Establishing energy consumption plans for green star-queries in data warehousing systems

Abstract

During the last few years many initiatives were taken away in response to high levels of energy consumption verified in data centers. We all know that this is a critical issue nowadays. Many studies carried out raised a lot of concerns about the energy demands of data centers, discussing solutions to reduce it effectively without affecting their day-by-day operation. In this work we made a small contribution to help that. We studied in a data warehousing system how could be possible to establish an energy consumption plan for a star-query. With these plans, we can establish in each phase of a star-query execution the energy consumed by all the elementary tasks that were executed to satisfy it. With this purpose in mind, and motivated by the usual methods and heuristics used on query execution optimization, we designed and developed a method to estimate the energy consumption of each element (operator) integrated in the execution plan of a query at compile time. With it, we also build up the corresponding energy consumption used on executing the star-query, taking into consideration the characteristics of the computational platforms used for.