HPC Consulting- Implementation: 3 Weeks

High performance computing is the ability to process data and perform complex calculations at high speed. CrunchYard consults with clients to help address their performance needs through HPC practices.

The process includes:

• Needs Analysis (i.e. what are the pain points)
• Proposed solution (i.e. how can we address the pain points)
• Implementation (we build the proposed solution)

Outcomes and deliverables:

• Comprehensive report on HPC needs analysis
• Comprehensive technical report on proposed HPC solution based on the HPC needs analysis.
• Implementation of proposed HPC solution making use of Azure VM's and associated resources
• Comprehensive technical report and user manual for implemented HPC solution

Highlight of Azure Services provided:

• Parallel file system optimization using Azure Virtual Machines and HDDs, SSDs or NVMes
• Network Latency reduced through Azure proximity groups and Azure Infiniband networks
• Azure VM's configured as HPC compute nodes, optimized for specific workloads
• Combining traditional resource managers such as PBSPRO and Slurm to work with Azure Batch
• Building network subnets and network security rules on Azure to provide security and performance
• Building hybrid HPC systems (as needed) to synchronize between Azure and on premises facilities

Solution examples include:

• Parallel file systems for financial services (Lustre, Gluster, BeeGFS used for SAS etc.)
• HPC clusters for Fluid Dynamics, Computational Chemistry, Exploration Geophysics etc.