Abstract—Query Optimization is at the core for contribution towards performance improvements in application systems. A lot of ideas have been proposed towards Query Optimization and there is lot of On-going research happening in this area. Virtually every commercial query optimizer chooses the best plan for a query using a cost model which is based on cardinality estimation. If cardinality estimation is inaccurate, then this may result in optimizer to choose a sub-optimal plan. But once the optimizer chooses an optimal plan for execution based on the approach of POP, the need for generating an optimal plan for subsequent execution of the same query at a later point in time can be minimized/reduced/exempted by storing the execution plan. This paper proposes a Model for building Dynamic Indexes & Storage and Re-Use of Optimal Query plans generated thru Progressive Optimization (POP) for performance gains. This approach is an extension to the work implemented in “Robust Query Processing through Progressive Optimization”. This paper proposes a model to build Learning system within the database to analyze the stream of incoming queries and project viable indexes as against the initial indexes created by the Administrator and also store and re-use of Optimal Query Plans generated thru Progressive Query Optimization (POP). Index Terms—QoS, PoP. I. INTRODUCTION Every enterprise relies on good decision-making for its sustainable and predictable growth in the market. Good business decisions are based on the analysis of huge amount of data. Therefore every enterprise (big or small) strives to ensure data integrity, availability of data and also has the need to cater to high degree of concurrent access to data. As organizations continue to evolve by expanding their operations by way of venturing into new business initiatives, small company acquisition that complement to their business capabilities, these result in increase of data volumes, leading for the need of high performance and scalable systems. A. Motivations In any automated business process, Performance of software systems is at the core of Customer satisfaction. It is quite obvious that any application system that meets all Manuscript received May 17, 2012; revised May 30, 2012. Sreekumar Vobugari and D. V. L. N. Somayajulu are with Department of Computer Science and Engineering at National Institute of Technology, Warangal, India (email: Sreekumar_vobugari@nitw.ac.in; somadvlns@gmail.com) B. M. Subraya is with Education & Research Unit of Infosys Limited, Mysore, India (email:Subraya@gmail.com) business requirements but fails to meet required Nonfunctional aspects (QoS parameters) will indeed lead to greater customer dissatisfaction. For non-real time application systems, though the business users expect for the best performance (one of the Non Functional Requirement) which is a nice to have feature, for real time machine critical application systems, Performance of the software system in terms of response time or through put becomes the qualifying criteria for acceptance of the application system by business users for Production roll-out. Adding to these challenges is the lack of standardized approaches for capturing and measurement of Performance requirements in application systems. Hence, given the criticality and limitations, we believe that looking for innovative approaches towards performance gains in the form of totally new ideas or improvements proposed to the existing ideas[1,2,3,4] that can bring in substantial improvements will surely be a differentiating factor to mitigate the “Performance” issue in software systems addressing the needs of larger audiences of IT industry. B. Performance Engineering Fundamentals For any mid-sized or larger applications, the Performance of the application systems is dependent on the quality of Architecture definition & implementation aspects. The objective of having more emphasis on Performance aspects is to avoid cost overruns by taking proactive measures than reactive approach. From an Application system perspective, Fig. 1 depicts a scenario of a reactive approach. Fig. 1. Performance influence in software life cycle stages From the above figure, two points are very clear and they are: 9 Absence of proactive approach adds to Costs. 9 Later the reaction, higher the costs. Clearly, we believe that there is a need for a Performance Management Process to overcome the crisis originating out of performance issues as depicted above. Based on our A Model for Building Dynamic Indexes & Storage and Re-use of Optimal Query Plans Generated thru Progressive Optimization (POP) Sreekumar Vobugari, D. V. L. N. Somayajulu, and B. M. Subraya International Journal of Machine Learning and Computing, Vol. 2, No. 4, August 2012 471