论文部分内容阅读
地震勘探资料数据处理是地震勘探重要而不可缺少的组成部分,当前地震勘探资料数据处理具有处理的数据量大、处理过程多以及资料的重新处理,反复处理等典型特征。因传统的地震勘探资料处理系统中各个处理模块之间采用临时文件的方式来传递数据,这些特征反映到传统的地震勘探资料处理系统中将面临的一个核心问题:巨大的磁盘I/O访问量。这个磁盘I/O访问量不仅跟地震勘探数据的大小有关,而且跟处理流程中的处理模块数有关。为此,我们设计数据流软件flobus来减少地震勘探资料处理流程中对磁盘I/O的访问量。在采用数据流flobus的地震勘探资料数据处理系统中,处理流程中总的磁盘I/O访问量只跟输入数据大小有关,而跟处理流程中的处理模块数无关,从而大大减少了磁盘I/O的访问量。数据流软件flobus充分考虑到提供多模块处理复杂系统的数据流共享机制,其主要特征包括:1)提供多任务的数据流控制,实现数据驱动的并发机制;2)提供多模块的数据聚合机制;3)支持同一级别模块间的数据依赖关系;4)支持任务流中模块的动态加载。
Seismic prospecting data processing is an important and indispensable part of seismic exploration. At present, seismic data processing has a large amount of data to be processed, many processing and data re-processing, repeated processing and other typical features. The transfer of data due to the use of temporary files among the various processing modules in a conventional seismic data processing system reflects a central issue that traditional seismic data processing systems will face: huge disk I / O traffic . This disk I / O traffic is not only related to the size of the seismic data, but also to the number of processing modules in the processing flow. To this end, we have designed the data streaming software flobus to reduce the amount of disk I / O accessed during seismic data processing. In the seismic exploration data data processing system using flobus data stream, the total disk I / O traffic in the processing flow is only related to the input data size and has nothing to do with the number of processing modules in the processing flow, thus greatly reducing the disk I / O’s traffic. Flobus full consideration to provide multiple modules to deal with complex system data flow sharing mechanism, the main features include: 1) to provide multi-task data flow control, data-driven concurrency mechanism; 2) to provide multi-module data aggregation mechanism ; 3) support for data dependencies between modules at the same level; and 4) support dynamic loading of modules in the task flow.