Design And Implementation Of Meteorological Database System Based On Adaptive Indexing

With the ceaselessly accelerating of the meteorological modernization process,the requirement is higher and higher for meteorological business such as information retrieving speed in all fields of society.However,the amount of meteorological data has increased rapidly as time goes on,which will affect the query speed directly.Creating the high performance database index is the most common way to speed up the query.Traditional database index relies on two core assumptions:(1)the patterns of query workload is available,and(2)there is sufficient idle time to create the indexes.Unfortunately,these assumptions are not valid in modern application environment.On the one hand,the patterns of query workload is no longer available or constantly changing.On the other hand,the data may be queried as soon as it arrives.Traditional database index is not only very time consuming when you create and update index but also it has the problem of not sensitive to the query,so the traditional database index cannot meet the requirements of the development of meteorological modernization.The emergence of adaptive indexing technology provides a good solution to these problems.This thesis first briefly describes the research background of adaptive indexing and the research status at home and abroad.Secondly,this thesis describes the basic principles of adaptive indexing in detail.Finally,based on the previous study and the characteristics of adaptive indexing,we proposed an improved hybrid adaptive indexing algorithm named HRCS,which optimizes the steps of ranking on the initial partition and incorporates the strategy of data reconstitution in database cracking algorithm.Additionally,it stores cracker index by using Partitioned B-tree.HRCS has some obvious characteristics:(1)There is no need to create an index in the process of data preprocessing;(2)The creation of the index is part of the query and the database can automatically adjust itself to fit the workload during query execution;(3)The algorithm only creates the indexes for part of the data and optimizes the index structure during query execution.Only the indexes of data which were frequently queried will be optimized,which results in higher speed when queried more frequently.The goal of adaptive indexing algorithm is to seek a lower initialization cost and faster convergence rate.HRCS algorithm overcomes the problem of slow convergence of database cracking algorithm as well as the big cost of initialization of the original adaptive merging algorithm.Additionally,we propose an optimization for robustness by performing an extra random split after the query-driven split,which makes our hybrid algorithm suitable for multiple patterns of query workload.Extensive experiments were conducted to demonstrate the effectiveness and efficiency of our method.The experimental results show that HRCS algorithm can solve the problems of difficulty on indexing and slow speed of querying faced by meteorological data when the amount of data is increasing.Various tests show that HRCS algorithm has very good performance in multiple queries on large amounts of data,and it has a good prospect in physical database design in the future.