HPC Workload Migration and Optimization to AWS
Context
DeepLife develops novel approaches to drug discovery based on state-of art multi-omics data, machine learning, and system engineering.
Their growing computational needs were held back by outdated orchestration paradigms running on-premises resources.
Aneo was mandated to analyze their HPC workloads and propose and execute a plan to migrate them to AWS.
Objectives
Aneo has accompanied DeepLife on a journey that has brought them from executing Snakemake workflows on an on-premises Slurm cluster, to functionally equivalent Nextflow workflows executed on a tailor-designed architecture on AWS.
Production workload continuity has been ensured during development with a custom-made AWS-Snakemake executor that allows to run Snakemake workloads on AWS without the need of third-party software.The now legacy workflows did not have fast and reliable access to genomic data sources. As part of the new orchestration design, Aneo has implemented the means to reliably procure genonic data from DeepLife selected providers.
Results
-
A new scalable, secure, reliable, optimized and available architecture on AWS designed for big genomics computing pipelines.
-
Reduced workload orchestration overhead leading to a significant reduction of computing times.
-
A secure, fast and easy to use system to download genomic data from remote archives.