Secure hardware resource monitoring, usage optimization and affirmation for database operations in virtualized cloud environment / Tan Chee Heng

• Main Author: Tan, Chee Heng
• Format: Thesis
• Published: 2014
• Subjects: T Technology (General)
• Online Access: http://studentsrepo.um.edu.my/4610/1/Thesis.pdf; http://studentsrepo.um.edu.my/4610/
• Institution: Universiti Malaya

Hardware resource management is an important topic in Information Technology (IT) industry. This is due to the increasing demand of computing power by ever-evolving applications, especially those which are Service Level Agreement (SLA)-bound. Undeniably, hardware cost has reduced significantly in recent time. However this does not translate into saving in capital and operational costs of businesses as the computing resource requirement from new applications overwhelms the reduction in hardware cost. Hence, cloud computing paradigm evolved from conventional grid and utility computing, to provide for the aggressive computational demands. To better serve the hosting in cloud environments, particularly in industries where data sensitivity and privacy is of major concern, better mechanisms are needed in the resource management arena. The proposed mechanisms in this research avoided access to real data in the database, to meet the objectives of effective hardware resource administration. Here, the hardware resource management in virtualized cloud environment is scrutinized. The topics of interest are in the area of resource utilization monitoring, optimization and affirmation. The proposed mechanisms provide alternatives to conventional methods which are commonly adopted by the wide IT industry today. The target is to provide more simplified approaches to these conventional tools, with faster and more accurate attributes in sight. In resource utilization monitoring area, metadata of the actual data is characterized to yield an understanding of the workload in the database, which then contributes to the decision in planning for hardware provisioning and de-provisioning activities, as well as resource scaling arrangement. Consequently, a mechanism is proposed to serve the resource utilization optimization objective. In this research area, hardware fault and failure analysis are investigated, in iv order to provide an optimal operating environment to database transactions. The analysis on the hardware fault and failure symptoms is performed against the output obtained from the iterative execution of Transaction Processing Performance Council (TPC)-H queries. Baseline is established and parameters’ values obtained from subsequent testing on the same set of queries are compared to baseline’s values to obtain insightful information on the hardware state. Next, the resource utilization affirmation theme deals with the proposition to establish stress-testing scenario in the Virtual Machine (VM). The work here strives to construct an environment in the VM whereby validation on transactions’ response time can be performed at the hypothetical resource constraining point in the VM. It serves this validation purpose in 2 situations: when the VM undergoes hardware change, or during normal operations. Verification is performed by stressing the VM to the resource constraining point using the proposed method; subsequently SLA-bound transactions are sent to the database and their respective response time is examined and compared to expected response time. The proposed mechanism also incorporates technique to determine the resource threshold from database transactions perspective. The resource utilization monitoring utilizes metadata from representative workload, whereas the resource utilization optimization and affirmation mechanisms utilize the hypothetical data and queries from TPC-H benchmark, hence achieving the objective of eluding access to real data. These deliveries focus on the consistency, stability and accuracy attributes.