A Design And Implement Of Job Execution Control System Used In Seismic Data Processing

The continuous development of information technology makes the demand ofcomputing power rising, the speed of computer hardware developing is seriouslylagging behind the pace of development in the calculation of the scale. Distributedcomputing expand a new way to improve computer’s performance: by running one taskon multiple computer nodes concurrently, achieving horizontal expansion, and avoidcomputers’ physical computing power bottleneck. Cloud computing developing rapidlyrecent years, which bring many new computing concept and new technology, includingdynamic scalable computer cluster system based on virtualization technology,efficiently parallel computing programming model, and so on. This thesis reference anduse the concepts or technologies, proposed a distributed computing solution of seismicdata processing, based on its own Characteristics. Design and implement a distributedjob control system used to process seismic data.Job control system is a computing framework used to process seismic data. Theexisting Job control system is centralized, which means each computing task can onlyrun on a single computing node, which makes computer hardware became a limit to thewhole system, also makes the cluster could not be fully utilized. This thesis use theconcept of horizontal expansion, running a seismic data processing task on multiplecomputing node concurrently, avoid the bottleneck of a single computer, which reducethe computing time a lot.The distributed job control system design and implement by this thesis solve theproblem existing in centralize job control system, such as low utilized of cluster system,too long to compute a big computing job, which greatly improved the systemperformance. The main work of this thesis:(1) By analyzing the existing distributed computing and cloud computing coretechnology, researching the development status of the current seismic data processingsoftware, selecting a appropriate technology solutions, design a distributed computingframework based on the characteristics of the seismic data processing, which containsuser interface, scheduling control layer and the job running framework. (2) Design and implement a communication solution used in distributed Jobcontrol system. Use boost::asio’s proactor mode to achieve a communication server,which can support a large number of high concurrently communication. Design a datatrans tunnel, which used to trans data between sub tasks. Data trans tunnel also providecommunication address query, Flow control, control signal.(3) Design and implement a way to allow user define group themselves, design ascheduling method to fit distributed job control system, to achieve schedule by group.(4) Check the efficiency of distributed job control system by running real seismicdata processing task, and compare with centralized job control system.