AUTONOMOUS NETWORK MANAGEMENT

Autonomous Network Management

Autonomous Network Management

Introduction

Telecommunications Network Management systems are of a class of systems which can be described as large and complex, with many interacting subsystems, considerable realtime requirements, and, extensive user interface and business interface requirements. This research focuses on network management systems as being large complex software systems and, therefore, argues that there is a benefit in using the techniques of design and analysis developed for large, complex systems. The theory, methods and techniques for large systems commonly grouped under the umbrella of software architecture. In particular, this research focuses on the practicality of the design of network management systems, the work that had been performed by others, to date, in this area, and, demonstrated some shortcomings in the software architecture techniques available. In recent years, autonomic, policy-based network management systems seen as the best approach to the management of networks. This paper is the first in a series of analyses and recommendations based on current proposals for autonomic network management systems. This paper analyses four currently proposed autonomic management designs, with regard to the validity of their designs, and, with regard to practicality, as expressed by scalability (focussed on time performance), reliability and maintainability.

Autonomic network management

The autonomic network management represents a crucial task in the development of Future Internet (FI) due to the ubiquity and large scalability of services and systems. In the present paper, we proposed autonomic network management architecture to provide self-management capabilities to any system, as well as the evaluation of a proof-of-concept implementation. This architecture is supported by new approaches to SOA, using an ESB to connect the services, and a CEP to detect desired situations. The evaluation results are promising on the basis for robust autonomic management. Autonomic networking has emerged as one of the most promising approaches towards the instantiation of the self managing future networked systems. The term of being autonomic seems to be very capacious, but it does not mean it makes the system either cognitive or autonomous. The primary role of autonomic is the support for auto-configuration without any explicit need for an external intervention. An autonomic network should behave like a living organism, so the centralised entity such as the base station only monitors the system status, but it refrains itself from any substantial interventions unless these are most needed. To map such a concept onto networked systems one needs to apply network engineering mechanisms which are currently undergoing pre-standardisation (AI-Shaer, 2009, p. 37).

Particularly, the idea of control loops is applicable to such systems, where a decision element (DE) is controlling a managed entity (ME) based on a closed information flow and with the use of external monitoring and policies related data. Here, the decision element would steer the cooperation protocol, which in that precise case would translate to the cross-layer approach outlined above. One should note that not only distinct nodes should express autonomic behaviours, but the network should be autonomic as a ...