An approach to design and develop generic integrated architecture for autonomic software system

摘要

The continuous increase in the system's software management and run-time exception problems in system's software leads to the development of self-managed autonomic software systems. The idea of developing autonomic software system was proposed by IBM in 2001. The approach involves prevention of security threats; system or any of the system's software failures, and will promise high performance software system. Autonomic computing approach is inspired by the autonomic human nervous system. The human nervous system is self-capable and sends instant control messages to the brain to control body temperature, to take unconscious decision, heal body wound, prevent body from danger, etc. The human body nervous system takes decision itself without the consciousness of a human. Similar approach was tried to develop in software system by IBM which followed by different IT industries such as Microsoft, HP, Oracle etc. The autonomic computing approach makes the decision-making process more reliable and responsive. With the development in the field of autonomic computing, run-time exception handling task, security threats, system failures related issues have been resolved at some extent but it compromises software systems' performance. The reason of low system's performance is the implementation of the autonomic features for different software systems' functionalities which makes the software system more complex. To manage system's software complexity, there is a need to define high-level administrative policies for system's software management. Several autonomic software systems' architecture ideas have been proposed by the researchers so far for system's software management. However, the existing architectures are designed and developed in such a way that it will retain the system's complexity due to which software systems' performance degrade. Also, the existing architectures are not designed with complete autonomic features. Compared to existing architectures, the authors in the present paper proposed architecture to reduce the software management complexity and designed a self-managed policy system. This will in result improve software systems' performance. The exceptions are categorized in two: autonomic and non-autonomic exceptions. To improve the performance of the proposed architecture, mapping of exceptions will be performed in terms of fully-matched, partially matched and not matched. The proposed work on self-management properties has been implemented using case study which gives promising results and confidence for the same in real-time complex applications also.