Previsão do tempo de resposta de aplicações de big data em ambientes de nuvem

• Autor(a) principal: Túlio B. M. Pinto
• Data de Publicação: 2018
• Outros Autores: Ana Paula Couto da Silva, Jussara M. Almeida
• Tipo de documento: Artigo de conferência
• Idioma: por
• Título da fonte: Repositório Institucional da UFMG
• Texto Completo: https://doi.org/10.5753/sbrc.2018.2440; http://hdl.handle.net/1843/74090; https://orcid.org/0000-0001-5951-3562; https://orcid.org/0000-0001-9142-2919
• topic: Armazenamento; Big data; Computação em nuvem

Data access heterogeneity and irregularity are typical properties of Big Data applications, and, therefore, turn hardware and software resource scheduling much more challenging. However, the flexibility and elasticity provided by cloud environments decrease the difficulty by allowing on-demand resource provisioning. Nonetheless, the performance prediction (e.g.: response time) of such applications increase in complexity as all these characteristics are combined. This work explores an analytical model for Spark applications’ response time prediction, a popular platform for large-scale data processing, parametrized by earlier execution logs. This model is evaluated in several scenarios and applications. The results show relative errors lower than 8% for response time prediction, in average.