This article was accepted into the corpus but its outbound wikilinks were never NER-processed — typical at the deepest BFS hop or when the run's entity cap was reached. No expansion funnel to show.
| Legion (programming system) | |
|---|---|
| Name | Legion |
| Developer | University of Virginia; Argonne National Laboratory |
| Released | 1998 |
| Programming language | C++ |
| Operating system | Linux; Unix |
| License | Proprietary (research) |
Legion (programming system)
Legion is a programming system designed for distributed heterogeneous computing, originating at the University of Virginia with collaborations involving Argonne National Laboratory and other research institutions. It provides a metasystem that unifies disparate resources across clusters, grids, and supercomputers, aiming to simplify application development for large-scale scientific projects such as those at Los Alamos National Laboratory, Lawrence Livermore National Laboratory, and Oak Ridge National Laboratory.
Legion was conceived to enable transparent access to remote resources across environments like TeraGrid, National Science Foundation projects, and computational facilities at NASA Ames Research Center. The project drew on concepts from distributed systems research at Massachusetts Institute of Technology, Stanford University, and Carnegie Mellon University, and influenced middleware efforts such as Globus Toolkit, CORBA, and MPI. Legion's goals were aligned with challenges faced by teams at Sandia National Laboratories, Princeton University, Harvard University, and other institutions involved in computational science.
Legion's architecture uses an object-based metasystem that maps application objects to resources across sites like Argonne National Laboratory and Lawrence Berkeley National Laboratory. Core components include an activation service influenced by work at Bell Labs and a naming system resonant with research from IBM Research and SUN Microsystems. The system integrates with resource managers and schedulers such as those used at European Centre for Medium-Range Weather Forecasts and adapts ideas from projects at CERN and Fermilab. Security and authentication drew on practices from National Institute of Standards and Technology and certificate models used in U.S. Department of Energy centers.
Legion exposes an API modeled in C++ that supports object abstractions reminiscent of Common Object Request Broker Architecture and influenced by language work at Bell Labs and AT&T Labs. The programming model supports distributed objects, remote procedure invocation, and data placement policies that echo projects at Oak Ridge National Laboratory and Los Alamos National Laboratory. Application teams from University of Illinois Urbana-Champaign, University of California, Berkeley, and Cornell University explored Legion APIs for scientific workflows, coupling tools similar to those used with ANSYS, MATLAB, and ParaView.
The implementation is primarily in C++ and targeted Unix/Linux platforms common at National Center for Supercomputing Applications and supercomputing centers. The runtime manages object activation, placement, and migration using distributed protocols informed by research at MIT Lincoln Laboratory and SRI International. Integration efforts paralleled middleware such as Globus Toolkit and resource managers like those at National Computational Science Alliance. Developers tested Legion on clusters at Princeton Plasma Physics Laboratory and grids associated with European Grid Infrastructure.
Legion was evaluated for scalability on systems comparable to those at TeraGrid, XSEDE, and national labs including Argonne National Laboratory and Lawrence Livermore National Laboratory. Performance studies compared Legion's overhead to implementations of MPI, PVM, and RPC systems researched at Stanford University and University of California, San Diego. Workload characterization drew on benchmarks and applications from NASA, NOAA, and scientific collaborations involving Columbia University and Yale University.
Researchers applied Legion to multiphysics simulations used by teams at Sandia National Laboratories, data analysis pipelines at CERN, and environmental modeling coordinated with NOAA and USGS. Collaborative projects with faculty from Duke University, University of Michigan, and University of Washington explored Legion for workflow orchestration in bioinformatics and climate science, integrating tools like R and domain codes used at Argonne National Laboratory.
The Legion project began in the late 1990s at University of Virginia with funding and collaboration from agencies including National Science Foundation and the U.S. Department of Energy. Influences included distributed computing work at MIT, UC Berkeley, and middleware initiatives like Globus Toolkit. The project progressed through deployments and demonstrations at laboratories such as Los Alamos National Laboratory, Lawrence Livermore National Laboratory, and Argonne National Laboratory before tapering as newer paradigms and platforms emerged at institutions like Microsoft Research and Google Research.
Critics noted that Legion's complexity and tight coupling to C++ made adoption harder compared with simpler interfaces like MPI and services from Amazon Web Services and cloud platforms pioneered by Google and Microsoft Azure. Portability concerns arose relative to containerization trends from Docker, Inc. and orchestration approaches exemplified by Kubernetes. Scaling to modern exascale efforts coordinated by DOE programs required different trade-offs than those Legion originally assumed, as seen in later projects at Oak Ridge National Laboratory and Argonne National Laboratory.
Category:Distributed computing systems