Development of new capabilities for the computer grid simulator iSPD: characterizing tasks and optimizing the simulation engine

• Grant number: 12/15127-3
• Support Opportunities: Regular Research Grants
• Duration: November 01, 2012 - October 31, 2014
• Field of knowledge: Physical Sciences and Mathematics - Computer Science - Computer Systems
• Principal Investigator: Aleardo Manacero Junior
• Grantee: Aleardo Manacero Junior
• Host Institution: Instituto de Biociências, Letras e Ciências Exatas (IBILCE). Universidade Estadual Paulista (UNESP). Campus de São José do Rio Preto. São José do Rio Preto , SP, Brazil
• Associated researchers: Renata Spolon Lobato

Abstract

The use of computer grids as infrastructure to achieve high performance computing has grown in recent years, both for scientific and business applications. In parallel with this growth comes the offer of performance evaluation tools oriented to grids, with a special note to grid simulators. One of such simulators is iSPD, {\bf i}conic {\bf S}imulator of {\bf P}arallel and {\bf D}istributed systems \cite{anss2012, iscc2012}, proposed and developed by Unesp's Parallel and Distributed Systems Lab, GSPD, which was partially funded by FAPESP (grant 08/09312-7). iSPD's primary goal is to provide a grid modeling interface that is easy-to-use by any kind of user, even those who have very little knowledge about computer programming or simulation tools. Its initial version attends this requisite in a large extension, but its simulation engine is somewhat slow while simulating systems with a large amount of elements (above fifty thousand grid nodes and one hundred thousand simulated tasks). Therefore, one of the goals of this research proposal is to find efficient solutions to speed up the simulation process in iSPD. This acceleration should come from two fronts: parallelizing the simulation through threads (what is not thoroughfully done by similar simulators) and identifyng algorithms for queue manipulation that have linear complexity. Besides this goal, other innovative aspects of this proposal include research to provide different forms of task manipulation, which occurs just as atomic elements in the current version. Particularly, two aspects of task's manipulation will be considered: their semantic distinction (allowing to treat them as services offered by the grid) and their interdependence (allowing for a more realistic modeling of concurrency and parallelism). (AU)