Performance Optimization Of The Crowdsourcing System Based On Main Memory Database-Redis

The meaning of the crowdsourcing is that enterprises or individuals outsource the task to the non-designated, and usually a large number of external workers. Because of this kind of characteristic, the capacity of dealing with high concurrency is very important. A bottleneck of improving system performance is the operation of the database.as many relational database will access the disk for all kinds of operations, the speed of operation is limited by the I/O speed. It is physically difficult to improve. It is precisely because of these shortcomings of traditional relational database, people developed a lot of main memory database, offering wide ways to enhance the database.In this paper, I design and implement a scheme to improve the performance of the crowdsourcing system which using MySQL by combining Redis. It is aimed at improving the performance of the system with the minor reconstruction of the key process. For task assignment and collection, I design the data model of redis which has universal significance; for accessing user characteristics, I design data model of redis which support rapid ranking by double attribute using atomic package features of Lua. In addition, using the MySQL trigger and user defined function, I develop a dynamic link library to achieve the mechanism which automatically transfer newly inserted data from MySQL to Redis. This paper also introduces the detail of the implement of improving the system. I test the task allocation and collect,access of users characteristics, and the whole server, respectively to testify the effect of improvement comparing with original system.