Simmulationswissenschaftliches Zentrum Clausthal - Göttingen - Numerically intensive simulations on an integrated computing infrastructure (SWZ)
The selection of the most suitable application and computing resources is an important success criterion for the usually computationally intensive and time-consuming scientific simulations. However, for users of simulation software, the selection of a suitable computing resource is not a trivial task in view of the heterogeneous infrastructure of modern data centers. In addition to the decision between IaaS, cloud services, HPC clusters and computers with special accelerator cards such as GPUs, learning the technical details also takes time.
The aim of the research project is to simplify access to heterogeneous computing infrastructures for the simulation sciences and to integrate various specialized computing resources on a user-friendly platform. The SWZ will use efficient methods and techniques for the IT-supported selection of the computing resource that is best suited for the specific application, whereby the provision and execution will be semi-automatic.
The framework of the required integrated computing environment will be a simulation-as-a-service (SimPaaS) component from the user's perspective, which automatically provides and executes simulation applications in a cloud environment. In the background, the application is defined in the application layer using a model. A distribution controller then distributes the mapping of the application to a specific resource type (in the resource layer), such as a cloud system, HPC cluster, GPU cluster or grid system. The application model is transformed into a resource-specific input model for the respective target system and transferred to the resource manager of the target system who executes the application on the corresponding infrastructure.
GWDG's Role in the Project
The main task of GWDG is to develop the distribution controller. This compares the requirements of a simulation described by the application model with the capabilities of the various computing resources and automates the selection of a suitable resource. It also transforms the application model into a platform-dependent execution model, depending on the selected resource.
GWDG is also responsible for the documentation of the developed component and its transfer to a production environment.
- ITM - Institute of Applied Mechanics, Technische Universität Clausthal
- Software Engeneering for Distributed Systems, Institut of Computer Sceince, Georg-August-Universität Göttingen
- Associated Partner: Research group Aerodynamics and Flight Mechanics, Dep. Aeronautics, Astronautics and Computational Engineering, Faculty of Engineering and the Environment, University of Southhampton, England