Framework for parallelisation on big data.

Area of Science:

• Geophysics and Computational Science
• Big Data Analytics
• High-Performance Computing

Background:

• Seismic data processing requires significant computational resources due to large data volumes.
• Traditional software struggles with the scale and complexity of modern seismic datasets.
• Implementing parallel processing for seismic applications is complex and hinders speed improvements.

Purpose of the Study:

• To develop a simplified framework for parallel processing of big seismic data.
• To address the limitations of traditional data processing software in geophysics.
• To enhance the performance of seismic algorithms through efficient parallelization.

Main Methods:

• Utilized the Apache Hadoop framework with its MapReduce function for parallel data processing.
• Conducted experiments on the RedHat CentOS platform to evaluate the system.
• Focused on seismic algorithms, specifically stochastic inversion, for performance testing.

Main Results:

• Successfully implemented a parallel processing framework for big seismic data.
• Identified system bottlenecks within the parallel processing environment.
• Achieved improved overall system performance for seismic data algorithms.

Conclusions:

• The proposed Apache Hadoop-based framework offers a viable solution for parallelizing big seismic data.
• The study demonstrates the effectiveness of MapReduce in addressing computational challenges in seismic data processing.
• Optimizing the system through bottleneck analysis leads to enhanced performance for complex geophysical algorithms.