Loading...

Modeling and Evaluation of Performability in Cloud Computing Considering Power Consumption

Ataie Dadavi, Ehsan | 2017

387 Viewed
  1. Type of Document: Ph.D. Dissertation
  2. Language: Farsi
  3. Document No: 50414 (19)
  4. University: Sharif University of Technology
  5. Department: Computer Engineering
  6. Advisor(s): Movaghar, Ali
  7. Abstract:
  8. Considering current problems in performance evaluation of the Cloud and its management policies and strategies, this thesis presents different approaches for simultaneous evaluation of the performance, dependability, and power consumption of Infrastructure-as-a-Service (IaaS) clouds. The proposed approaches use Stochastic Reward Nets (SRNs) and Stochastic Activity Networks (SANs) to model and evaluate performability and power consumption of cloud systems. In all of the proposed approaches, the cloud resource management system is power-aware. It means that idle Physical Machines (PMs), network equipment, and supporting subsystems are kept in powered-off or standby modes, as much as possible, to save power. Furthermore, the approaches take different features of real-world IaaS clouds into account including different pools of PMs each one containing different power consumption, failure/repair behavior, and provisioning delay. As other important features of the proposed models in this thesis we can mention modeling failure and repair of PMs and Virtual Machines (VMs), considering separate provisioning and servicing steps of VMs, and having support for VM multiplexing. The first approach investigates the scalability of the models proposed for evaluating cloud systems. To achieve this, a monolithic model for a cloud data center, which contains two hierarchical levels is presented. In order to cope with the scalability problem of the monolithic model, two approximate models are presented that improve the scalability without showing any significant loss of accuracy. In the second approach, we study the method of dispatching incoming IaaS requests to the racks of a data center. To this end, several power-aware and structure-aware resource allocation policies are introduced to efficiently dispatch user requests between the underlying racks. The third approach considers self-adaptation of the cloud resource management system when workload bursts occur. To this end, a self-adaptive power-aware and Service Level Agreement (SLA)-aware resource management system is presented that adapts the employed resources according to the rate of incoming requests. In all of the approaches, different performance, dependability, and power consumption measures are introduced and evaluated
  9. Keywords:
  10. Cloud Computing ; Virtualization ; Performance ; Dependability ; Power Consumption

 Digital Object List

 Bookmark

No TOC