Performance Tuning Of Transaction Intensive Real-time Reporting Systems

As software becoming more and more complex and the load of system increases, software performance has become one of the most important qualities of software. Performance tuning is the technology to guarantee the quality of software; and after several years' development, it finally becomes software performance engineering.Transaction Intensive Real-time Reporting System is a kind of systems with the same features, which has both features of transaction intensive system and reporting system. The traditional performance tuning work is made only on transaction intensive system, or only on reporting system. For the systems with both of the features, there is rarely any performance tuning work done in the area.For transaction intensive real-time system, this paper proposes the performance tuning strategy by comprehensive utilization of parallel computing and cache technology. Parallel computing can use the advantage of multi-processor and distributed computing environment, and cache technology can reduce duplicate computing and the waste of resources. Based on these two technologies, this paper presents the synchronous and asynchronous computing framework of dividing a large problem into several small problems computed parallel. With this framework, the complexities of reporting procedures can be transferred to the processing of transactions. So the resource utilization is optimized and the system performance will be improved.