1. What are the contributions in this paper?
The authors present a scenario for a fully distributed computing environment in which computing, storage, and I/O elements are configured on demand into `` virtual systems '' that are optimal for the solution of a particular problem.. The authors also describe present two pilot projects that illustrate some of the elements and issues of this scenario.. The goal of this work is to make the most powerful computing systems those that are logically assembled from network based components, and to make those systems available independent of the geographic location of the constituent elements.. Not only will this allow powerful capabilities will provide a `` just in time '' approach to assembling the resources needed to solve specific to be brought to bear on large problems, it instances of problems in computational simwill also allow access to these capabilities by ulation, data acquisition, data analysis, and a much wider community of people than is archiving.. This environment will be architectures for the solution of specific probenabled through software av. d hardware architecture advances expected over the next ' lems, and then, by using network based reseveral years, including: au order of magnisources, to logically assemble and use the required elements only for the time during which tude increase in workstation I/O and memory they are needed, bandwidth ; the routine incorporation of coprocessors for special tasks ( e. g. video com * This work is supported by the Director, Office of pression ) ; the emerging collaboration between Ener_j Research.. Be easily configured into cooperating elements i | i that form virtual systems ; easy access to masA Case Study sive unique data archives enabled through adThe Lawrence Berkeley Laboratory ( LBL ) vances in data management and mass storage Information and Computing Sciences and Re, systems, and ; user interface paradigms evolved search Medicine Divisions have collaborated to allow non-computer specialists to easily aswith the Pittsburgh Supercomputer Center semble the above dements into effective tools ( PSC ) to demonstrate the possibilities for such to attack scientific problems.. In the following sections, the authors describe two application is the interactive visualization of pilot projects that illustrate some of the elelarge 3D scalar fields ( voxel data sets ) by using ments and issues of this new computing model, a combination of heterogeneous supercomputers and low cost workstations for display and RESEARCH IMAGING control.. The authors believe, however, that the tive '' here is taken to mean the ability to gen.
read more