Adaptive Scheduling for Task Farming with Grid Middleware
Scheduling in metacomputing environments is an active field of research as the vision of a Computational Grid becomes more concrete. An important class of Grid applications are long-running parallel computations with large numbers of somewhat independent
- PDF / 237,706 Bytes
- 14 Pages / 431 x 666 pts Page_size
- 90 Downloads / 202 Views
Department of Computer Science and Engineering, University of California at San Diego, La Jolla, CA 92093-0114, USA 2 School of Computing, SoongSil University, Seoul, 156-743 Korea 3 Department of Computer Science, University of Tennessee, Knoxville, TN 37996-1301, USA 4 Mathematical Science Section, Oak Ridge National Laboratory, Oak Ridge, TN 37831, USA
Abstract. Scheduling in metacomputing environments is an active field of research as the vision of a Computational Grid becomes more concrete. An important class of Grid applications are long-running parallel computations with large numbers of somewhat independent tasks (Monte-Carlo simulations, parameter-space searches, etc.). A number of Grid middleware projects are available to implement such applications but scheduling strategies are still open research issues. This is mainly due to the diversity of both Grid resource types and of their availability patterns. The purpose of this work is to develop and validate a general adaptive scheduling algorithm for task farming applications along with a user interface that makes the algorithm accessible to domain scientists. Our algorithm is general in that it is not tailored to a particular Grid middleware and that it requires very few assumptions concerning the nature of the resources. Our first testbed is NetSolve as it allows quick and easy development of the algorithm by isolating the developer from issues such as process control, I/O, remote software access, or fault-tolerance. Keywords: Farming, Master-Slave Parallelism, Scheduling, Metacomputing, Grid Computing.
1
Introduction
The concept of a Computational Grid envisioned in [1] has emerged to capture the vision of a network computing system that provides broad access not only to massive information resources, but to massive computational resources as well. Such computational grids will use high-performance network technology to connect hardware, software, instruments, databases, and people into a seamless web that supports a new generation of computation-rich problem solving environments for scientists and engineers. Grid resources will be ubiquitous thereby justifying the analogy to the Power Grid. Those features have generated interest among many domain scientists and new classes of applications arise as being potentially griddable. Grid resources P. Amestoy et al. (Eds.): Euro-Par’99, LNCS 1685, pp. 30–43, 1999. c Springer-Verlag Berlin Heidelberg 1999
Adaptive Scheduling for Task Farming with Grid Middleware
31
and their access policies are inherently very diverse, ranging from directly accessible single workstations to clusters of workstations managed by Condor [2], or MPP systems with batch queuing management. Furthermore, the availability of these resources changes dynamically in a way that is close to unpredictable. Lastly, predicting networking behavior on the grid is an active but still open research area. Scheduling applications in such a chaotic environment according the end-users’ need for fast response-time is not an easy task. The concept
Data Loading...
