Reducing the Time to Science of Ambient Noise Tomography with ANXCOR and Kubernetes

Session: Applications and Technologies in Large-Scale Seismic Analysis
Type: Oral
Date: 4/28/2020
Time: 09:15 AM
Room: 120 + 130
Description: 

We have developed the Python package ANXCOR: Ambient Noise X-CORrelation to help Ambient Noise Tomographers produce extensible and reproducible science. ANXCOR’s PyPI and Anaconda-based dependency management system make it an attractive option for both local and cloud-based high-performance computing environments.

Kubernetes is a popular container orchestration tool able to deploy Python packages to both local and cloud HPC platforms such as Azure, Amazon Web Services and Google Cloud. Here we demonstrate ANXCOR’s workflow deployed in a Kubernetes orchestrated container. We report on metrics related to data scope such as compute time and resource needs and provide insights on the added knowledge domain complexity needed to deploy ANXCOR to cloud-compatible HPC platforms.

Presenting Author: Kevin A. Mendoza

Authors