Real-time transaction processing for autonomic grid application

Sarangi, Banalata (2007) Real-time transaction processing for autonomic grid application. MTech thesis.



The advances in computing and communication technologies and software have resulted in an explosive growth in computing systems and applications that impact all aspects of our life. Computing systems are expected to be effective and serve useful purpose when they are first introduced and continue to be useful as condition changes. With increase in complexity of systems and applications, their development, configuration, and management challenges are beyond the capabilities of existing tools and methodologies. So the system becomes unmanageable and insecure. So in order to make the systems selfmanageable and secure the concept of Autonomic computing is evolved. Autonomic computing offers a potential solution to these challenging research problems. It is inspired by nature and biological systems (such as the autonomic nervous system) that have evolved to cope with the challenges of scale, complexity, heterogeneity and unpredictability by being decentralized, context aware, adaptive and resilient. This new era of computing is driven by the convergence of biological and digital computing systems and is characterized by being self-defining, self-configuring, self-optimizing, self-protecting, self-healing, context aware and anticipatory. Autonomic computing is a new computing model to self manages computing systems with a minimal human interference. It provides an unprecedented level of self-regulation and hides complexity from Users. The Autonomic computing initiative is inspired by the human body’s autonomic nervous system. The autonomic nervous system monitors the heart- beats, checks blood sugar levels and maintains normal body temperature with out any conscious effort from the human. There is an important distinction between autonomic activity in the human body and autonomic responses in computer systems. Many of the decision made autonomic elements in computer systems make decisions based on tasks, which are chosen to be delegated to the technology. The influences of the autonomic nervous systems may imply that the autonomic computing initiative is concerned only with lowlevel self-managing capability such as reflex reaction. The basic application area of autonomic computing is grid computing. Both autonomic computing and grid computing are proposed as innovations of IT. Autonomic computing aims to present a solution to the rapidly increasing complexity crises in IT industry, as grid computing tries to share and integrate distributed computational resources and data resources. Basic aim is to implement the autonomic computing in grid related study like autonomic task distribution and handling in grids, and autonomic resource allocation. In this thesis paper we presents methods of calculating deadlines of global and local transaction And sub transaction by taking EDF algorithm and measure the performance by taking miss ratio in Different workload. We implement this work in an existing grid. The basic aim is to know autonomic computing better. It is a model to self manage computing Systems with minimal human interference. Self manage has properties like self-configuration, self-optimization, self-healing, self-protection. Autonomic grid computing combines autonomic computing with grid technologies to help companies to reduce the complexity associated with the grid system and hides the complexity from their grid user. Autonomic real-time transaction services incorporate fault tolerance into autonomic grid technology by automatically recovering systems from various failures. Here in this paper Deadlines of global transaction, sub transaction and local transaction are calculated by taking parameters arrival time, execution time, relative deadline, and slack time. We are taking a periodic transaction having λ (transaction arrival rate per second) Tasks are generated at different nodes with Poisson ratio with λ as workload. Miss ratio is the performance metrics. With increase in workload miss ratio first decreased and then rose. The reason was each sub transaction acted as a unit to compete for resources so that more workload the more system resource they consumed. So more transaction missed their deadlines, as they could not get enough resource in time. EDF algorithm has both less global and local miss ratios then other scheduling algorithm. If EDF is compare with FCFS or SJF or HPF it is apparent that both algorithms perform almost identically until no of transaction is low, then EDF misses fewer dead lines than other. Real-time transaction can handled by the grid in autonomic environment and satisfy properties of autonomic computing.

Item Type:Thesis (MTech)
Uncontrolled Keywords:Autonomic grid application, IT, EDF, FCFS, SJF, HPF
Subjects:Engineering and Technology > Computer and Information Science
Divisions: Engineering and Technology > Department of Computer Science
ID Code:4367
Deposited By:Hemanta Biswal
Deposited On:12 Jul 2012 10:34
Last Modified:18 Jul 2012 16:21
Supervisor(s):Sahoo, B D

Repository Staff Only: item control page