Year: 2024
Type: conference-paper
Abstract: Hybrid cloud, integrating public and private clouds, presents a promising environment for scientific applications by combining scalability with cost efficiency. However, the complexity of these environments requires tools to support infrastructure planning and optimization prior to actual deployment. This study evaluates two simulators: SimGrid and CloudSim Plus. The focus is on assessing their suitability for simulating the execution of scientific applications within hybrid cloud environments; particularly regarding scalability. Scientific workflows modeled as directed acyclic graph, were used to evaluate a hybrid cloud infrastructure, and both simulators were assessed on workload representation, cloud infrastructure, and scheduling strategies. The results suggest that SimGrid offers greater flexibility in network modeling, while CloudSim Plus excels in resource allocation policy simulation. This comparative analysis aims to help researchers select the most appropriate tool for simulating and optimizing scientific applications and cloud scheduling strategies. Future research should explore the integration of emerging technologies, such as containers and microservices, within these simulators.
Authors: João Antonio Soares, Rafael Burlamaqui Amaral, Gerson Cavalheiro
Institution: Universidade Federal de Pelotas
Language: English
Subfield: Information Systems
Field: Computer Science
Domain: Physical Sciences
Open Access status: gold
OpenAlex ID: W4411183623