Team AlGorille

Members
Overall Objectives
Scientific Foundations
Application Domains
Software
New Results
Other Grants and Activities
Dissemination
Bibliography

Section: Software

SimGrid

Participants : Pierre-Nicolas Clauss, Fekari El Mehdi, Martin Quinson, Lucas Nussbaum, Cristian Rosa, Christophe Thiéry.

The SimG rid framework aims at being a scientific instrument to the evaluation of algorithmic solutions for large-scale distributed experiments.

The SimG rid framework is the result of a collaboration with Henri Casanova (Univ. of Hawaii, Manoa) and Arnaud Legrand (MESCAL team, INRIA Grenoble-Rhône-Alpes, France). Simulation is a common answer to the grid specific challenges such as scale and heterogeneity. SimG rid is one of the major simulators in the Grid community.

The main strong point of this is its carefully assessed model validity. To this end, the simulation kernel relies on a blend of analytical models and coarse-grain discrete event simulation. It proves several orders of magnitude faster than usual packet-level simulators used in the networking community (such as ns2 or GTNetS) while providing an good level of accuracy [59] .

The SimG rid framework is currently extremely fast. Independent authors demonstrated its superior scalability over its main concurrence [45] , [52] . In addition to the efficiency of the simulation models, this scalability is ensured by a layered architecture, with a simulation kernel computing the time taken by actions which need to consume resources to complete. Another layer of abstraction introduces the notion of processes and network routing between hosts. On top of this come the user interfaces aiming at providing the syntactic sugar easing the tool usage.

Several such user interfaces exist, ensuring the versatility of the SimG rid framework by adapting to the user goal: MSG helps the study of distributed heuristics. This is the historical interface of SimG rid, and remains the most used one. SMPI is a new interface which allows the simulation of MPI programs designed for multi-processor systems on a single computer [2] . SimDag eases the study of scheduling heuristics for DAGs of (parallel) tasks, which helps the work on parallel task scheduling. GRAS (Grid Reality And Simulation) eases the development of Grid services and infrastructures [8] through a specific interface implemented twice: once on top of the simulator for the comfort of development, and once using regular sockets for live deployments.

SimG rid can be freely downloaded SimGrid and its user base is rapidly growing. Over the last decade, it grounded the experimental section of more than eighty scientific publications, only twenty of them being co-authored by members of the development team.


previous
next

Logo Inria