Abstract:
Cloud computing is an emerging computing platform with flexible computational
architecture and an enormous collection of heterogeneous systems. Resource
management is an important strategy in enhancing the overall performance of cloud
computing. Currently, effective resource management in cloud computing is
becoming an inevitable demanding topic and it is very challenging to implement
efficient resource management based on QoS demands by optimizing the number of
objectives. In this research, resource management problem can be considered from
three optimization aspects: single objective optimization, multi-objective
optimization and many objective optimizations. We discuss two significant levels in
this emerging resource management paradigm: one is task scheduling, that maps
tasks to VMs and the other is VM placement, which maps VMs to physical servers.
The nature inspired algorithm approaches can obtain feasible solutions than others.
Firstly, this research explores the research in the area of resource management in
cloud computing to gain an understanding of related work. Secondly, this research
proposes single-objective efficient resource management and task scheduling
methods in cloud platforms to minimize response time, makespan and data center
processing time. Another main target of the thesis is to research multi-objective
energy aware, efficiency aware and QoS aware resource management as VM
placement mechanism for simultaneously optimizing three objectives. Most of the
previous research methods, either single objective or multi-objective, cannot
provide suitable solutions when the number of the objectives exceeds three.
Therefore, there is a tremendous demand to use the many-objective optimization
mechanism to deal with the resource-management problem. Finally, this research
proposes a many-objective approach for VM placement to attain equilibrium among
five objectives. In summary, we propose energy aware, efficiency aware, cost aware,
QoS aware load balancing and utilization aware resource scheduling approaches in
cloud computing using nature-inspired algorithms. The performance evaluation
from three optimization aspects, i.e., single-objective, multi-objective and manyobjective optimizations, demonstrate that the proposed approaches are capable of
enhancing the state-of-the-art techniques in the environment of enormous cloud
data centers by optimizing different objective functions.
