2011-12-06 Sinterklaas goes cloudy and HPC

Just like last year, AZUG planned an event on December 6, the day "Sinterklaas" delivers toys and candy to a lot of children. To reach such a broad audience quickly, he requires some high-performance computing. And that's exactly what Kurt Claeys joins us for on this tuesday!

Session info

High Performance Computing with Azure

High performance computing (or HPC, formerly known as TC) is all about massive number crunching, performing loads of parallel calculations, solving complex simulation models with many different parameters, rendering images … These are workloads that typically need loads of CPU power to perform and where the time to finish the calculation is important. So why don’t we use Azure compute for this ? Azure provides compute nodes on demand. Typically these calculations only need to be done once, so Azure could solve the problem without an investment in hardware. Scenarios in HPC are : financial simulations, insurance predictions, video rendering, chemical calculations, … even weather and earthquake forecasts.

In this session we see how Windows HPC server (https://www.microsoft.com/hpc) can use Azure to spread out the calculation logic across many workerroles. We discuss the typical scenarios and show how they are implemented in some demos : a number crunching executable, rendering images and a financial model in Excel offloading calculations to Azure. We’ll see both the developer aspect and configuring the HPC Server part.

After this session you’ll realize that Azure is not only about hosting web apps or storing data. It’s also about massive compute power on demand. HPC could be one of the most important workloads for Azure in the future.

Speaker: Kurt CLAEYS, working for Microsoft as Technology Solution Professional for Windows Azure where he guides clients and partners in their adoption of the Azure platform on the technical level. Before joining Microsoft he was a .NET solution Architect/Trainer and active in the .NET Community as MVP Connected Systems.