Run high-performance computing (HPC) applications on Azure

In this module, you will:

• Identify the HPC and batch solutions available on Azure
• Identify the scenarios where HPC and batch technologies are a fit

After you complete this module, you'll be able to:

• Create an Azure Batch job by using the Azure portal
• Run an Azure Batch job by using the Azure portal

In this module, you will:

• Create an Azure Batch job in the Azure CLI
• Run an Azure Batch job in the Azure CLI
• Check status and results of Batch job in Azure CLI
• Monitor a Batch job with Batch Explorer

In this module, you will:

• Build and run an Azure Batch job from a .NET app
• Add an application package to a Batch account
• Interact programmatically with Batch and Storage resources
• Manage a Batch programmatically

By the end of this module, you'll be able to:

• Install and configure Azure CycleCloud to work with your Azure subscription.
• Integrate a third-party scheduler.
• Create and deploy a basic HPC cluster on Azure.

By the end of this module, you'll know more about how to:

• Customize HPC cluster architecture.
• Customize software and applications that you use with Azure CycleCloud clusters.

In this module, you will:

• Document the file-access architecture for your HPC use case
• Consider the implications of your architecture requirements
• Compare several file-access architectures

By the end of this module, you'll be able to:

• Describe the factors that determine throughput limits of Azure NetApp Files volume.
• Choose the best service level of Azure NetApp Files for HPC applications.

By the end of this module, you'll be able to:

• List the best practices that improve Azure NetApp Files performance.
• Describe the performance impact of the best practices on FIO and EDA benchmark suites.

By the end of this module, you'll be able to:

• Evaluate whether HPC Cache is appropriate for improving the productivity of your high-performance computing applications.
• Describe how HPC Cache enables you to accelerate, scale, and manage your file-based high performance computing workloads in Azure.

By the end of this module, you'll be able to:

• Understand what Azure HPC Cache is.
• List the pre-requisites for Azure HPC Cache.
• Create an Azure HPC Cache instance.
• Configure Azure HPC Cache to accelerate a Linux-based file system.
• Attach virtual client machines to read from the cache.

In this module, you will:

• Choose the best and most cost-efficient HPC SKU for each type of tightly coupled HPC application.
• Optimally build and run each type of tightly coupled HPC application on Azure HPC SKUs.

In this module, you'll learn how to:

• Troubleshoot runtime failures for tightly coupled HPC applications.
• Troubleshoot tightly coupled HPC applications that are underperforming.

• Introduction
• Azure Batch
• Azure VM HPC instances
• Microsoft HPC Pack
• Scenarios for HPC and Batch technologies
• Summary

• Introduction
• Create a Batch job by using the Azure portal
• Exercise - Create an Azure Batch job and virtual network
• Run tasks and add application packages
• Exercise - Add a new application package to a Batch account in the Azure portal
• Consider features for Azure Batch security
• Knowledge check
• Summary

• Introduction
• What is Azure Batch?
• Design an Azure Batch job to run your parallel task
• Exercise - Create an Azure Batch job in the CLI to run your parallel task
• Exercise - Monitor your Azure Batch job in the CLI
• Visualize batch jobs with the Batch Explorer
• Exercise - Monitor and visualize your batch job in Batch Explorer
• Knowledge check
• Summary

• Introduction
• Manage Azure Batch workloads from a .NET app
• Exercise - Setup Batch and Storage accounts in the Azure portal
• Access your Batch account using the .NET client library
• Exercise - Access your Batch account using the .NET client library
• Exercise - Create a pool of compute nodes to run our jobs
• Create and run a Batch job using the .NET client library
• Exercise - Upload resources to be used by a Batch with the .NET storage library
• Exercise - Manage and deploy applications on compute nodes
• Exercise - Add code to monitor and log Azure Batch job progress in your app
• Summary

• Introduction
• Describe the basics of Azure CycleCloud
• Describe Azure CycleCloud scheduler integration and autoscaling
• Exercise - Install and configure Azure CycleCloud
• Exercise - Create an HPC cluster from built-in templates
• Knowledge check
• Summary

• Introduction
• Describe Azure CycleCloud cluster templates
• Exercise - Add a disk to an HPC cluster's scheduler node
• Exercise - Add a custom node array to an HPC cluster
• Customize software installations
• Exercise - Create a cloud-init script for use with Azure CycleCloud
• Knowledge check
• Summary

• Introduction
• File access for HPC jobs
• File-system performance considerations
• File-system traffic considerations
• File-system authentication and authorization
• NFS overview
• Parallel file systems
• Hybrid file access requirements
• Summary

• Introduction
• Identify the decision criteria
• Choose your service level
• Knowledge check
• Summary

• Introduction
• Overall performance suggestions
• Mount options and client VM configurations
• Benchmark results
• Knowledge check
• Summary

• Introduction
• What is Azure HPC Cache?
• How Azure HPC Cache works
• When to use Azure HPC Cache
• Knowledge check
• Summary

• Introduction
• Identify the requirements for HPC Cache
• Create an HPC Cache instance
• Connect local storage to HPC Cache
• Connect HPC clients to HPC Cache
• Knowledge check
• Summary

• Introduction
• Select the appropriate HPC SKU
• Knowledge check - Select the appropriate HPC SKU
• Process pinning considerations
• Exercise - Find the correct NUMA mapping for an HPC SKU
• Compilers and numerical libraries
• Knowledge check - Select the right compiler and numerical libraries
• Select the right MPI library
• Knowledge check - Select the right MPI library
• Summary

• Introduction
• Health checks for HPC virtual machines
• Knowledge check
• Debug performance bottlenecks
• Knowledge check
• Summary