Index Of Seismic Data File System

The rapid development of economy and energy shortages led to the growth of oil demand, under this requirement, as the most effective way of oil exploration, seismic exploration is developing towards high precision, high density, multi-dimensional.Resulting large scale seismic data need to be stored, analysis and processing, and the seismic data has a characteristics of small piece and large number. The processing speed of read mode used in the conventional disk is slow, so not only to improve the underlying storage system for seismic data, to join index mechanism to improve the speed of data access is also essential.So study a new index structure and algorithm for the seismic data storage system becomes very meaningful.Indexing technology can effectively improve the ability of handling the data query, reduce the query time, reduce the number of disk I/O access. Access for seismic data,the current plan is to use an index which store separated from data,index server adopt centralized management,this cause memory overflow,single point of failure and other issues. Some scholars puts forward different schemes for large data index,uses hierarchical index based on the characteristics of keyword,make the time and space costs become lower,but the index after classification is also store in the same node.Seismic Data File System(SDFS)system make the chunk of data in partition and make different pieces evenly distributed to each disk to store.This thesis take the advantage of the characteristics of distributed and hierarchical type and special access requirements for seismic data of SDFS, put forward a two levels of index structure to record data block after position based on the Shot number and gather number. After the study of two level index, this thesis adopt the Trie tree, red and black tree and AVL tree index for the implementation. Finally put forward a kind of T*-Trie tree to achieve the primary index. Results show that the two level indexes which implementation of T*-Trie tree can meet the needs of system random read and sequential read, and increase the speed, make the the memory consumption of metadata server got lower.