Performance and Energy Consumption Aspects in Migrating Big Data Systems to the Cloud

Abstract

Massive data processing areas (big-data) have taken advantage on the availability of cloud computing. Applications processed in virtualized environments have become usual. Since datacenters are responsible for demanding around of 2 % of the global energy, understand the impact of virtualization on big-data applications is necessary, as well its energy consumption. In this work, we evaluated two resources scheduling policies to process big-data in three different sizes of virtual machines. Our experiments showed that execution time (performance) and consumption of a big-data tasks can result in relevant discrepancies even with the same resources budget.