High Performance Computing on AWS Cloud for Engineering Simulation

Manufacturing companies across sectors such as Automotive, Industrial Products, and Heavy Equipment are increasingly reliant on Engineering Simulation through CAE tools like LSDYNA, Abaqus, Pam Crash, Fluent, among others.

This heightened dependence has led to a surge in demand for High-Performance Computing (HPC) capacity. However, this demand is compounded by the variable nature of demand, driven by scheduled programs. Consequently, accurately projecting capacity requirements has become an intricate and often cost-prohibitive endeavor.

To tackle these complex challenges effectively, an on-demand and adaptable capacity model comprising savings plan emerges as the most fitting solution. We have meticulously crafted a proven methodology designed to assess, activate, and seamlessly migrate on-premises engineering simulation workloads to the AWS cloud. This structured process encompasses the following key steps:

• Develop a business case for migration: Lay the foundation for the migration project by outlining its rationale, benefits, and expected outcomes.
• Compile an inventory of tools and technologies: Discover and assessment of the comprehensive list of on-premises cluster components and software applications essential for migration planning.
• Discover job submission workflows: Identify and document the processes and mechanisms currently used for job submissions on the existing cluster.
• Document software versions and dependencies: Catalog the distinct software versions in use and their interdependencies to ensure seamless migration.
• Assess license server requirements: Evaluate the licensing needs of software applications and tools to guarantee compliance during migration.
• Setup Landing zone architecture: Design and implement the sophisticated landing zone architecture considering the network efficiency, security and reliability.
• Setup the HPC cluster in AWS cloud: Establish the High-Performance Computing (HPC) environment in the AWS cloud using custom-built scripts.
• Shortlist pilot jobs for migration: Select a small, representative set of jobs for initial testing and migration.
• Migrate pilot workloads: Execute the migration of the chosen pilot jobs to validate feasibility and address potential issues.
• Validate pilot migration outcomes: Thoroughly review the results of the pilot migration to ensure the intended benefits are achieved.
• Define a plan for full workload migration: Develop a comprehensive strategy for migrating the entire workload, incorporating lessons learned from the pilot phase.
• Execute the full workload migration plan: Implement the migration plan to move the entire workload to the AWS cloud, following best practices and minimizing downtime.

Post-migration, we offer operational support for tasks such as cluster management, software version upgrades, and MIS/cost reporting.