Concerning the problem that there is a lot of data which need to be real-time processed during the production process, the local processor, based on multi-thread and co-processing architecture and two data buffer mechanisms was accomplished. As a reference, multi-functional thread in Hadoop's parallel architecture has an impressed impact on the design of the local processor, especially MapReduce principle. Based on the user-defined architecture, the local processor ensures data concurrency and correctness during receiving, computing and uploading. The system has been put into production for over one year. It can meet the enterprise requirements and has good stability, real-time, effectiveness and scalablility. The application result shows that the local processor can achieve synchronized analysis and processing of mass data.