Date of Award


Document Type


Degree Name

Doctor of Philosophy (PhD)


Electrical and Computer Engineering (Holcomb Dept. of)

Committee Member

Associate Professor Haiying (Helen) Shen, Committee Chair

Committee Member

Professor Adam Hoover

Committee Member

Associate Professor Walter B. Ligon, III

Committee Member

Professor Amy W. Apon


Datacenter clouds (e.g., Microsoft's Azure, Google's App Engine, and Amazon's EC2) are emerging as a popular infrastructure for computing and storage due to their high scalability and elasticity. More and more companies and organizations shift their services (e.g., online social networks, Dropbox file hosting) to clouds to avoid large capital expenditures. Cloud systems employ virtualization technology to provide resources in physical machines (PMs) in the form of virtual machines (VMs). Users create VMs deployed on the cloud and each VM consumes resources (e.g., CPU, memory and bandwidth) from its host PM. Cloud providers supply services by signing Service Level Agreement (SLA) with cloud customers that serves as both the blueprint and the warranty for cloud computing. Under-provisioning of resources leads to SLA violations while over-provisioning of resources leads to resource underutilization and then revenue decrease for the cloud providers. Thus, a formidable challenge is effective management of virtual resource to maximize energy efficiency and resource utilization while satisfying the SLA.

This proposal is devoted to tackle this challenge by addressing three fundamental and essential issues: i) initial VM allocation, ii) VM migration for load balance, and iii) proactive VM migration for long-term load balance. Accordingly, this proposal consists of three innovative components:

(1) Initial Complementary VM Consolidation.

Previous resource provisioning strategies either allocate physical resources to virtual machines (VMs) based on static VM resource demands or dynamically handle the variations in VM resource requirements through live VM migrations. However, the former fail to maximize energy efficiency and resource utilization while the latter produce high migration overhead. To handle these problems, we propose an initial VM allocation mechanism that consolidates complementary VMs with spatial/temporal-awareness. Complementary VMs are the VMs whose total demand of each resource dimension (in the spatial space) nearly reaches their host's capacity during VM lifetime period (in the temporal space). Based on our observation of the existence of VM resource utilization patterns, the mechanism predicts the lifetime resource utilization patterns of short-term VMs or periodical resource utilization patterns of long-term VMs. Based on the predicted patterns, it coordinates the requirements of different resources and consolidates complementary VMs in the same physical machine (PM). This mechanism reduces the number of PMs needed to provide VM service hence increases energy efficiency and resource utilization and also reduces the number of VM migrations and SLA violations.

(2) Resource Intensity Aware VM Migration for Load Balance.

The unique features of clouds pose formidable challenges to achieving effective and efficient load balancing. First, VMs in clouds use different resources (e.g., CPU, bandwidth, memory) to serve a variety of services (e.g., high performance computing, web services, file services), resulting in different overutilized resources in different PMs. Also, the overutilized resources in a PM may vary over time due to the time-varying heterogenous service requests. Second, there is intensive network communication between VMs. However, previous load balancing methods statically assign equal or predefined weights to different resources, which leads to degraded performance in terms of speed and cost to achieve load balance. Also, they do not strive to minimize the VM communications between PMs. This proposed mechanism dynamically assigns different weights to different resources according to their usage intensity in the PM, which significantly reduces the time and cost to achieve load balance and avoids future load imbalance. It also tries to keep frequently communicating VMs in the same PM to reduce bandwidth cost, and migrate VMs to PMs with minimum VM performance degradation.

(3) Proactive VM Migration for Long-Term Load Balance.

Previous reactive load balancing algorithms migrate VMs upon the occurrence of load imbalance, while previous proactive load balancing algorithms predict PM overload to conduct VM migration. However, both methods cannot maintain long-term load balance and produce high overhead and delay due to migration VM selection and destination PM selection. To overcome these problems, we propose a proactive Markov Decision Process (MDP)-based load balancing algorithm. We handle the challenges of allying MDP in virtual resource management in cloud datacenters, which allows a PM to proactively find an optimal action to transit to a lightly loaded state that will maintain for a longer period of time. We also apply the MDP to determine destination PMs to achieve long-term PM load balance state. Our algorithm reduces the numbers of SLA violations by long-term load balance maintenance, and also reduces the load balancing overhead (e.g., CPU time, energy) and delay by quickly identifying VMs and destination PMs to migrate.

Finally, we conducted extensive experiments to evaluate the proposed three mechanisms.

i) We conducted simulation experiments based on two real traces and real-world testbed experiments to show that the initial complementary VM consolidation mechanism significantly reduces the number of PMs used, SLA violations and VM migrations of the previous resource provisioning strategies.

ii) We conducted trace-driven simulation and real-world testbed experiments to show that RIAL outperforms other load balancing approaches in regards to the number of VM migrations, VM performance degradation and VM communication cost.

iii) We conducted trace-driven experiments to show that the MDP-based load balancing algorithm outperforms previous reactive and proactive load balancing algorithms in terms of SLA violation, load balancing efficiency and long-term load balance maintenance.