Distributed processing environments are more complex than centralized systems and require a new breed of performance management framework and tools. The selection and development of distributed performance management mechanism should consider the tech...
Distributed processing environments are more complex than centralized systems and require a new breed of performance management framework and tools. The selection and development of distributed performance management mechanism should consider the technical problems such as large number of nodes, heterogeneous environments, large volume of data, multiple data sources, changing configurations, and response time measurement.
So, with the growing heterogeneity and complexity of these environment, it becomes increasingly important to relieve system managers from tedious tasks and raise the level of abstraction by adding intelligence to the management resources and applications.
Management policies give structure to the use of technology and promise to add significant value towards the management of today's complex distributed computing environment. And, due to the intelligent agent concepts ability to facilitate the delegation of management functionality to autonomous agents operating behalf of a management application, it has become one of the key technologies envisaged to realize the vision of distributed system management.
In this paper, a framework for applying the concepts of intelligent agents to implement performance management policies is proposed. And, we discuss a method to provide a common set of definitions so that performance information can be communicated among different managing agents.
Future works will focus on the implementation of this framework in terms of Web and Java technologies. Also, the automation, or at least a systematical approach to refine the abstract performance requirements down to applicable operational rules is needed.