A distributed computing environment with support for constraint-based task scheduling and scientific experimentation
- Univ. of Washington, Seattle, WA (United States). Dept. of Computer Science and Engineering
This paper describes a computing environment which supports computer-based scientific research work. Key features include support for automatic distributed scheduling and execution and computer-based scientific experimentation. A new flexible and extensible scheduling technique that is responsive to a user`s scheduling constraints, such as the ordering of program results and the specification of task assignments and processor utilization levels, is presented. An easy-to-use constraint language for specifying scheduling constraints, based on the relational database query language SQL, is described along with a search-based algorithm for fulfilling these constraints. A set of performance studies show that the environment can schedule and execute program graphs on a network of workstations as the user requests. A method for automatically generating computer-based scientific experiments is described. Experiments provide a concise method of specifying a large collection of parameterized program executions. The environment achieved significant speedups when executing experiments; for a large collection of scientific experiments an average speedup of 3.4 on an average of 5.5 scheduled processors was obtained.
- Research Organization:
- Los Alamos National Lab. (LANL), Los Alamos, NM (United States)
- Sponsoring Organization:
- USDOE, Washington, DC (United States)
- DOE Contract Number:
- W-7405-ENG-36
- OSTI ID:
- 505354
- Report Number(s):
- LA-UR-97-473; CONF-970835-1; ON: DE97004798; TRN: AHC29716%%106
- Resource Relation:
- Conference: 6. IEEE international symposium on high peformance distributed computing, Portland, OR (United States), 5-8 Aug 1997; Other Information: PBD: [1997]
- Country of Publication:
- United States
- Language:
- English
Similar Records
Center for Technology for Advanced Scientific Componet Software (TASCS)
Parallel processing of filtered queries in attributed semantic graphs