基于数据库的并行抽道集技术

Parallel gathering based on database

针对目前地震海量数据处理的需求,本文提出一种高效并行道集排序方法。利用地震数据按线和道多级关键字管理的特点,将数据分块,分发给多个节点同时进行初级关键字排序,利用数据库存储过程,控制并发进程之间的同步和异步,由最后一个进程将按初级关键字排序后的道集信息进行合并,继而再次分块进行下一级关键字的并行排序,实现了海量地震数据分块多级并行排序。实际地震数据排序结果表明该方法切实可行,效果良好。

For mass seismic data management in the current seismic processing system,a rapid gathering method is put forward in this paper using parallel sorting.Because seismic data is organized by multi-level keywords such as line and trace,it can be divided into several parts for parallel sorting by primary keywords on computation notes.The last task merged all of the firstly sorted results at the control of the inter-process synchronization through the storage procedure of database.The merged information can also be divided to perform parallel sorting by the next level keyword.Finally,the multi-level parallel sorting is rapidly achieved by parting the massive seismic data.Based on this sorting method,we develop the parallel data loading and parallel gathering modules in our seismic processing system and achieve good results in real seismic data processing.