ENERGY EFFICIENT RESOURCE ALLOCATION IN DATA CENTERS

The Hong Kong University of Science and Technology
Department of Computer Science and Engineering


PhD Thesis Defence


Title: "ENERGY EFFICIENT RESOURCE ALLOCATION IN DATA CENTERS"

By

Miss Xiangming DAI


Abstract

Recent years have witnessed a tremendous increase in the popularity of 
cloud computing services in supporting business, communication, online 
customer service and helping make life more productive and efficient. 
Naturally, this has been accompanied by a constant expansion of data 
centers in scale and in geographical outreach worldwide, resulting in a 
dramatic growth of the energy consumed to power such data centers. Several 
studies indicate that the energy consumed today by data centers is 
equivalent to the annual output of 34 large (500-megawatt) coal-red power 
plants in the US alone.This number is forecast to reach double in less 
than 10 years. The huge amount of energy consumption not only costs data 
center providers billions of dollars, but also generate hundreds of 
millions of tonnes of carbon pollution per year. Energy consumption in 
data centers comes from several aspects: i) computing and networking 
equipments, ii) cooling equipments, and iii) power draw and other 
ancillary equipments. Any reduction of such consumption is seen as such a 
boon that some data center providers such as Facebook and Google have 
built data centers in as far flung area as the Arctic circle, while others 
like Microsoft are considering undersea data centers, to cut cooling 
costs. In this thesis, we consider several important problems of resource 
allocation in data center while optimizing the energy consumed by 
computing and networking equipments.

The thesis is structured in three parts. The first part falls within the 
area of the socalled platform-as-a-service (PaaS) cloud service model and 
deals with job scheduling in the MapReduce massive-data 
parallel-processing framework. In this part we consider energy efficiency 
as a by-product of minimizing the jobs . More specifically, we first 
propose a new scheduling algorithm called Multiple Queue Scheduler (MQS) 
to improve the data locality rate of tasks as a means to curbing the 
costly data migration delays. Then, to take into account the intricate 
details of MapReduce framework such as the early shuffle problem, we 
propose the Dynamic Priority Multiple Queue Scheduler (DPMQS) to further 
improve MQS. DPMQS dynamically increases the priority of jobs that are 
close to completing their Map phase to speed up the start of the reduce 
phase, thus reducing further the expected job holding time and thus the 
makespan. We implemented both algorithms in Hadoop and compared their 
performance to other existing algorithms. The second part falls within the 
realm of infrastructure-as-a-service (IaaS) and deals with energy 
efficient virtual machine (VM) scheduling in data centers. In this part we 
formulate the minimum energy VM scheduling as a non-convex optimization 
model, prove its NP-hardness, then explore two greedy approximation 
algorithms, minimum energy VM scheduling algorithm (MinES) and minimum 
communication VM scheduling algorithm (MinCS), to reduce the energy while 
satisfying the data center tenants service level agreements. Current IaaS 
service providers only support rudimentary network topologies that simply 
consist of a super-sized virtual switch interconnecting a tenant's VMs. To 
increase the business potential of such platforms by supporting more 
intricate network topologies as specified by the tenants, we consider in 
the third part of this thesis the problem of embedding virtual clusters 
into a data center in a energy efficient manner. We carefully provide a 
mathematical optimization model of this problem, then given its 
NP-hardness, we propose an approximate algorithm minimum energy virtual 
cluster embedding (MinE-VCE) to solve the problem. We tested all proposed 
algorithms using real data traces as well as synthetic ones to demonstrate 
their performance.


Date:			Thursday, 2 June 2016

Time:			2:00pm – 4:00pm

Venue:			Room 5501
 			Lifts 25/26

Chairman:		Prof. J.S. Kuang (CIVL)

Committee Members:	Prof. Brahim Bensaou (Supervisor)
 			Prof. Kai Chen
 			Prof. Jogesh Muppala
 			Prof. Chin Tau Lea (ECE)
                         Prof. Samee Khan (Elec. & Comp. Engg.,
 					  North Dakota State Univ.)


**** ALL are Welcome ****