Design And Implementation Of The Integration And Management On Heterogeneous Scheduling System Of The Mobile Reading

China Mobile's digital reading service is an innovative practice in the field of big data. Digital reading platform is built on the data warehouse which provides a large number of user's basic data. Data warehouse implements the personalized recommendations by collecting and analyzing these data, mining user's reading preferences, and recommending similar resources to the user.With the continued expansion of the users and service scale, the data warehouse become larger and more complicated, many problems appear during the system routine maintenance. The low task scheduling efficiency and the high system resource consumption become the main problems of the data warehouse. Besides, the searching and updating of the task configuration become more difficult, the tasks execution fails more frequently.Based on the existing problems of the data warehouse, this thesis focuses on optimizing the scheduling mechanism, designed and implemented the schedule-merged subsystem, integrated the heterogeneous scheduling resources, screened the differences between the two scheduling systems and unified the task scheduling specification.Besides, this thesis integrates the static and the dynamic resources of the warehouse system as well. In terms of the static resource management,the configuration management subsystem is designed and implemented.In order to facilitate warehouse users to configure and retrieve the task of the warehouse, the subsystem extracted all task configuration into an unified configuration platform. In terms of the dynamic resources of the warehouse, the task monitor subsystem is designed and implemented, this subsystem is responsible for monitoring and managing the task states and operation results, allowing the warehouse manager to grasp the scheduling system's health states opportunely. At the end, these subsystems' functions are verified by test instances, and indeed provide solutions to the problems in tasks scheduling, configuration, and monitoring. The thesis also points out the weakness of the work as well as the directions to optimize.