A Self-management Approach to Service Optimization and System Integrity through Multi-agent Systems

Abstract

The primary goal of this thesis is to make simple approaches to implement selfmanagement functions based on the regular administration tools as a way to simplify tedious tasks for the system administrators. In order to improve the efficiency of system administration, this work mainly focuses on how to combine ideas from the area of selfmanaging with the more well-established concepts from virtualization technology. A fully functioning self-management system was developed for two essential administration tasks in this project: service optimization and system integrity. In these scenarios, each virtual machine can make its own decision when and where to live migrate itself between the physical nodes in order to reach the predefined policies. This method minimizes the maintaining tasks for the system administrators, only leaving the high-lever administration policies for the users to decide. From a higher point of view, this project can be used as an example for other system administrators, showing how to optimize the regular administration tools for different self-management purposes.