SGISL: a distributed computational environment supporting environmental and ecological research

Abstract

SGISL) at the University of Montana was recentlyformed to support a number of related environmental and ecological research projects.Each of these projects requires access to large information and computational resources, in addition to sharing "modeling expertise" contained in the form of various computerized data transformations and simulations. Historically, the modus operurdi for such projects was for a small group of researchers to work in relative isolation, familar to some extent with the nature of work being conducted by other groups but primarily interested in constructing a stand-alone information system suited specifically to the needs of their own application,With a few small projects this computational environment can be effective in producing application-specific results, but as the size and number of projects increases this decentralized structure ignores the potential benefits that can be achieved with a structure that promotes sharing of information system costs and resourcqs. SGISL combines modern data modeling techniques and networking technology to provide a conceptual form of information system centralization, while allowing the information system components and users to continue to work in the distributed. decentralized environment.Permission to copy without fee all or part of this matarial is granted provided that the copias ara not made or diatributad for diract commercial advantage, tha ACM copyright notica and tha title of the publication and its date appaar, and notica is given that copying is by parrnission of the Association for Computing Machinary. To copy otharwisa, or to republish, raquiras a fea and/or apacific permission. 01992ACM 089791 -472-4/92/0002/0165 $1.50

Using the Linda distribution paradigm to address a variety of computational balancing concerns

“…The Scientific Geographic Information System Laboratory (SGISL) Project [5] was formed to support the computational and software development needs of several environmental and ecological research projects, each involved in experimentation with large scale (i.e., at the regional level) ecosystem simulation models (e.g., [8,12,13,14,15]). The ecosystem simulation models typically require access to large data resources, execution of potentially lengthy computations, and utilization of very speciahzed input/output interfaces.…”