Seminar by Srishti Srivastava
Research Approaches Towards a Technology for Performance Optimization via Reliable and Cost-Efficient Autonomic Computing Systems
Srishti Srivastava
Mississippi State University.
Date: Tuesday, December 13th, 2011
Time: 4PM
Venue: CS102.Abstract:
The rapid development of computing technology has increased the complexity of computational systems and the ability to solve large, more complex high performance computing (HPC) applications. Today's HPC systems have rapidly evolved in size (from multi-core to many-core and to petascale and exascale), increased in complexity of the interconnection networks and of the processor hierarchies, have become more expensive in terms of energy consumption and performance per watt, and have evolved from clusters to grids to data-centers to the most recent cloud computing systems. Nowadays, time to solution consists of more factors than just the execution time of the application and requires incorporation of factors such as, scalability, efficiency, and reliability. This increase in the complexity of computing systems, which is expected to move beyond the capability of the system administrators and managers, led to the proposed autonomic computing (AC) approach.
The primary goal of the research at the National Science Foundation Center for Autonomic Computing (NSFCAC) at Mississippi State University (MSU) is to accept this grand challenge of developing and deploying such AC systems. A number of approaches, such as, machine learning and model based computing are being used to enable system integration and automation of management for independent operation, minimization of cost and risk, and accommodation of complexity and uncertainty of systems with large numbers of components. Automation of dynamic load balancing of scientific application execution on HPC systems and autonomic performance management of resource allocation in cloud computing systems are the main focus areas of this research at NSFCAC at MSU. The overall contribution of our research is to develop generic autonomic management systems such that they are applicable to a wide range of computing systems such as traditional data centers or the recent cloud computing systems, and also applicable to a wide range of applications (data intensive or compute intensive). The automation of performance management also addresses the cost, efficiency, availability, and reliability of the computing systems in a holistic manner using a utility measure.
About the speaker:
Srishti Srivastava is a PhD student at the department of computer science and engineering at Mississippi State University since August 2010. Srishti Srivastava received the Bachelors degree in Technology (Computer Science and Engineering) from Uttar Pradesh Technical University, India, in 2007 and the M.S. degree in Computer Science from Mississippi State University, USA in 2010. Her research interests include dynamic load balancing, high performance computing, performance and reliability analysis, optimization, and prediction, and autonomic computing.