The Research On Parallel Processing Model Of Private Cloud Towards Specific Domain

The rapid development of the Internet gave birth to the emergence of cloud computing, "cloud computing" is a integration of parallel computing, distributed computing, grid computing, grid storage, virtualization technology, gradually became a massive data computing and storage standard in reality. The core idea of cloud computing is used extensively for unified computing resource management and scheduling and constituting a pool of computing resources to provide on-demand service to users. Seismic data processing because of its large amount of data, large computing tasks, cloud computing technology to solve the problem of course is an important technology.Currently public cloud computing platform has not been fully applied, precisely because of its security is not very reliable. Therefore, taking into account the special nature of the field and the confidentiality of the data information, we use a scalable, more secure private cloud computing as a platform for seismic data processing. On the other hand, in order to ensure effective seismic data is processing correctly and completely, we propose a parallel processing model in cloud environment with fault tolerance ability. After experimental analysis, and achieved good results.Since the seismic data itself is generally large amount and low correlation, so the seismic data processing will be executed concurrently and pursued better efficiency. In practice, the cloud nodes can be added and canceled in accordance with the size of the work to make a private cloud in this paper with good scalability. Based on the available resources in cloud environment and features to divide jobs, real- time tracking resource load of each computing node, assign the appropriate computing tasks for it and achieve a higher concurrent execution efficiency in this paper. In order to ensure the completeness and accuracy of the data results, the parallel scheduling model proposed in this paper can track the completion of each gun data. If the data is not correctly used, then re-scheduling and imposed calculated to obtain the ideal experimental results.