Cloud CFD Course
CFD Direct announces the release of their new 1 day Cloud CFD course, the fourth course in its portfolio of OpenFOAM Training. The course enables users to run CFD with OpenFOAM on AWS. It trains the tools and processes for inexpensive, secure and efficient cloud use.
The course has been created by Chris Greenshields (OpenFOAM co-founder) and is initially scheduled as follows:
- 25 May 2018: Virtual, Americas
- 5 July 2018: London, UK and remote attendance
Course Modules | Who Should Attend | Further Details | Schedule and Booking
CFD Direct From the Cloud™
The case for using cloud for CFD is compelling, based on cost — see “Cost of Cloud CFD” below. Therefore we provide the tools to make it easy to run OpenFOAM in the cloud:
- CFD Direct From the Cloud (CFDDFC): a complete OpenFOAM cloud computing platform that includes the latest version of OpenFOAM and supporting software running on the latest long-term support (LTS) version of Ubuntu GNU/Linux.
- A command line interface (CLI) for CFDDFC to make it easier and quicker to launch, manage and run CFD on cloud instances on Amazon Web Services (AWS), the market leader in public cloud.
Aims of the Course
The Cloud CFD course teaches users to run CFD with OpenFOAM using these tools. It addresses the main concerns of organisations considering the use of public cloud:
- Cost Management: breakdown of the costs of cloud CFD; cost reduction, including use of spot pricing, reducing data transfer, efficient storage; monitoring cost with budgets.
- Security: administration of AWS; identity and access management (IAM); managing and storing key pairs; strong security group settings; firewalls.
- Efficient CFD: configuring the CLI; making cloud local; managing instances; running applications; production CFD; hardware configuration; parallel performance.
- Data Management: objective CFD data, plugin post-processing, quick visualisation, transferring data, data transfer costs.
Cost of Cloud CFD
CFD is a software application with fluctuating demand on computer resources. For example, a user might create a prototype CFD simulation with thousands of cells running on 1 CPU core, then scale up to many millions of cells on multiple cores, e.g. 4-32. They might next want to conduct a parametric study, running several muti-core simulations concurrently, before a long period of no simulations while they reflect on results.
Purchasing hardware for CFD presents a dilemma: if it meets peak demand, average utilisation is generally low; if it does not meet peak demand, productivity is low. Cloud, however, is cost-effective by maintaining high utilisation by sharing resources across multiple applications. Applications like CFD do not need high (100%) availability, e.g. unlike a web server, so can additionally exploit the cheaper price of spare capacity, e.g. using spot pricing on AWS.