Processing Capacity and Response Time Enhancement by Using Iterative Learning Approach with an Application to Insurance Policy Server Operation

摘要

In this study, computing system performance enhancement by using iterative learning technique is presented. Computational response time and throughput of the computing system is improved by introducing computational cost model and selection probability for each individual job. Excepted gain by enforcing dynamic caching is maximized in terms of classifying the arriving computing jobs on an selective manner and dynamically replacing them in a limited memory space. Gain maximization is performed by tuning the window size which helps to compare the computing jobs in terms of their individual selection and occurence probabilities. Fairly special computing work in insurance risk investigation is chosen for experimental validation of the proposed approach. Aspect Oriented Programming (AOP) methodology on Java platform is used for the experimental setup. AOP allows to identify computational jobs and their parameters based on the marked annotations. Experimental results show that the developed iterative learning based caching algorithm performs better than the other well known caching techniques. The design and development of iterative learning based dynamic caching minimizes the necessity of developers' decision about job results to be cached in the memory.