globalcomputinglab/somospie_flux repository overview

⁠Overview

SOMOSPIE⁠ is a spatial inference engine consisting of modular stages for processing spatial environmental data, generating fine-grained soil moisture predictions with machine-learning techniques, and analyzing these predictions. Given the critical role of soil moisture and this engine, we extend its execution to Kubernetes clusters. To this end, we use the Flux Operator⁠ which is a Kubernetes Cluster Operator that can be installed on a Kubernetes cluster to create and control Flux Framework⁠ MiniClusters to launch jobs.

The goal is to facilitate users to run SOMOSPIE on an HPC cluster in Kubernetes. The flux operator sets up a Flux instance to run across pods in a Kubernetes cluster orchestrating storage resources as needed and executing the SOMOSPIE's jobs. We will provide the instructions to execute the SOMOSPIE workflow with the Flux Operator.

⁠Acknowledgments

This project is a collaboration between the Global Computing Lab at the University of Tennesee, Lawrence Livermore National Laboratory, IBM, Sandia National Labs, and the University of Delaware.

⁠Websites

• SOMOSPIE⁠
• Flux Framework⁠
• Flux Operator⁠