AbstractsComputer Science

Enhancing self management capabilties for software systems;

by Mallikarjun M Math




Institution: Graphic Era University
Department: Computer Science
Year: 2014
Keywords: Computer Science
Record ID: 1191635
Full text PDF: http://shodhganga.inflibnet.ac.in/handle/10603/19770


Abstract

The ever increasing complexity of modern computer networks and systems, combined with the extremely dynamic environments in which they operate, it is beginning to outpace our ability to manage them. Information systems that respond to changing conditions, regulate and adjust themselves are the dream behind Autonomic computing paradigm. This initiative depends on bringing together advanced technologies and self management techniques in a holistic way. Autonomic computing presents with an approach to build systems with a capability to sustain and optimize them in way that provides a stable working environment. Such an environment can be defined in terms of self sustaining features of Autonomic computing namely, self-Configuring, self-Healing, self-Optimizing and self-Protecting.In the first phase, MAPE-K Autonomic computing reference model was implemented that has a Self-healing capability. The structure of Symptom catalog is complex because of its mesh like structure and to reference the symptoms and generate corrective actions many XML paths have to be accessed. This results in delay in initiating the control action. Therefore, in the proposed approach, the structure of symptom catalog is simplified by using a variable length record-field structure and these records are stored in hast-table that has constant access time. It is observed that, the estimated access-time for accessing the symptoms stored in the hash-table was and#8776;2.5 units of time in the worst case. On the other-hand if the XML structure is used for the same, the estimated time turned out to be and#8776; 3.5 units of time. Hence, a reduction of and#8776;30% in access time was achieved. It is also observed that the symptoms can be referenced easily because of its simpler structure and actions can be executed in real-time whenever the symptoms are matched.In the second phase of the research, an improved version of traditional approach wherein, an alternate way of writing Generic Log Adapter has been proposed.%%%References p. 159-169