Research On Storage And Query Method Of Mobile Object Deputy Database

Nowadays,mobile terminals,such as mobile phone,pad and smartwatch,have become indispensable in people's daily life,and the data types stored in mobile terminals are more and more abundant,and the data magnitude is more and more huge.Traditional file systems can no longer meet the needs of data management in mobile terminals,embedded database for mobile terminals have appealed much attention of researchers.Mobile terminals have the characteristics of finite internal and external storage capacity and limited computing performance.In addition,mobile terminals use NAND flash memory as storage medium which has the features of instant reading,unbalanced reading and writing efficiency and high erasure cost,which lead the design of database oriented to mobile terminals needs to be compatible with the features above.Currently,the mainstream embedded database in mobile terminals,like relational database,Key-Value database and graph database,performs excellently in various application scenarios.These shortcomings are mainly reflected in the following aspects:(1)Relational database can hardly afford high performance in grouping and aggregating query and query of entities' association.(2)Key-Value database may cause large amount of redundant data under the condition of guaranteeing high querying efficiency.(3)Graph database needs traverse all nodes to screen entities,which performs unsatisfactory efficiency.An entity is regarded as an object in Object Deputy Model,and objects with identical attributes and functions are abstracted into a class.In Object Deputy Model,two concepts are proposed: deputy class and deputy object.A deputy object can inherit several attributes from an object.The relations between deputy objects and objects are recorded in bilateral pointers,which store class IDs and object IDs,causing small extra storage overhead.Object Deputy database can provide remarkable efficiency in grouping and aggregating query and search of entities' association.Objects are connected into a net with bilateral pointers,which maintains object deputy database can obtain the data of an object by its associated objects,which is called cross class query.In order to solve the existed problems above in current embedded databases oriented to mobile terminals,based on the limited computing efficiency,finite external and internal storage capacity and storage hardware characteristics of mobile terminals,this paper proposed to optimize the storage and query strategy of Mobile Object Deputy database.The research contents and innovations of this paper mainly include the following aspects:(1)Based on the limited computing performance of mobile terminals,the overall architecture and functions of Mobile Object Deputy database are proposed.(2)A large number of bilateral pointers are stored in object deputy database,thus,the storage and reading efficiency of bilateral pointers has significant influence on the query performance of object deputy database.Based on the finite internal and external storage capacity,this paper proposes to compress bilateral pointer clusters by simplify the storage of redundant class IDs,which can reduce the I/O overhead caused by reading bilateral pointers in queries,and to optimize the cluster strategy of JOIN deputy class to avoid reading multiple bilateral pointer pages when querying a JOIN deputy object.(3)The optimized querying strategy of Mobile Object Deputy database is proposed,including the process of querying virtual attributes of deputy classes based on the bilateral pointer cluster mentioned above,the optimized data scanning method of cross class query to improve cache hit rate,rewriting the path expression in crossclass query to further reduce reading redundant data pages.(4)Based on the imbalance between reading efficiency and writing efficiency of NAND flash memory,a new query cost model of Mobile Object Deputy database is proposed and compared with the query cost of relational database.Finally,this paper has optimized the storage and query method of Mobile Object Deputy database,which is implemented based on open-source embedded database SQLite.In addition,this paper tests the storage overhead,the efficiency of query of virtual attributes,the performance of cross-class query and the impact of cache size on query performance of these three kinds of database,verifying the validity of the optimizing scheme mentioned above.