SDIS: A PB-level seismic data index system with ML methods

Seismic data processing is one of the important technologies in the oil exploration industry. The original seismic data volume is collected from the field, its size has exceeded petabyte level. The seismic data processing requires random access to the large seismic data set and usually selects different keywords for multiple index construction. Multi-keyword (in Trace Header, as shown in Fig. 1) range query and read-only are main features of seismic data access. A high-performance index is the key to improve the application processing efficiency. In PB-level data situation, the major challenges are: