Optimization And Testing Of Large - Scale Networked Multimedia Information Publishing System

With the rapid development of information technology, Multimedia Publishing System (MPS) has been widely used in hotels, hospitals, banks, shopping malls, transportation, media, community and so on, provides modern city with convenient, variety, real-time transmission of information. Large-scale commercial multimedia information publishing system has large amount of resource data and large number of terminal,the performance of the system also is particularly important. At present, most of the commercial multimedia information publishing system are based on the layered Web architecture with a large number of data access and data communication, database calls is much more costly than the system memory operations, which is also the main factor that influence the performance of system.Based on the detailed characteristic analysis of the commercial multimedia information publishing system, it is concluded that the system has a complex business logic and complex data relationship and high user operating linkage. Combined with the observation and analysis of the actual multimedia information publishing system, it is concluded that the editing and publishing of the program list is the most frequent user operation, also is the most significant problem of the system performance, which is cost by the complex business logic and high data persistence workload and large amount of database SQLs. Based on the analysis above, I optimized the system business logic layer and data persistence layer and database layer, these three aspects to improve the system data performance. In business logic layer, I focused on the program list editor, which is also the core function of the system, the optimized design of program list editor simplifies the program initialization process, reduces the coupling of programs with other data, thereby reducing the query of other associated data. In data persistence layer, I chose Hibernate framework as the persistence layer design based on the detailed analysis. In addition, I summarized the reasonable and efficient way to use the Hibernate framework in this kind of system, including using the reasonable fetching strategy to reduce the correlation data query; and using Hibernate cache mechanism; and using the Servlet filters to manage the Hibernate Session cache, in order to reduce the number of the Session object in system and make full use of the Session cache. In this way, the database connection can be greatly reduced. In the database layer, I improved the performance of the database server by the work of the database design optimization, database SQL optimization and database parameters tuning. In the end, experience are did to evaluate the multimedia information publishing system before and after the optimization, besides, I used the Hibernate statistical mechanism to statistic the resource consumption of the system. Results of the comparative experiment show that the optimized system has less page response time, the user experience has significantly improved, and the system database connection number with the amount of the Session object has greatly reduced. Experiments show that the optimization methods mentioned above can be integrated effectively improve the performance of the multimedia information publishing system.