SimGrid

SimGrid

Initial release	1998 (1998) [1]
Stable release	3.36 / September 9, 2024;5 months ago (2024-09-09)
Repository	https://framagit.org/simgrid/simgrid
Written in	Core: C++; Bindings: Python.
Platform	Unix, Mac OS X, Microsoft Windows
Type	Distributed system simulator, Network simulator, Model checking
License	GNU Lesser General Public License [2]
Website	simgrid.org

SimGrid is a framework for evaluating algorithms and heuristics in distributed systems using simulation. ^[3] It provides tools for prototyping, evaluating, and comparing different platform configurations, system designs, and algorithms. By simulating distributed computing environments, SimGrid provides researchers and developers with a way to test and analyze system behaviors.

SimGrid facilitates the study of distributed systems without the need for extensive physical infrastructure. ^[4] This allows developers to experiment with various scenarios and configurations in a virtual environment.

APIs

The framework supports multiple programming paradigms through its APIs. This makes it suitable for examining scheduling strategies, resource allocation policies, fault tolerance mechanisms, and other aspects of distributed computing. SimGrid is used in research and applications in various fields such as cloud computing, high-performance computing (HPC), grid computing, and peer-to-peer networking. It can be used to identify bottlenecks and assess system efficiency, which is useful for performance analysis and optimization.

There are four SimGrid user interfaces:

• SimDag: This allows prototyping and simulation of scheduling heuristics for applications providing task graphs. Users can create tasks, add dependencies between tasks, retrieve information about the platform, schedule tasks for execution on particular resources, and compute the DAG's execution time. ^[3]
• MSG: Added in SimGrid v2, this allows study of CSP applications. Initially intended for studying scheduling algorithms, it may be used in other contexts as well. ^[3]
• Grid Reality And Simulation (GRAS): This enables development of distributed applications within the simulator, which can then be deployed on real-world platforms without requiring code modifications. ^[5]
• SMPI: This allows simulation of unmodified MPI applications by intercepting MPI primitives.

See also

• BIGSIM

References

1. Casanova, Henri (May 2001). "A Toolkit for the Simulation of Application Scheduling". First IEEE International Symposium on Cluster Computing and the Grid (CCGrid'01). Brisbane, Australia. pp. 430–441. doi:10.1109/CCGRID.2001.923223.
2. "Official SimGrid Page" . Retrieved October 17, 2024.
3. Casanova, Henri; Legrand, Arnaud; Quinson, Martin (March 4, 2008). "SimGrid: a Generic Framework for Large-Scale Distributed Experiments". Inria. IEEE – via HAL Open Science.
4. "Lowering entry barriers to developing custom simulators of distributed applications and platforms with SimGrid".
5. Quinson, Martin (November 2006). "GRAS: a Research and Development Framework for Grid and P2P Infrastructures". Inria– via HAL Open Science.

External links

• SimGrid – official project homepage
• Reference publication: H. Casanova, A. Giersch, A. Legrand, M. Quinson, and F. Suter, “Versatile, Scalable, and Accurate Simulation of Distributed Applications and Platforms,” Journal of Parallel and Distributed Computing, vol. 74, no. 10, pp. 2899 – 2917, 2014.