UC-NGI-NW-III
From EGI Knowledge Base
Use Case Title: Virtual Research Environments for Scientific Computing
Short Description: With this use case we suggests to employ the grid infrastructure to facilitate interdisciplinary collaboration between scientists interested in Scientific Computing using High Performance Computing facilities. The intention is to create a virtual research arena as a host for locally distributed researchers. The focus is on the sharing of resources by the distributed community, as it is common for grid applications. In contrast to the common approach, our intention is to use the grid infrastructures as a shared work space. The shared entities are computing resources, data and software. Communication channels (message boards, collaborative work spaces, video-conferencing) will be added. The intended workspace should bring together scientists from different research areas, namely application engineers, applied mathematicians and computer scientists. Computing centers will support the activities. The collaborative workspace will be realized as a Virtual Organization (VO), a key concept in grid computing. It is the ideal structure to realize distributed environments. The research focus of the intended Virtual Organizations is on compute-intensive simulations in science and technology, where numerical investigations have proved to be an indispensable tool to understand the nature of underlying processes. Solving real-world problems requires this deep knowledge of the processes and of the complex mathematical models. An interdisciplinary approach involving experts from different research communities is often indispensable. The target numerical simulations require usually the usage of High Performance Compute resources. A cooperation with High Performance Computing centers is essential.
Implementation: Virtual work arenas shall be established on the base of the European grid infrastructure. The initial set-up includes the establishment of a corresponding Virtual Organization, the assignment of grid resources and the realization of a transparent web-access via web-service based portals. The web solutions will realize communication channels to enable a vibrant exchange. In this framework, grid applications related to resource-intensive engineering applications will be established. The virtual research environment starts with the implementation of a simulation environment, which runs on massively parallel distributed memory machines with high scalability. Griddification of the software is the first task. This comprises the realization of standard interfaces for I/O as well as interfaces for general model configuration. Standard model problems will be realized being the research base of the related investigation both in direction of scientific computing and model development and research. We can identify the following tasks:
- Task 1. Creation of the Virtual Organization and assignment of initial resources. The Virtual Organization will be used to manage the computational resources and software available to users. The Virtual Organization will also allow administrators to install and configure grid applications.
- Task 2. Development of a web-service based portal. Realization of communication channels (message boards, collaborative work spaces, video-conferencing, etc.). The portal will allow for a transparent access to data-repositories, with the potential for a "single sign-on". Single sign-on means that user authentication is still secure and uses the certification methods already in place, but some of the complexity of handling certificates is taken care of by the portal infrastructure.
- Task 3. Grid-enabling of the simulation software. This comprises the development of standard interfaces for file-I/O and model configuration. These interfaces will be abstract and will enable the virtual community to apply and compare simulation results with results from other simulations.
- Task 4. Training of engineering partners in the usage of grid infrastructures. The training includes a basic understanding of grid concepts, the usage of basic (command line) tools as well as the usage of the portal solution. The training in usage of the simulation software includes a basic understanding of the conceptual models and mathematical formulations as well as the usage of the program-environment on application level.
- Task 5. Development of a representative set of model problems. This set will be a testbed to investigate engineering as well as mathematical/numerical questions. Engineering questions may be related to real-world applications, fundamental questions to issues of mathematical/numerical nature.
This implementation plan is understood to realize the basic structure for a fruitful work environment and will be established in the first implementation period. In a second step, the continuation will lead to a scientific computing work arena in which development of grid-infrastructure, mathematical methodology, software tools and models take place with mutual interaction.
Perspectives: The implementation of research infrastructures as described above will ease the development of European cooperations based on computational resources and interdisciplinary exchange. Scientific Computing as the key computationally-oriented scientific approach will be supported by the large European compute-resources. The approach bringing together leading experts of application areas with the most powerful compute resources will lead to a new dimension on scientific investigations. Investigations of fundamental importance to the European community, like risk assessment related to CO2-sequestration or high-throughput imaging in systems-biology can be treated appropriately in the presented environment.
