Loading...

Analysis and Optimization of Architectural Quality Requirements for Federated Clouds

Pourreza, Maryam | 2016

637 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 48711 (19)
  4. University: Sharif University of Technology
  5. Department: Computer Engineering
  6. Advisor(s): Habibi, Jafar
  7. Abstract:
  8. Proposing a comprehensive architecture for cloud federation with the aim of satisfying stakeholders the most, is one of the most challenging issues in this context. For an architecture to be comprehensive, requirements should be elicited thoroughly in early stages of architecture design. Since covering all requirements is a very complicated task, prioritization is considered to be a vital step which can help architects to focus on more important requirements in their design as well as to evaluate existing architectures. In this research, architecturally signifcant quality requirements of cloud federation for aggregated cloud architectures are studied and categorized. Moreover, elicited requirements are also prioritized in this research with the help of a number of experts through a predetermined methodology. According to the prioritization results, performance, availability, and security are the three most important requirement categories for federated clouds. In the next step of this research and with the aim of satisfying the elicited requirements in previous step, the data center selection problem for federated clouds is analyzed. Finding an optimized solution for this problem in a federated environment can help all cloud providers to have an increase in their customers’ satisfaction. Therefore, in this research the data center selection problem based on cloud customers’ preferences of quality attributes is frst introduced as an integer programming. This problem and its different variations are also analyzed computationally. Then, fve algorithms for fnding suboptimal solutions for the proposed problem are introduced. Evaluation of these algorithms indicates the superiority of the Ant Colony System algorithm compared to the other presented algorithms in the sense of the overall utility as well as the running time. Furthermore, this algorithm provides at least 1.232 higher utility when compared with previous solutions in this context. It should also be mentioned that previous solutions for the data center selection problem have the potentianlity for the capacity violation and consequently the violation of service level agreements in a federated environment with small cloud providers
  9. Keywords:
  10. Cloud Computing ; Software Architecture ; Optimization ; Federated Cloud ; Architectural Requirements ; Data Center Selection

 Digital Object List

 Bookmark

...see more