This workshop will explore challenges to the effective use of cluster and grid computing posed by several emerging trends: the rise of multi-core processor architectures, the use of hybrid architectures that combine multiple types of microprocessors, and the increasing data demands of scientific applications. Challenges to be considered include scalability, component interoperability, reliability and fault-tolerance, new programming paradigms to simplify working with highly parallel systems, performance monitoring, and system management. A motivation for the activity is the growing gap between the application codes for scientific research, many of which use algorithms originally designed for serial computers, and the increasing scale and hierarchical depth of the parallelism in modern large-scale computational clusters and grids. The design of the workshop is intended to exploit the commonality of some of the challenges associated with programming and managing clusters and programming and managing computational grids. The workshop will: survey and analyze the key issues associated with deployment, operations and use; document the state of the art in these areas; and identify the promising directions for research and development.
The workshop organizers will actively encourage participation by members of under-represented groups. The workshop will foster dialog between the members of the academic and industrial communities interested in its theme. A report of the workshop will be published and openly available.
