The Research And Implementation Of CRP Query Optimization For Economic Dynamic System

The multiprocessor architecture have the ability to support high performance requirements of large databases and all of that relies on its excellent scalability.In a multi-processor system,an important factor affecting the efficiency of its query is the query scheduling.Although scholars at home and abroad have endlessly researched on query scheduling,most of the researches have focused on how to effectively process a single query.The study on the scheduling of multiple queries is extremely scarce.Because of that,the research on multiple query scheduling is carried out in this thesis.Our research can give full play to the performance of multi-processor systems and effectively improve the efficiency of query.We decide to adopt the multi-processor architecture after analyzing the existing large-scale database query system.And then,we propose an algorithm which is based on the Consecutive Retrieval Property.The algorithm can quickly find the query combination that occupies all or most of the processing nodes from the reconstructed query queue.The execution of this combination as a batch can fully exert the performance of the multi-processor system,thereby effectively speeding up the query of the database.The key of the algorithm lies in applying the reorganization strategy,so that the query data can be reorganized into a continuous relationship with C-R property,and then the data is continuously stored in the memory.CRPScheduling can solve the problem of idleness of some system resources in the existing algorithms..The CRP-Scheduling algorithm does not take into account the memory requirements of the system and the possible I/O time.It also does not consider the impact of slow queries on query performance.Therefore,we consider combining these factors with the CRP-Scheduling algorithm and try to reusing data that already exists in memory and avoiding idling resources due to executing slow queries.Finally,an improved algorithm called ICRP-Scheduling is proposed.We have verified through experimentation that regardless of changing memory size,resource size,the number of resources,and the frequency of slow query,the improved algorithm is significantly superior to other algorithms in terms of system utilization and system performance.Especially when the memory capacity is larger than the query processing memory requirement and the size of the resource are less than 320 K,the algorithm has more than 3 times higher system performance than the conventional Largest-Fit-First algorithm.This proves that the improved algorithm is practical and feasible,and it is of great help to improve the overall system performance of the database.We apply the improved query scheduling algorithm to the query-related module of the newly-built economic dynamic system,which greatly improved the system query efficiency.Then we also introduce the overall architecture of the entire system and the detailed design of each functional module.Finally,we demonstrated the operating status at the end of the thesis.