AIMES - An Integrated Middleware Framework to Enable Extreme Collaborative Science
Large-scale scientific applications are usually distributed across multiple diversified resources in terms of compute, storage and network services. They require that the usage of underlying infrastructures be both convinient and efficient. However, a structured and standard approach to addressing this need is not routinely available. This leads to many ad-hoc, repeated and non-extensible solutions that are effort wasting and lag the process of scientific research. This project aims to bridge the gap between application requirements and diverse and heterogeneous platforms, by developing a middleware framework that can support the needs of tools and services in support of distributed scientiﬁc collaborative applications at extreme scales.
This project itself is also a collaborative one participated by researchers from Rutgers University, University of Minnesota and University of Chicago. Our contribution to this project is a system called Bundle.
The design goal of Bundle is to high-level abstraction of resources. Bundle make multiple platforms such as batch-controlled supercomputers, grid and cloud appear to be a unified allocable resource pool. It also enables automatic, on-demand selection of appropriate resource that meets performance requirements.
Click to view larger image
- Students: Francis Liu
- PI: Jon Weissman