Design And Development Of ACM Task Scheduling System

ACM/ICPC (International Collegiate Programming Contest) is a contest of programming held by ACM organization yearly. In order to hold this contest, many universities around the world have developed their own ACM/ICPC contest platforms. Most of these platforms collect tasks via web service, and run the tasks on an independent server, then send the results to users after running. However, because the system could only run the tasks one by one, when a large amount of concurrent tasks are submitted to the system, it can't finish all task in time, thus the users have to spend too much time waiting for the result of their tasks.Aimed to solve the problem, the design also collects tasks via web service. It uses grid tasks scheduling technique to obtain reasonable allocation of all the tasks to multiple servers, achieving a balanced load and higher system efficiency. The system includes three parts. The first part is Portal Entrance module, the second part is System Management module, and the last part is Compute Resource module, which are functional independent. Core functions of the Portal entrance module are for the collection of tasks and the publication of information. Core functions of the System Management module are for the detection and management of resources, the forecasting and scheduling of tasks, and data management. The System Management module mainly include: the obtaining of remote server load, the maintenance of machine available time, the forecast of task run-time and the current mainstream grid task scheduling algorithm. Core functions of the Compute Resource module are for the acquisition of the tasks, the judging and the feedback of the tasks.This thesis introduces the background situation and research at home and abroad, and then introduces the Grid basic environment on which the design is based. The processing of task, system architecture, modules and functions of each module are described in the overall design chapter. In the detailed design chapter, design and implementation of the system and each functional module are described. In the chapter of performance evaluation, the thesis analyses the data generated by running, evaluates each task scheduling algorithms, and probes the suitable task scheduling algorithm for the system under different situations. Finally, the thesis makes a summary of the whole design, analyzes the existing problems concerning the work, and presents some enlightenment for future research.