is that since the lower layer paths supporting higher layers may be physically very long, the recovery process may involve network elements at great distances from the original failure. Top-down escalation strategy starts at the upper-(most) layer and escalates downward to the lower layer(s); depending on the working level, the top-down strategy is suitable when each of the layers define its own Class-of-Services (CoS) so that the escalation strategy might take this CoS into account when executing the recovery process. In this case, the message exchange starts in the optical network, which passes the failure notification until reaching one of the points designated as the boundary of the restoration domain. Packet Layer <==> PSC interface <==> GMPLS Signalling | ^ | | + | | + | | (1) + Message | (1) | + | v + v Lambda Layer <==> LSC interface <==> GMPLS Signalling | ^ | . + . | (2) + | (2) . + Message . | + | v + v Fiber Layer <==> FSC Interface <==> GMPLS Signalling Papadimitriou D. (Editor) - Expires November 2001 7 draft-pbh-packet-optical-escalation-01.txt May 2001 3.5 Hybrid Strategy The hybrid strategy is a combination of a top-down and bottom-up approach. By using this approach, the layer n+1 (for instance, the packet layer) and the layer n (for instance, the lambda layer) cooperates by initiating after synchronization their own restoration procedures. Notice that the layer n-1 (for instance, the fiber layer) does not simultaneously participate to the common restoration process. If for instance, we focus on the hybrid process between the packet and the lambda layer then this strategy can be represented as follows: Packet Layer <==> PSC interface <==> GMPLS Signalling | ^ ^ v + | + | (1) + Message | (1) + | ^ + | | + v Lambda Layer <==> LSC interface <==> GMPLS Signalling | ^ | . + . | (2) + Message | (2) . + . | + | v + v Fiber Layer <==> FSC Interface <==> GMPLS Signalling The lower layer that triggers the ôerror conditionö sends a message to the upper layers (in this case the packet layer). This error condition message flows from the fiber to the packet layer. Then, the receipt of the message by the upper layer û it could be a summary or an aggregation of lower layer messages û triggers the synchronization of the network resources (i.e. the links, the LSPs, etc.) that each layer will restore. So that both packet and lambda layer initiates their own restoration procedure but only after being synchronized at the initial time T0. After a certain period of time determined by a restoration timer, at T(RTR), each layer re-synchronize its actions by exchanging the status of the LSPs they planned to restore at the initial time T0. Then if all the LSPs have been restored the process stops otherwise they restart the previous process by eliminating the already restored LSPs. The process is completed when any network resource identified during the initial synchronization exchange has been restored. Papadimitriou D. (Editor) - Expires November 2001 8 draft-pbh-packet-optical-escalation-01.txt May 2001 This procedure can be consequently described through the following steps: .====================. .=========== . Layer n+1 ^ | |\ /| ^ | | . . \ / . | . | | | \ / | | | | . . /.\ . | . | | | / | \ | | | v . ./ . \. V . |====================| | |=========== | Layer n . . . . . | | | | | . . . . . Time --|--------------------|---|---|------------|--------------> Start T0 < T(RTR)> T(RTR)+s The restoration period is determined by k x T(RTR) + s, as illustrated in the above figure. In the hybrid model, the timing plays a very important role. The idea is to define a restoration hold-off timer per layer but also a restoration quantum time (i.e. a restoration interval or timer RTR) during which synchronization of the restoration between layers can occur. 4. Conclusion: Applicability of the ôEscalation Strategiesö In this document, we describe two well-known escalation strategies referred to as bottom-up and top-down strategies. When considering packet-overûoptical networks, the bottom-up strategy is the most ôusedö since coming directly from ôdigitalö transmission TDM networks. However, with the current profound and fundamental change in transmission technologies where all-optical channels replaces SDH/Sonet circuits (and SDH/Sonet is only use as a framing procedure), switching capabilities are now concentrated on two layers: the packet and the optical layer. From these considerations, and the generalization of the mechanisms for IP-based control planes, well-known concepts in the packet layer will be applied to optical layer as well, such as Class-of-Services. Moreover, since now the ôcircuitsö are to be considered as huge pipes, the multiplexing capabilities are concentrated at the packet layer (i.e. using LSP multiplexing). Depending on the LSP multiplexing rate and the available discrete bit-rate at the optical layer, a top-down approach can be considered when for instance the optical layer does not provide restoration mechanisms or signal quality monitoring capabilities so that the re-routing must be performed at the upper-layer. Papadimitriou D. (Editor) - Expires November 2001 9 draft-pbh-packet-optical-escalation-01.txt May 2001 Nevertheless, it doesnÆt seem to be the most efficient way to perform restoration within the network since for instance, a huge number of packet LSPs can be multiplexed into a 40Gbps lambda-LSP. Consequently, if we consider (and when the optical layer enables it) simultaneous restoration at optical and packet layer then, advantages of both layers can be combined in order to achieve a faster restoration comparing to one achieved when using a top-down approach. This hybrid process can be considered as the best method when for instance, some of the packet-LSP are ôin sizeö comparable to the bit-rate of the optical channel into which they are multiplexed (for instance, 1:N multiplexing with N < 8) while the majority of the packet-LSP are ôin sizeö very small compared to the bit-rate of the optical channel into which they are multiplexed (for instance, 1:N multiplexing with N > 128) but using a comparable number of lambda- LSP. In the first case a ôtop-downö approach can be considered while in the second case a ôbottom-upö approach can be efficiently considered. Therefore, when a physical layer failure occurs implying to consider simultaneously both scenarios then an hybrid packet- optical restoration mechanism is the most suitable. 5. Further Studies This analysis results in potential control protocols enhancements required in order to achieve the fast-restoration targets. Control Protocol enhancements need to be refined which can support the following requirements: - They need to accommodate each model (bottom-up, top-down, hybrid) - They need to function which different number and types of layers in the network - They must support a wide range of service classes with different reliability requirements 6. References [GMPLS-ARCH] E.Mannie et al., ôGeneralized MPLS Architectureö, Internet Draft, Work in progress, draft-ietf-many-gmpls- architecture-00.txt, February 2001. [GMPLS-CRLDP] P.Ashwood et al., ôGeneralized MPLS Signaling û CR-LDP Extensionsö, Internet Draft, Work in progress, draft-ietf-mpls- generalized-cr-ldp-03.txt, May 2001. [GMPLS-RSVP] P.Ashwood et al., ôGeneralized MPLS Signaling û RSVP-TE Extensionsö, Internet Draft, Work in progress, draft-ietf-mpls- generalized-rsvp-te-03.txt, May 2001. [GMPLS-SIG] P.Ashwood et al., ôGeneralized MPLS û Signaling Functional Descriptionö, Internet Draft, Work in progress, draft- ietf-mpls-generalized-signaling-04.txt, May 2001. Papadimitriou D. (Editor) - Expires November 2001 10 draft-pbh-packet-optical-escalation-01.txt May 2001 [RES-MLN] P. Demeester, M. Gryseels, A. Autenrieth, C. Brianza, L. Castagna, G. Signorelli, R. Clemente, M. Ravera, A. Jajszczyk, D. Janukowicz, K. Van Doorselaere, Y. Harada, ôResilience in multilayer networksö, IEEE Communications Magazine, Vol. 37, Nr. 8, August 1999, pp. 70-76. [IPO-REST] J. Hahm et al, ôRestoration Mechanisms and Signalling in Optical Networks,ö Internet Draft, Work in progress, draft-many- optical-restoration-00.txt, February 2001. [IPO-RFR] N. Ghani et al, ôArchitectural Framework for Automatic Protection Provisioning in Dynamic Optical Rings,ö Internet Draft, Work in progress, draft-ghani-optical-rings-01.txt, February 2001. [IPO-RMR] D. Papadimitriou et al, ôOptical Rings and Hybrid Mesh- Rings Topologies,ö Internet Draft, Work in progress, draft- papadimitriou-optical-rings.txt, February 2001. 7. Acknowledgments The authors would like to be thank Bernard Sales, Emmanuel Desmet, Fabrice Poppe and Mike Sexton for their constructive comments and inputs. 8. Author's Addresses Papadimitriou Dimitri (Editor) Alcatel IPO-NSG Francis Wellesplein, 1 B-2018 Antwerpen, Belgium Phone: +32 3 240-8491 Email: dimitri.papadimitriou@alcatel.be Jim Jones Alcatel TND-USA 3400 W. Plano Parkway, Plano, TX 75075, USA Phone: 1 972 519-27-44 Email: Jim.D.Jones1@usa.alcatel.com Debashis Basak Accelight Networks 70 Abele Road, Bldg.1200 Bridgeville, PA 15017, USA Phone: +1 412 220-2102(ext115) Email: dbasak@accelight.com Riad Hartani Caspian Networks 170 Baytech Drive, San Jose, CA 95134, USA Papadimitriou D. (Editor) - Expires November 2001 11 draft-pbh-packet-optical-escalation-01.txt May 2001 Phone: +1 408 382-5216 Email: riad@caspiannetworks.com Papadimitriou D. (Editor) - Expires November 2001 12 draft-pbh-packet-optical-escalation-01.txt May 2001 Full Copyright Statement "Copyright (C) The Internet Society (date). All Rights Reserved. This document and translations of it may be copied and furnished to others, and derivative works that comment on or otherwise explain it or assist in its implementation may be prepared, copied, published and distributed, in whole or in part, without restriction of any kind, provided that the above copyright notice and this paragraph are included on all such copies and derivative works. However, this document itself may not be modified in any way, such as by removing the copyright notice or references to the Internet Society or other Internet organizations, except as needed for the purpose of developing Internet standards in which case the procedures for copyrights defined in the Internet Standards process must be followed, or as required to translate it into languages other than English. The limited permissions granted above are perpetual and will not be revoked by the Internet Society or its successors or assigns. This document and the information contained herein is provided on an "AS IS" basis and THE INTERNET SOCIETY AND THE INTERNET ENGINEERING TASK FORCE DISCLAIMS 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." Papadimitriou D. (Editor) - Expires November 2001 13