Power Model And Power-efficient Query Execution Plan

With the popularity of network applications and the development ofinformation collection, the ability to generate and obtain data was improved rapidly,which makes the massive amounts of data almost were distributed everywhere. Inorder to support the massive data management and computing effectively, largescale and ultra large scale data centers was brought out. However, with thedevelopment of data center technology, the energy consumption of data centers wasbecoming very big, the proportion of energy costs in data center operational costrise gradually. Therefore, the energy consumption in data centers has become aproblem cannot be ignored, and the energy management technology is becomingmore and more important, which has become an important metrics and designfactors of data center.How to predict the power consumed by the system accurately is the basis foreffective energy management, and predicting system power was mainly completedby building power model. However, the existing power models were only concernedfor single-core CPU, but didn’t consider the multi-core CPU. However, with thedevelopment of hardware technology, multi-core CPU servers are becoming morepopularly. Therefore, in this paper, we first proposed the multi-core CPU powermodel based on multi-core CPU. Secondly, we proposed a new power model ofmemory by considering the CPU utilization. Finally, with the proposed powermodel of multi-core CPU, memory and disk, a new model of whole machine wasbuilt. Experiments in real machines show that our model is more precise inpredicting the power consumption of the machine, compared to the existing powermodel.DBMS is an important type of software in data center which bears most of thedata center workloads, and also consumes most of the energy in the data center.However, little work has been dated on the power-efficiency of query executionplans generated by the DBMS. Traditional optimal query execution plan does nottake power-efficiency into account. This paper tries to generate a power-efficientquery execution plan by redesigning the query optimizer of the database. We build a power model to estimate the power consumption of a query execution plan andpropose an algorithm to generate the power-efficient query execution plan. Doingexperiments based on the TPC-H benchmark, we evaluate the accuracy of our powermodel and compare the power-efficient query execution plan with the traditionaloptimal query execution plan.