Carlos Samitier Joaquin Luque Fernando Gonzalo DIMAT S.A. Manuel Mejías Grupo Endesa Spain Universidad de Sevilla (Spain) Spain NETWORK MANAGEMENT ARCHITECTURE FOR POWER UTILITY NETWORKS ABSTRACT Telecommunication Networks in the Power Utilities environment are becoming a strategic asset. They give Power Utilities not only the way of managing the core business but also the possibility of becoming a Telecommunication Service Provider. In this new situation, Network Management is a key component in order to achieve and guarantee the required Quality of Service offered to the final user. The paper explores different Network Management architectures and the interaction between them. Network management products in the market follow two major management tendencies, the Simple Network Management Protocol architecture,, defined by the Internet community, and the ISO/ITU-T architecture, based on the Common Management Information Protocol CMIP. Differences and interworking possibilities of these management architectures are analysed. This new approach makes the most of the advantages of both systems and offers a cost-effective solution to the problem of deploying modern Network Management centres. The architecture of a new Network Management Centre that includes this new idea is presented. Key words: ISO TMN architecture, CMIP/CMIS, Internet Network management 1. INTRODUCTION The competitive environment set by the new deregulation scenario and the open market has fostered new concepts in the way Telecommunication Networks are designed and operated. Concepts such as profitability, effective administration of network resources, service provision to third parties, etc are playing a relevant role in this new environment. The effective implementation of these new concepts will require an optimisation of the investment, a reduction of operational costs and the global optimisation of the network resources. This optimisation has to be carried out in terms of the cost and the Quality of the service provided.
On the other hand, the network concept has evolved from being a collection of individual devices to a set of equipment that cooperates to offer a service optimising the resources used in all cases. This new image of the network requires some functions to supervise and control the characteristics of the service offered. The way to achieve the above mentioned goals is by including management functions that allow the resources of the networks to be administrated and thus, making it possible to offer a committed QoS lightening the work of the network manager. The concept of Network Management, although it is not new, is still in evolution. Existing networks based on legacy technologies do not include this kind of functionality whereas emerging transport and networking technologies, such as SDH or ATM, embed management functionality as one of their core components. An important aspect of the implantation of Network Management Functions is the incremental cost introduced in the network, this increment is related to the network technology. The integration of NM functions on legacy networks usually requires a data management network, some Mediation Devices and the development of proprietary proxy agents. Whereas, in existing and new technologies only the NM centre has to be added as these technologies integrate network management as a part of their functionality. As can be seen, the extra cost of including NM functions in networks based on existing or new technologies is negligible. Whereas the advantages obtained from a better resource management and the improvement of the QoS of the services provided could bring a considerable profit. This paper is going to focus on this second approach as it is of interest for new and existing networks. The election of a certain technology to implement a network has a direct implication in the management architecture. For instance, SDH networks are usually managed according to the OSI model; ATM Networks can be managed in OSI style, for public networks or using Internet in the case of private application; and IP networks are managed using. The real management scenario is still more complicated as modern telecommunication networks are based on the integration of different networking technologies SDH for transport, ATM for switching and IP for service delivery. This situation has brought about the need to integrate different management architectures in order to get an homogeneous network management scheme. The described environment pushes the technology to the limit and compels network designers to find practical solutions that are fully compliant with the operational requirements. This practical approach will be based, depending on every case, on a combination of different components of different management architectures taking those that best suit the actual requirements of the network. This paper describes network management technologies used to manage modern and new coming networking technologies and the way they can be integrated over a common management platform, the NOMOS TMN system. 2. EVOLUTION OF NETWORK MANAGEMENT The management of communication networks has evolved from a scenario of Management by Complaints, where corrective actions were taken after breakdown situations, to the future approach of active management. In this scenario, management functions are modelled as a new service that supervises and controls the QoS of the services provided by the network. As far as the management architecture is concerned, there has also been an evolution from proprietary approaches to standard architectures. In the first case, existing networks based on legacy technologies, such as a mixture of analogue and digital technologies, could only be managed using proprietary management approaches. These solutions use proprietary protocols to manage some equipment whereas some others can only be managed through Mediation Devices. This situation raises the problem of management of heterogeneous networks. To tackle this problem the ISO defined the manager-agent architecture as the base of the Open System Interconnection (OSI) standards. The manager-agent principle has been successfully used by the ISO and the Internet Engineering Task Force (IETF) to develop the OSI management and the Simple Network Management Protocol () standards
Manager Agent Management Applications Managed Resources Manager MIB Agent MIB Management Services Management Protocol Management Services Management Protocol Protocol Messages Figure 1. Native Management Architecture Figure 1 shows this generic management architecture. The goal of this model is to unify the access to a distributed collection of heterogeneous management elements. Management elements are represented as objects. The manager can interact with these objects through the agent. The interaction between the manager and the agents is carried out by means of a management protocol. So that a common management interface between the management centre and the devices under management is achieved regardless of the type of device, the type of object and the way it can be accessed inside the device in which it is contained. It is important to mention at this stage that the management architecture is based on two main requirements. The first is that some communications facilities have to exist between the management centre and the managed devices, allowing management centre and the devices under management to be connected and then transfer management information. Whereas, the second is that managed devices have to be equipped with a management interface that allows management centre to interchange management actions and notification with the local agent. The integration of management information over a common platform that also allows management action to be executed is a very important requirement for the network manager. NOMOS TMN system [1] offers a solution that has been designed taking into account the Power Utility requirements, as it is able to integrate different types of management over a common management platform. Although this is a giant leap in management technology, the evolution of networking technology is introducing new architectures and concepts that will make management systems evolve towards a new scenario of integration of services. The services will be provided over a common networking technology based on standard switching and transport technologies that will require seamless homogeneous management architectures. In the following chapters, we are going to analyse the advantages of both, ISO and IETF management frameworks and the way that both approaches can interact establishing a unified management scheme. 3. THE OSI MANAGEMENT ARCHITECTURE The OSI management architecture was a joint development effort of the ISO and the ITU-T for the standardisation of the management of the OSI protocols and network devices in general [2]. The OSI management architecture follows the basic idea shown in figure 1. Management interactions between the management centre and the managed device are performed in terms of operations and notifications that are communicated by means of management services offered by the management protocol. The agent performs management operations on the managed object as a consequence of the operations requested by the management centre.
A managed object is the view of a resource, whether physical or not, that is subject to management. The OSI framework includes three different aspects: - A set of functional components that define standard management operations. - A protocol that specifies the management communication and offers a set of management services. - And the information models that define the attributes, operations and notifications performed upon the managed objects as well as the structure of the Management Information Base (MIB) that contains them. 3.1 Functional components The set of management functions supported by a TMN is grouped in five functional areas: configuration, fault, performance, security and accounting Configuration Management This functional area includes the necessary functions to control and set the provision of resources and services, the operation of the managed object and the configuration of the managed system. Fault Management Fault management includes functions to detect, isolate and correct abnormal operation of managed systems. The capability of alarm logging, fault traces and diagnostic tests is also included. Performance Management Performance management includes functions to evaluate the behaviour and performance of the resources and the effectiveness of communications. Accounting Management This functional area includes the functions that perform the accounting for the services provided including charges and the identification of costs. Security Management This functional area includes two aspects, the management of security policies and the security of the management functions. 3.2 Communication components Management interactions between the manager and the managed object are carried out by means of the communication services CMIS supported by the management protocol CMIP [3]. CMIP protocol has been designed to work over a reliable transport mechanism. Thus, when management operation has to be carried out an association is established using a connection-oriented transport protocol and the necessary resources are dedicated to this service even when there is no management traffic to be transported. One of the outstanding features of CMIS is that of scoping and filtering. Thanks to this, it is possible to apply a management action over a set of managed objects establishing logical operators expressed as a filter [4]. Table 1 describes services offered by CMIS. It is important to mention the fact that managed objects can be created and deleted so that the management can be configured on the fly, being possible to adapt it to the changing conditions of the services supported by the network. SERVICE M-EVENT-REPORT M-SET M-GET M-CANCEL-GET M-ACTION M-CREATE M-DELETE DESCRIPTION Reports an event from a managed object Requests the modification of management information Requests the retrieval of management information Request to cancel an outstanding get-request Requests that an action be performed by a managed object Requests an instance of a managed object to be created Requests the deletion of an instance of a management object Table 1. CMIS service primitives
3.3 Information Components The resources of the network that are subject of management are modelled as management objects. These objects are defined by means of their attributes, the management operations that can be performed upon them, behaviour, relationship with other objects, and the notifications that they can emit. The set of managed objects, together with their attributes, constitutes the management information base (MIB) [5]. The properties of managed objects are defined by means of object classes. Each managed object is an instance of a class, which includes all managed objects that share the same definition and properties. A managed object class is defined as a collection of attributes, operations, notifications and related behaviour as well as the relation with other object classes. The OSI management architecture has been defined with a powerful object class framework that allows any kind of managed object to be defined. For instance, a service or a virtual network could be defined as an object. In order to understand how these kinds of objects are defined and managed, the main properties of object classes and therefore of the managed objects, are going to be mentioned. One of the most relevant properties of the object classes is the Inheritance. Inheritance allows a hierarchy of object classes to be defined. The subclass inherits attributes, notifications, operations and behaviour from the superclass. It is also possible to define multiple inheritance so that a subclass will acquire attributes, notifications, operations and behaviour from more than one superclass. A managed object of one class can contain other managed objects of the same or different classes. This relationship is called containment. This containment relationship is defined between managed object instances, not classes. The implementation of the properties defined above requires quite a complex management information base. The OSI MIB includes three different trees. The registration tree, the inheritance tree and the naming tree. The registration tree is used to store the definition and identification of the managed object classes. The definition includes class name, attribute types etc. These definitions are used in management protocols to uniquely identify aspects of managed objects and their associated attributes, operations and notifications and also to create new instances of any type of object. The inheritance tree contains the inheritance relation between managed object classes. This tree makes up the inheritance hierarchy of the object classes stored in the registration tree. The naming tree contains the managed objects that have been defined using the object classes stored in the registration tree. The naming tree stores managed objects taking into account the containment relationship amongst them. The architecture we have seen so far offers a very powerful management scheme. It is the best solution for the NM Centre as it allows the management policy and structure to be defined by the user, but its integration on the managed equipment represents a tremendous complexity for small/medium size devices. Management of private networks that are built using small/medium size equipment will require a simpler management scheme. 4 THE IETF MANAGEMENT ARCHITECTURE The Internet management architecture was defined by the Internet community and approved by the IAB as an Internet standard [9] with the aim of covering the management functionality requested to manage the devices that form the Internet. The simplicity of the proposed architecture and the small amount of resources asked of the managed equipment have driven its massive diffusion. Nowadays, any kind of equipment, not only the ones forming the network, can be managed using Internet architecture, this concept has been extended to auxiliary devices such as host, printers, etc. The working principle of the IETF management architecture follows the basic idea shown in figure 1. Network Management Station NMS performs management task over the managed device that is equipped with a management agent that interacts with the managed object stored in the MIB of the device. Although the architectural concept is the same, the facilities of both systems as well as the complexity are very different.
The IETF network management model includes four key elements: management station, management agent, management information base (MIB) and network management protocol. The management station executes management applications which monitor and control managed elements, the Graphical User Interface that monitors the management information, the management protocols and contains the central MIB that contains the consolidated management information stored in the managed devices. Only the management protocol and the MIB are the subject of the standardisation. Management agent and the MIB are the management components of managed devices. The agent responds to requests of management information or management actions also being able to provide unsolicited management information. Management resources are represented as objects. Each object represents one aspect of the managed system. The collections of objects that define a managed device are stored in the MIB. Management operations over the managed object are performed using the service of the Simple Network Management Protocol (). is an application level protocol, it operates over the datagram transport protocol UDP, unlike CMIP, protocol does not require any kind of association or reservation of network resources. Management Centre Managed Equipment Management Application Managed Objects GetRequest GetNextRequest SetRequest GetResponse Trap GetRequest GetNextRequest SetRequest GetResponse Trap Agent UDP IP Agent UDP IP NETWORK Figure 2. messages Due to the fact that managed object classes are fixed and static the services provided by the are limited to the interaction over the managed objects. Managed object classes are defined by the Structure of Management Information (SMI)[10]. SMI uses ASN.1 syntax to specify object classes. Objects in a MIB are identified by a sequence of integers that are used to traverse the MIB tree identifying an object type. Managed objects are instantiated by giving a value to an object type. For instance, the notion of the bitrate of a data port is an object type that should be included in the MIB, the bit-rate of every port of a certain device are the instances of this object type. This is a quite rigid but simple structure. MIBs cannot be dynamically defined at run-time and new managed object classes cannot be included. Although this is a limitation to the management capabilities, the resources required by the managed device to support functionality are almost negligible. IETF management architecture is a simple and efficient scheme to access and control managed objects distributed in a network. It covers and offers a cost-effective solution for the problem of managing distributed devices. Nevertheless, It does not include any kind of standardisation for the management centre or for the management functions required in the NM centre. The fact that the managed object classes are not dynamic makes the customisation of the management platform only possible by using proprietary solutions.
Taking into account the Power Communication Network management requirements, it is crystal clear that the best management approach would be one based on the OSI architecture but able to manage devices by mapping the OSI services on the IETF architecture. 5 INTEGRATION OF OSI AND IETF MANAGEMENT ARCHITECTURES We have seen in previous chapters the advantages of the two main network management schemes. If we consider the functionality asked of network management centres of the Power Utility Networks only OSI management architecture is able to offer the functionality required. Nevertheless, as many of the managed devices can only be managed using architecture, the integration of both technologies has become a must. Although functionally similar, both management schemes differ in terms of their complexity and management services offered. Nevertheless, it is possible to integrate both technologies in such a way that a unified view of the network, despite protocol and SMI differences is presented. The Network Management Forum has been working on the definition of an interworking architecture able to integrate CMIP and management schemes [11]. The model proposed for the integration of and CMIP environments is based on the principle of proxy management. In this architecture, a new type of agent, the proxy agent, is defined. This agent performs all the necessary protocol and data format translation in such a way that the management centre interacts with the managed devices using CMIP management so that the management centre has a CMIP view of the managed devices. This architecture can be used to integrate any kind of management approach over a common CMIP management centre. The proxy agent can be installed in the management centre or in any mediation device. The basic model for ISO CMIP to IETF proxy management is depicted in figure 3. Manager Proxy Agent Management Applications GDMO MIB Internet MIB Managed Resources Service Emulation ISO/CCITT GDMO (scoping) (operations) (filtering) Internet Internet Manager MIB (message transformation) Agent MIB CMIS Services CMIS Services "Services" "Services" CMIP CMIP CMIP Messages Messages Figure 3. Proxy management architecture
The proxy agent has two different sides. One behaves like an ISO agent communicating with the ISO manager by means of CMIP protocol. The other behaves like an Internet manager, communicating with the agent of the managed device using protocol. The proxy agent provides emulation of CMIS services by mapping/generating the corresponding messages required to carry out the requested service. The functionality of the proxy agent is based on the existence of a pair of MIB definitions that have been created by translating MIB objects to GDMO format. This translation is manually carried out before the proxy agent software is installed and cannot be changed at runtime. These MIBs are used to provide run-time translation of the management information. 6 INTEGRATED MANAGEMENT CENTRE The possibility of integration of the two main management architectures open up the way to the advent of a new generation of network management centres. The management of the Power Utility telecommunication Networks requires a reliable technology able to unify over a common platform powerful easy-to-use management functions. This management platform should allow every device involved in the network to be managed regardless of its technology. Due to the fact that modern networks are built using heterogeneous technologies, it will be a required facility, for the new generation of Network Management centres, will be the ability to integrate these technologies over a common platform. As we have seen in previous chapters, OSI CMIP management architecture offers the functionality required by a management centre, on the other hand, most networking devices are managed using the approach. This situation will require the management centre to integrate both management technologies. This new concept of management centre is now possible thanks to the new development that is being carried out on the NOMOS management system which is now near completion. The new NOMOS concept integrates the functionality of a proxy agent in the same platform so that the management centre will be able to manage devices from the OSI world. NETWORK () PROXY AGENT Network Protocols Conversion CMIS/CMIP OSI GDMO Data Base GAB GUI Adaptation USER INTERFACE Specific Processes Distributed Communications ( CMIS ) Figure 4. Network management centre architecture Figure 4 shows NOMOS architecture. It can be seen that the integration of the proxy agent does not modify the other components of the system, so that the existing functionality and the ability of successfully integrate proprietary management on the OSI world will remain available to the user.
7. CONCLUSIONS We have presented the actual networking scenario in which two main management technologies are being used. The integration of both schemes has also been included as it responds to the actual requirements of seamless unified solutions stated by the final users. Network management technology is probably one of the most profitable investments. It plays a capital role in the improvement of the quality of the services offered by the network as well as in the reduction of the maintenance costs since it offers the tools to identify and isolate faulty elements. The main benefits obtained with this new approach will be achieved by: - Reduction of training costs - Automation of management processes that improves the QoS of the service offered to the final customer - Increased flexibility due to the usage of the appropriate management technology in every case - Integrated management of heterogeneous networks - Coherent end-to-end management The work presented in this paper describes the research line that is following the evolution of the fully compliant TMN system NOMOS. Its OSI management kernel will be complemented with a specific proxy agent that will allow managed devices to be integrated in the same platform. This research has been made possible by a grant from the National Research Plan ATYCA. 8. REFERENCES [1] J. Luque, M. Mejías, F. Gonzalo and C. Samitier. Managing Power Utilities Telecommunication Networks. 3 rd National Symposium on Power System Management. CIGRE. Croatia Oct 98. [2] CCITT Recommendation X.700. Management Framework Definition for Open Systems Interconnection (OSI) for CCITT Applications./ ISO/IEC 7498-4:1989, Information processing systems Open Systems Interconnection Basic Reference Model Part 4: Management framework. [3] CCITT Recommendation X.711 (1991), Common management information protocol specification for CCITT applications. / ISO/IEC 9596-1:1991, Information technology Open Systems Interconnection Common management information protocol Part 1: Specification. [4] CCITT Recommendation X.710 (1991), Common management information service definition for CCITT applications. ISO/IEC 9595:1991, Information technology Open Systems Interconnection Common management information service definition. [5] CCITT Recommendation X.721 (1992) ISO/IEC 10165-2:1992, Information technology Open Systems Interconnection Structure of management information: Definition of management information. [6] CCITT Recommendation X.720 (1992) ISO/IEC 10165-1:1992, Information technology Open Systems Interconnection Structure of management information: Management information model. [7] CCITT Recommendation X.722 (1992) ISO/IEC 10165-4:1992, Information technology Open Systems Interconnection Structure of management information: Guidelines for the definition of managed objects. [8] CCITT Recommendation X.730 (1992) ISO/IEC 10164-1:1992, Information technology Open Systems Interconnection Systems Management: Object management function. [9] IAB. Internet-standard Network Management Framework. RFCs 1155 (STD 16), 1157 (STD 15), and 1212 (STD 16). [10] M. Rose, K. McCloghrie and Hughes. RFC 1155. Structure and Identification of Management Information for TCP/IP-based Internets. May 1990. [11] OMNIPoint Component Set. CMIP/ INTERWORKING. NMF CS341