Internet Draft D. G. Kim Document: draft-kim-ccamp-gmpls-nsid-01.txt S. W. Ryu Expiration Date: April 2003 KT (Korea Telecom) Korea University J. K. Choi ICU C. H. Kang Korea University October 2002 A Carrier Requirement of the Network State Information Database for Traffic Engineering over GMPLS Status of this Memo This document is an Internet-Draft and is in full conformance with all provisions of Section 10 of RFC-2026 [1]. 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 obsolete 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 This document presents a set of requirements of the Network State Information Database (NSID) for Traffic Engineering over Generalized Multiprotocol Label Switching (GMPLS) in the view of service providers. The Network State Information Database is required to implement the network architecture for network models that introduce the control element of IP and to optimize the utilization of network resources. And this document includes discussion about the considerations and necessity of the several attributes to construct NSID for Traffic Engineering over GMPLS that are extended from the requirement for Traffic Engineering over MPLS [4]. These attributes can be used to maximize the utilization of network resources and to enhance resource oriented Traffic Engineering techniques. Kim et al Expires - April 2003 [Page 1] draft-kim-ccamp-gmpls-nsid-01.txt October 2002 Conventions 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]. Table of Contents 1. Introduction.....................................................3 1.1. Terminology.................................................3 1.2. The Network State Information Database......................3 1.3. Document organization.......................................4 2. Traffic Engineering over GMPLS...................................4 2.1. Traffic Engineering Performance Objectives in GMPLS.........4 3. GMPLS Architecture for Traffic Engineering.......................6 3.1. Network State Monitoring and Analysis Stage.................7 3.2. Required Resource Estimation Stage..........................7 3.3. Reconfiguration Decision Stage..............................7 3.4. Logical Topology Design and Modification Stage..............8 3.5. Network Topology Migration Stage............................8 4. GMPLS Architecture for Traffic Engineering in Overlay Model......8 5. GMPLS Architecture for Traffic Engineering in Integrated Model..10 6. Network State Information Database for GMPLS....................11 6.1. Resource Attribute.........................................11 6.2. Policy Attribute...........................................12 6.3. Traffic Attribute..........................................13 6.4. Adaptivity Attribute.......................................13 6.5. Priority Attribute.........................................13 6.6. Preemption Attribute.......................................14 6.7. Resilience attribute.......................................14 7. Implementation Considerations...................................17 8. Security Considerations.........................................17 References.........................................................17 Author's Addresses.................................................18 Kim et al Expires - April 2003 [Page 2] draft-kim-ccamp-gmpls-nsid-01.txt October 2002 1. Introduction According to the recent rapidly increasing of IP traffic, Wavelength Division Multiplexing (WDM) technology is rapidly becoming a technology-of-choice to meet the tremendous bandwidth demand of IP traffic. Traffic Engineering (TE) of an IP network is concerned with performance optimization of operational networks and mapping the flow of traffic to present a physical topology of network. A major goal of Internet Traffic Engineering is to facilitate efficient and reliable network operations while simultaneously optimizing network resource utilization and traffic performance [5]. But because the attributes of present IP routing protocols have been poor, the supplementary means for traffic engineering over GMPLS must be prepared. To overcome this problem, two schemes of the researches for traffic engineering in IP networks have been introduced – one scheme is to control explicitly the data path and other scheme is the adaptive load balancing of usable path – the results of these research came to produce the MPLS technology for traffic engineering in IP networks and a GMPLS technology added the control plane of IP for introducing the traffic engineering over MPLS in WDM networks. For introducing the traffic engineering over GMPLS, we can consider two schemes, one is the extended routing protocol for traffic engineering and the other scheme is the allocation of traffic through a network resource database for the efficient utilization of network resources. This document describes the latter scheme that allocates the traffic to the network resources through the construction of the network state information database (NSID). 1.1. Terminology The reader is assumed to be familiar with the GMPLS terminology as defined in [3]. The Network State Information Database is defined as the information of characteristics and its control values related to the optical resources that can be adopted as the control element in IP over optical network. 1.2. The Network State Information Database In this document we define the network state information database (NSID) that support the traffic engineering database (TED) for the function of the traffic engineering and the management information base (MIB) for the function of network maintenance. The main function of the network state information database (NSID) is the control and the management of network state information data. The Network State Information Database can be classified as an essential element to implement the functions of the traffic Kim et al Expires - April 2003 [Page 3] draft-kim-ccamp-gmpls-nsid-01.txt October 2002 engineering in optical network. It is different from the MIB that is configured to control individually each device dependent on the characteristics of a system. The Network State Information Database can be implemented by using the extension of the routing protocols and the signaling protocols. In the present MPLS router architecture, the information distribution module establishes the database of traffic engineering that is used to store the information of topology, the resource utilization, and the metrics etc. Therefore, the concept of network state information database is required to extend and implement the database of traffic engineering in the optical network. In the overlay model, the information of the resource characteristics and the resource controls should be established to exchange the information between the traffic engineering database (TED) of LSR (Label Switched Router) in MPLS and the Network State Information Database of WXC (Wavelength Crossconnect) in optical system. In the integrated model, the traffic engineering database in MPLS and the Network State Information Database in optic network need be integrated for the unified management. 1.3. Document organization The reminder of this document is organized as follows: Section 2 provides the Traffic Engineering over GMPLS and section 3 presents the GMPLS Architecture for Traffic Engineering. Sectin4,5 present the overview of the characteristics and the requirements of GMPLS architecture required for an extended overlay and integrated model. Section 6 describes the necessity and the structural elements of a network state information database (NSID). Finally, Section 7 contains and the implementation of NSID. 2. Traffic Engineering over GMPLS This section describes the Traffic Engineering objectives in GMPLS and when traffic engineering of MPLS is extended to optical network, the correlation of optical resources and the requirements in traffic mapping are presented. 2.1. Traffic Engineering Performance Objectives in GMPLS The key performance objectives associated with traffic engineering in GMPLS can be classified as being either in MPLS [4]: - traffic oriented or - resource oriented Kim et al Expires - April 2003 [Page 4] draft-kim-ccamp-gmpls-nsid-01.txt October 2002 Traffic oriented performance objectives include the aspects that enhance the QoS traffic streams. In the present Internet network providing best effort service, the key traffic oriented performance objectives can include: minimization of packet loss, minimization of delay, maximization of throughput, and enforcement of service level agreement. These elements are used in optical networks as well as in the IP network. Resource oriented performance objectives include the aspects pertaining to the optimization of resource utilization. Efficient management of network resources, such that subsets of network resources do not become over utilized and congested while other subnets along the alternate feasible paths remain underutilized, is the vehicle for the attainment of resource oriented performance objectives. Bandwidth is a crucial resource in contemporary networks, but in the consideration of optical networks the key resources oriented performance objectives can include: the number of wavelengths, the number of Optical Crossconnects (OXC), the number of fibers, and the number of optical transceivers. For the enhancement of the resource oriented performance the congestion control and the load balancing can be chosen. The objective of load balancing is to minimize the maximum congestion or alternatively to minimize the maximum resource utilization, through the efficient resource allocation. Also for the maximization of resource performance the resource management is needed. For the resource management it is necessary to construct the resource control signaling and the network state information database that is able to manage the network resource through the control signaling. Some key extensions brought by GMPLS to MPLS TE are highlighted in the following [3]. 1) In MPLS-TE, links traversed by a Label Switched Path (LSP) can include an intermix of links with heterogeneous label encoding (e.g. links between routers, links between routers and ATM-LSRs, and links between ATM-LSRs). GMPLS extended this by including links where the label is encoded as a timeslot, or wavelength, or fiber. In MPLS-TE, an LSP that carries IP has to start and end on a router. GMPLS extends this by requiring LSP to start and end on several types of Label Switching Routers (LSR) capable of several kinds of labels. Therefore the type of a payload that can be carried in GMPLS by LSP is extended to allow such payloads as SONET/SDH, G.709, 1GbE or 10GbE, etc. The use of Forwarding Adjacencies (FA) that can improve bandwidth utilization is considered. When the bandwidth allocation can be performed only in discrete units, FA allows the number of required labels to be reduced. GMPLS allows for a label to be suggested by an upstream node to reduce the setup latency and to limit the range of labels that is selected by the downstream node. This feature is useful in photonic networks where wavelength conversion may not be available. While traditional TE-based (and even LDP-based) LSPs are unidirectional, GMPLS supports the bi-directional LSPs. This feature will be useful in resource management. Kim et al Expires - April 2003 [Page 5] draft-kim-ccamp-gmpls-nsid-01.txt October 2002 GMPLS supports the termination of LSP on a specific egress port, i.e. the port selection at the destination node. For TDM, Label Switching Capable (LSC) and Fiber Switching Capable (FSC) interfaces in GMPLS, the bandwidth allocation for an LSP can be performed only in indiscrete units. There are much fewer labels on TDM, LSC, and FSC than on PSC or L2SC links, because the former are physical labels instead of logical labels. Therefore the resource oriented traffic engineering is needed. There are basically three fundamental problems of traffic mapping in MPLS: mapping packets onto forwarding equivalent classes (FEC), mapping forwarding equivalent classes onto traffic trunks, and mapping traffic trunks onto the physical network topology through LSP. But in GMPLS including optical resources there are some problems of the hierarchical LSP with how to map traffic trunks onto lambda, and mapping lambda onto LSP, mapping lambda LSP onto waveband LSP, and mapping waveband LSP onto fiber LSP. The hierarchical mapping stages to optical switching are depicted in Figure 1. Packet--> FEC--> Traffic Trunk--> LSP--> Label Stacking | | | | V | ---------> Lambda < ---- | V Waveband | V Fiber [FIGURE 1] The hierarchical mapping stages to optical switching 3. GMPLS Architecture for Traffic Engineering In this section, the basic frameworks for the proposed traffic engineering that can allocate the optical resources on demand, the wavelength assignment, and the network reconfiguration are focused. Main functions of these frameworks consist of the following stages in shown in Figure 2. +---------------------------------------+ + Network State Monitoring & Analysis + +---------------------------------------+ | V +---------------------------------------+ + Required Resource Estimation + +---------------------------------------+ Kim et al Expires - April 2003 [Page 6] draft-kim-ccamp-gmpls-nsid-01.txt October 2002 | V +---------------------------------------+ + Reconfiguration Decision + +---------------------------------------+ | V +---------------------------------------+ +Logical Topology Design & Modification + +---------------------------------------+ | V +---------------------------------------+ + Network Topology Modification + +---------------------------------------+ [FIGURE 2] The Framework for Traffic Engineering 3.1. Network State Monitoring and Analysis Stage This stage is responsible for collecting the traffic statistics from the network elements. Then the statistics are analyzed for the traffic engineering and the network reconfiguration related decision- making. 3.2. Required Resource Estimation Stage This stage estimates the resources, i.e. wavelength, the bandwidth of each wavelength, in the near future based on the measurement, the collection and the analysis of the past and present traffic characteristics. 3.3. Reconfiguration Decision Stage This stage consists of a series of operations that decides when a network level reconfiguration is performed for the resource allocation from the resource estimation stage. This decision element is based on traffic conditions, the number of network resources, and other operational issues, e.g., suppressing the influence of transitional factors and reserving the adequate time for network to converge. Network reconfiguration can be triggered when any element value exceeds the thresholds. Kim et al Expires - April 2003 [Page 7] draft-kim-ccamp-gmpls-nsid-01.txt October 2002 3.4. Logical Topology Design and Modification Stage This stage computes a network topology based on the measurement and predictions about the traffic state and the network resource. This can be considered as optimizing a layered graph (i.e. IP routers connected by lights in the WDM layers) for specific objectives (e.g. minimum hops and maximizing throughput), subject to certain constraints (e.g. nodal degree, interface capacity), for a given load matrix (i.e. traffic load applied to the network), which in general is a NP-complete problem. Since the network reconfiguration can be triggered by periodically changing traffic pattern, a practical approach is to develop methods that focus on cost-effective, fast convergence, and/or minimal impacts on ongoing traffic in the view of global optimality. There are two methods considered. One method is that sub-TLV that can become aware of the optical characteristics is added to the present IP routing protocol and the other method is that signaling protocol of IP network is introduced to the optical network so that network topology is changed. 3.5. Network Topology Migration Stage This stage consists of algorithms to schedule the network migration from old topology to a new topology. Even if WDM layer resources are sufficient to support the allocation of resource and the change of network resource, there are still other issues concerning the migration. For example, as WDM reconfiguration deals with large- capacity channels, changing allocation of resources in this coarse granularity has significant effects on a large number of end user traffic. The procedure of reconfiguration resource for topology migrations is rising to critical issues. In general a migration procedure consists of a sequence of establishing and taking down individual WDM light paths. Traffic flows have to adapt to the changes of the light path during and after each migration stage. The algorithm is needed so that the network configuration gives minimal change on present network state. 4. GMPLS Architecture for Traffic Engineering in Overlay Model In the overlay model traffic engineering over MPLS is extended to GMPLS using the system architecture based IP control. In the GMPLS network, the router is required to add the sub-TLV related optical attributes to traditional IGP routing protocol to perceive and control the state of optical resources. The Network State Information Database (NSID) is required to control and manage the information of network topology and resources in the optical network. The update routing protocol manages the whole state of IP and optical networks and passes the change of state information to Traffic Engineering Database (TED) of MPLS. Kim et al Expires - April 2003 [Page 8] draft-kim-ccamp-gmpls-nsid-01.txt October 2002 [Router] +---------------------------------------------+ | +-----------+ +---------+ | | | IGP Route | |LSP Path | | | | Selection |=> |selection| | | | Module | | +-----+ | +-----------+ | | +-----------+ | | TED | |=> | signaling | | | ^ | +-----+ | | Module | | | | +----^----+ +-----------+ | | | | | | +----------------+-------+ | <==== | IS/IS & OSPF Routing |===============> | +------------------------+ | +--------------^------------------------------+ | [OXC] | +--------------v----------------------------------+ | +--------------------------------+ Control | | | WDM NC & M Module |<=============> | +--------------------------------+ | | Resource ^ ^ ^ | | status Info. | | | Connection Info. | | +--------------v---+ | +-v---------+ | | | Resource control | | | Light Path | Data | | | Module NSID | | | Connection |<======> | +------------------+ | | Module | | | ^ | +------------+ | | | +---v-----------+ ^ | | | | Protection | | | | Fault -------->| & Restoration |<- Fault | | Notification | Module | Detection| | +---------------+ | | | +-------------------------------------------------+ [FIGURE 3] The Architecture of traffic Engineering in Overlay Model According to the architecture of GMPLS in the overlay model as shown in Figure 3, the optical crossconnect (OXC) in GMPLS network consists of four modules as followings: - The WDM Network Control and Management Module informs the state of the optical network to the routing protocol of GMPLS and manages the optical network by using the control signaling. Kim et al Expires - April 2003 [Page 9] draft-kim-ccamp-gmpls-nsid-01.txt October 2002 - The Resource Control Module manages the optical resources and the topology information for traffic engineering. - The Light Path Connection Module establishes and releases the light path, selects the optical fiber. - The Protection and Restoration Module provides the fast restoration through the protection and restoration algorithms. And the Resource Control Module has the function of managing the limited resources attributes and the network topology by using the network state information database that is created by the Routing and Wavelength Assignment (RWA) algorithm, the control signaling, and the function of informing the change of optical resources to the MPLS domain through the WDM Network Control and Management Module. Also the Resource Control Module informs the available optical resources to the Light Path Connection Module so that the table of light path can be maintained. In the optical domain the WDM Network Control and Management Module takes charge of establishing and releasing the light path from the Light Path Connection Module by using the control signaling and informs the values of change related to the resource information to the Resource Control Module. The WDM Network Control and Management Module reports the faults to the Protection and Restoration Module so that the restoration algorithm is active and the Protection and Restoration Module reports the fault information to the Resource Control Module so that new resources are allocated. 5. GMPLS Architecture for Traffic Engineering in Integrated Model The architecture of an integrated GMPLS model has an IP addressed system for traffic engineering and a lambda router is needed to completely support the network control and traffic management of the MPLS and optical network. Also it is necessary to introduce the wavelength-MIB to consist of the information about not only fibers but also wavelength and the wavelength continuity/interchange constraints of each node. And for the wavelength-MIB network control consists of three modules: wavelength routing, wavelength signaling, and wavelength access control. The wavelength routing module needs an update link-state protocol with the suitable optical layer extensions. The wavelength signaling module fulfills the wavelength routing decisions achieved by the wavelength routing module to perform the wavelength assignment, the setup/teardown optical light path, the priority arbitration with preemption, and the adaptive QoS with QoS negotiation. The implementation approach of wavelength routing needs to be in optical layer extended RSVP or CR-LDP. The wavelength access control module manages the physical connection between the MPLS router and lambda router, and maps MPLS label to wavelength. For traffic engineering, each system needs to be equipped with migration scheduling of the network topology, reconfiguration algorithm, and statistics collection and analysis. Also the network state Kim et al Expires - April 2003 [Page 10] draft-kim-ccamp-gmpls-nsid-01.txt October 2002 information database should be constructed to manage the integrated domain of IP and WDM for maintenance of resources change. For traffic engineering in the integrated model, the construction of centralized traffic engineering needs to manage wavelength-MIB in the optical layer and IP-MIB in the IP layer to strictly manage traffic and network control. 6. Network State Information Database for GMPLS The network state information database is needed for the control and management of the MPLS and optical network. The network state information database is constructed and maintained based on the traffic-engineering model and application method described in section 4 and 5. For example, in the integrated model an entire integrated network state information database is collected from each site and is centralized; in the overlay model the MPLS traffic engineering database and optical network state information database are stored, maintained, and managed separately. The network state information database is involved with two aspects: resources and their usage. Conventional representation of network resources can be simply just the topological information. However, traffic engineering requires more information, e.g. total bandwidth, and current usage on each link. Traffic engineering in an optical layer is interested in not only the utilization status of network resources but also the optical characteristics of wavelength connections and optical signal quality. When overlay traffic engineering is attempted, the objectives functions at different layers may even be different. In the case of integrated traffic engineering, traffic control and resource allocation are considered together so that optimization objectives must be coordinated. Although different traffic engineering models require different design and implementation of network state information database, many common attributes are shared in both cases as discussed in this section. The attributes of MPLS traffic engineering are extended to include the attributes of an optical network. Requirements of common attributes and extended attributes are classified and elements of each attribute and application scheme are observed. 6.1. Resource Attribute When the resource attribute for state information of network topology is extended to the optical network, optical resources can be included in this attribute as followings: - the number of fiber, wavebands and wavelengths per link - the number of LSP, traffic trunk per wavelength - wavelength, LSP coloring - bandwidth of link, wavelength, and LSP and their usage - min, max of current usage related to link, wavelength, LSP - the number of optical transceivers in optical system Kim et al Expires - April 2003 [Page 11] draft-kim-ccamp-gmpls-nsid-01.txt October 2002 In the view of traffic engineering, the optimal allocation of physical and logical optical resources is an important element in determining the performance of the network and this information about resources is fundamental to the network state information database. Also resource class attributes can be used to specify the class of resources and depend on the quality of the optical resources – for example, the quality of the light path, end-to-end BER attributes can be an element to determine the quality of optical resources- which become an important requirement for differentiated services and these are applied differently according to policy attributes. 6.2. Policy Attribute The policy attribute provides differentiated services and this attribute uses a resource class attribute. The policy attribute can have the following attributes. - allowable end to end hops - the number of resources offered - protection and restoration mechanism for network survivability - optical signal noise ratio (OSNR), bit error rate (BER) - network provisioning For example, if the ranges of services according to resource grade present Si, Tj, Uk, those ranges can be presented in the expressions below: S = { S1, S2, S3, ... Si} T = { T1, T2, T3, ... Tj} U = { U1, U2, U3, ... Uk} Ranges of services and same grade of services offered are classified according to the expressions listed above and differentiated services can be provided according to the network policy and optical characteristics. For example, through the limitation of end-to-end hops network management can be done. Services included in Si can allow a maximum of 30 hops; Services in Tj can allow a maximum of 20 hops so that the limitation of hops can be operated as an element of network management. The number of available wavelengths per link is applied for network management. For example, the number of wavelengths per link included in Si can be 10; the number of wavelengths per link included in Tj can be 4. According to the range of services, the finite number of wavelengths can be limited and applied to the network management. This scheme that the service range depends on service grade establishes the parameter for traffic engineering and is capable of implementation. As the provisioning that depends on the service range can construct the optical system, the differentiated services can be provided according to the resource grade attribute of the optical network. Kim et al Expires - April 2003 [Page 12] draft-kim-ccamp-gmpls-nsid-01.txt October 2002 6.3. Traffic Attribute The Traffic Attribute is the element that captures the characteristics of the traffic streams. In MPLS the bandwidth of LSP, the maximum throughput allowed, and the minimum data rate guaranteed are included in the classification of attributes [4]. If this definition is extended to the optical network, traffic attributes can be extended to physical and logical values of wavelength, waveband, and fiber. There are important traffic attributes as follows: - end to end BER - the optical characteristics of fiber: optical frequency, optical signal power, optical signal-to-noise ratio - the number of wavelengths per fiber - the number of wavelength channels in a waveband - the bandwidth of a wavelength channel - the maximum or minimum of data rate These traffic attributes include common elements of resource attributes and utilize the characteristics of resources. 6.4. Adaptivity Attribute The states and resources of network change over time. These changes result from the availability of new resources and reactivation of failed resources, and de-allocation of allocated resources. In the view of traffic engineering, an administrative function that can control the resources is required because of this dynamism of network. An adaptivity attribute is a part of the path maintenance parameters and can be presented as re-optimization. According to this adaptivity attribute, the execution of an optimization algorithm depends on the network state in considering stability. This adaptivity attribute is the necessary attribute at the stage of Network Topology Migration described in section 3.5. The adaptive algorithm that has a minimum effect on the network state should support this adaptivity attribute and can provide provisioning at the time of implementation and control the network state. 6.5. Priority Attribute Priority attribute gives priority to the emergency data and real time data and should be considered with the above resource class attribute and policy attribute. For the implementation of the priority attribute optical resources with excellent quality characteristics- for example the resource that its end-to-end BER is lower is allocated or the backup path leased for restoration is allocated- are allocated to high priority. And the routing and signaling protocol that are updated to occupy optical characteristics can be considered to include the function that can allocate and control the priority. Also for the priority services the preemption that depends on priority should be allowed. Kim et al Expires - April 2003 [Page 13] draft-kim-ccamp-gmpls-nsid-01.txt October 2002 6.6. Preemption Attribute The preemption attribute should be considered with the priority attribute and resilience attribute and it is the function that can give itself using resources to high priority data according to network priority. This function is useful for the efficiency of resources and costs and especially for the implementation of various prioritized restoration policies. However low priority traffic should be used for the optical link and path with preemption attribute, and the provisioning function should be able to control the preemption attribute. 6.7. Resilience attribute The resilience attribute for network stability can be defined as a function that when a fault of link or node occurs, a used traffic stream is transferred to a new path through an alternate path and the path generation algorithm. The type of recovery mechanism of the link can be classified as below according to the time of capacity allocation and the decision about backup capacity [6]. 1) Dedicated protection is that the capacity allocation and the decision of backup path that are concluded before a fault occurs. For example, a 1+1 recovery scheme gives a dedicated protection path for a link or wavelength. The advantages of this scheme are the simple implementation and the fast recovery, but its disadvantage is the poor utilization of resources. 2) Preplanned restoration is that the backup capacity is decided before a failure but the allocation of backup capacity is decided after a failure. For example, when high priority traffic and low priority traffic are used together and high priority traffic has failed, low priority traffic is removed. A 1:1 recovery scheme is a link of low priority traffic that is given to high priority traffic. And 1:N recovery scheme is the other example that recovery depends on priority traffic and one backup path is established for the N link through the control signal assuming that a failure in all links does not occur simultaneously. This scheme is achieved by signaling so that recovery time is slower than dedicated recovery and more complex but is excellent in resource utilization 3) Dynamic restoration is that the backup capacity and decision are executed after failure and is most excellent but has the disadvantage of having recovery time is low and the implementation is complex. In the optical network fast recovery is required and therefore a fast path generation algorithm is required for introducing dynamic restoration into the optical network. Also the concept of Protection Link Group (PLG) is required for providing the differentiated services of protection according to the priority of the link or data in optical network. The Protection Link Group provides differentiated recovery time when the priority link Kim et al Expires - April 2003 [Page 14] draft-kim-ccamp-gmpls-nsid-01.txt October 2002 and its backup path are established by using the three recovery mechanisms above. This PLG scheme can provide differentiated services in the view of the resource utilization and the network policy because the 1:N or M:N protection group based on the traffic priority can be created in optical network. Therefore, PLG is defined as the management of link group to provide differentiated recovery schemes by allocating optical information of each light path to the cost of each path and maintain the consistent recovery ratio. For example, we can provide the differentiated service of data traffic classified by three classes, such as mission-critical, best- effort, and low-priority flows in optical network. The priority of these Optical LSPs (O-LSPs) can be allocated as high, medium, and low priority respectively. Through the allocation of these priorities, each light path has the different recovery scheme according to its priority. Namely, the light paths with high priority have the dedicated protection mechanism, the light paths with medium priority have the preplanned restoration mechanism, and the light paths of low priority have the dynamic restoration mechanism. These priority based recovery mechanisms can provide the differentiated recovery speed and the resource utilization according to the characteristics of each recovery mechanism illustrated previously. Each light path has the information of fiber segments that belong to its own light path and this information can be stored network state information database to compose of PLG for the purpose of network management. The concept of PLG in mesh optical network is shown in Figure 4. +---+ 1 | | 2 ----------|wxc|---------- | | | | | +---+ | | | | | +---+ +---+ +---+ | | 3 | | 4 | | |wxc|--------|wxc|--------|wxc| | | | | | | +---+ +---+ +---+ * | | * * | | * Kim et al Expires - April 2003 [Page 15] draft-kim-ccamp-gmpls-nsid-01.txt October 2002 * | +---+ | * * | 5 | | 6 | * * ---------|wxc|--------- * * | | * * +---+ * * * * +---+ * * 7 | | 8 * ************|wxc|************ | | +---+ Primary Lightpath with high priority P1 {1,2} Primary Lightpath with medium priority P2 {3,4} Primary Lightpath with low priority P3 {5,6} Backup Lightpath B {7,8} Protection Link Group {P1,P2,P3,Backup} [FIGURE 4] The concept of Protection Link Group (PLG) in mesh optical network The fiber ID {1,2} includes the primary light path P1 with the high priority and the fiber ID {7,8} includes the backup light path B for the purpose of backup of primary light path P1. The fiber ID {3,4} includes the primary light path P2 with the medium priority and fiber ID {5,6} includes the primary light path P3 with low priority. All these fiber ID is established as one Protection Link Group (PLG) to maintain the consistent recovery ratio in the case of occurring the fault of light path. If the fault of light path is occurred, each recovery mechanism uses this PGL and the new recovery backup path is made by control signaling for the purpose of consistent recovery ratio within the PLG. Then the failed light path and the newly generated light path are updated this PLG to maintain the consistent PLG ratio. The differentiated recovery scheme according to three priority uses these light path lists included PLG. The scheme of Protection Link Group has some advantage. - It can improve the resource utilization of network compare with legacy recovery schemes Kim et al Expires - April 2003 [Page 16] draft-kim-ccamp-gmpls-nsid-01.txt October 2002 - It can provide differentiated recovery speed according to the traffic characteristics of light paths. 7. Implementation Considerations In considering the construction of the GMPLS network based on these attributes in IP and WDM domain, the following requirements of each layer should be considered. 1) IP layer - information about virtual topology (set of light path connected between routers) of IP network - information about link state of IP network 2) WDM layer - information about physical topology (set of fiber connected between WDM devices) - information about the continuity of wavelength allocated to each link, e.g. wavelength converter is equipped or not - information about switching capability and port usage (information of capability of fiber, waveband, and wavelength switching for each port) - information about fiber, the physical link, for example the number of available wavelengths per fiber, directionality, optical SNR of wavelength, BER - information about the light path, for example, the identity (ID) of the component at the destination network, port ID of adding light path, port ID of dropping, directionality, Fiber ID, throughput per wavelength, end-to-end SNR, Protection Link Group (PLG) ID etc. 8. Security Considerations This document does not introduce new security issues beyond those inherent in GMPLS. It is, however, possible that using the suggested network management attributes provisioning can be done as administrative usage. References [1] S. Bradner, "The Internet Standards Process -- Revision 3", RFC 2026, October 1996. [2] Brander, s., " Key words for use in RFCs to Indicate Requirement Levels", BCP 14, RFC 2119, March 1997. Kim et al Expires - April 2003 [Page 17] draft-kim-ccamp-gmpls-nsid-01.txt October 2002 [3] E. Mannie, "Generalized Multiprotocol Label Switching (GMPLS) Architecture", Internet Draft, draft-ietf-ccamp-gmpls- architecture-03.txt (work in progress), August 2002. [4] D. Awduche, J. Malcom, J. Agogbua, M. O'Rell, J. McManus, "Requirement for Traffic Engineering over MPLS", ITEF RFC 2702, September 1999. [5] D. Awduche, A. Chiu, I. Widjaja, X. xiao, "Overview and Principles of Internet Traffic Engineering", IETF RFC 3272 May 2002. [6] Eric Mannie,D. Papadimitriou "Recovery (Protection and Restotation) Terminology for GMPLS", IETF Draft, draft-ietf- ccamp-gmpls-recovery-terminology-00.txt (work in progress), June 2002. Author's Addresses Dae-Gun Kim Korea Telecom, Korea University 1,5-ka, Anam-dong, Sungbuk-ku, Seoul,136-701, Korea Phone : 82-2-927-6116 e-mail : dkim@kt.co.kr Sung Woo Ryu Korea Telecom, Korea University 206 Jungja-dong Bundang-gu, Songnam-city Kyonggi-do ,463-711, Korea Phone : 82-2-929-5625 e-mail : isdn@kt.co.kr Jun Kyun Choi Information and Communication University (ICU) 58-4, Hwaam-dong, Yuseong-gu, Daejeon, 305-732, Korea Phone: 042-866-6122 e-mail : jkchoi@icu.ac.kr Chul-Hee Kang Korea University 1,5-ka, Anam-dong, Sungbuk-ku, Seoul, 136-701, Korea Phone: 82-2-927-6116 e-mail: chkang@widecomm.korea.ac.kr Full Copyright Statement "Copyright (C) The Internet Society 2002. 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 Kim et al Expires - April 2003 [Page 18] draft-kim-ccamp-gmpls-nsid-01.txt October 2002 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 Document: draft-kim-gmpls-nsid-01.txt Expiration Date: April 2003 Kim et al Expires - April 2003 [Page 19]