Despachador de Simulações Tolerante a Falhas
Abstract
Scientific applications often require a lot of computational resources to achieve their results. Understanding and proving scientific results implies the performance of appropriate statistical analyzes. For this, researchers commonly need to perform several simulations by correctly analyzing the differences obtained in each replication. In some scenarios, this makes the conduct of research an impractical process. The automation of this process implies saving the time spent both in configuration and in possible human errors in data analysis. As a way to overcome such challenges, this work proposes an automation system for the dispatch of simulations with statistical analysis of the results. In addition, the proposed system provides distributed computing resources and assists the organization of tasks in a distributed system, with load balancing and fault tolerance. Examples of running the dispatcher are presented through an intuitive interface for system management and the display and export of results.
References
MongoDB (2018). Mongodb, Inc. https://www.mongodb.com/.
NodeJS (2018). Node.js Foundation. https://nodejs.org/en/.
