Massively Parallel, Adaptive and Fault-Tolerant Execution of Data Flow Programs on Dynamic Clouds

平流层研究单位B项目的目标是研究和实施一个大规模并行、自适应和容错的可重新配置云环境执行框架。作为整个研究小组的底层，该项目将解决如何将现代信息管理应用程序的需求与当今云系统提供的可扩展性和计算能力结合在一起的挑战。特别是，该项目将研究虚拟环境中数据局部性的拓扑感知、这种环境中数据流程序的高效和自适应执行，以及在出现永久性或暂时性节点故障时确保容错的方法。所有研究成果将与研究小组的前期工作结合起来，通过实验进行评估，并将促进平流层项目的更高水平的研究工作。

The goal of Project B within the Stratosphere Research Unit is to research and to implement a massivelyparallel, adaptive and fault-tolerant execution framework for re-configurable Cloud environments. Forming the bottom layer of the entire research group, the project will address the challenge of how to bring together the demands of modern information management applications with the scalability and computational power offered by today’s Cloud systems. In particular, the project will conduct research on topology awareness for data locality in virtualized environments, the efficient and adaptive execution of data flow programs in such environments as well as methods to ensure fault tolerance in presence of permanent or transient node failures. All research results will be integrated with preliminary work of the research group, evaluated experimentally and will facilitate higher level research efforts of the Stratosphere project.