Uriarte, Rafael Brundo (2015) Supporting Autonomic Management of Clouds: Service-Level-Agreement, Cloud Monitoring and Similarity Learning. Advisor: De Nicola, Prof. Rocco. Coadvisor: Tiezzi, Prof. Francesco . pp. 208. [IMT PhD Thesis]
RafaelBrundoUriarte_Thesis_Final_A5.pdf - Published Version
Available under License Creative Commons Attribution Non-commercial.
Download (4MB) | Preview
Cloud computing has grown rapidly during the past few years and has become a fundamental paradigm in the Information Technology (IT) area. Clouds enable dynamic, scalable and rapid provision of services through a computer network, usually the Internet. However, managing and optimising clouds and their services in the presence of dynamism and heterogeneity is one of the major challenges faced by industry and academia. A prominent solution is resorting to selfmanagement as fostered by autonomic computing. Self-management requires knowledge about the system and the environment to enact the self-* properties. Nevertheless, the characteristics of cloud, such as large-scale and dynamism, hinder the knowledge discovery process. Moreover, cloud systems abstract the complexity of the infrastructure underlying the provided services to their customers, which obfuscates several details of the provided services and, thus, obstructs the effectiveness of autonomic managers. While a large body of work has been devoted to decisionmaking and autonomic management in the cloud domain, there is still a lack of adequate solutions for the provision of knowledge to these processes. In view of the lack of comprehensive solutions for the provision of knowledge to the autonomic management of clouds, we propose a theoretical and practical framework which addresses three major aspects of this process: (i) the definition of services’ provision through the specification of a formal language to define Service-Level-Agreements for the cloud domain; (ii) the collection and processing of information through an extensible knowledge discovery architecture to monitor autonomic clouds with support to the knowledge discovery process; and (iii) the knowledge discovery through a machine learning methodology to calculate the similarity among services, which can be employed for different purposes, e.g. service scheduling and anomalous behaviour detection. Finally, in a case study, we integrate the proposed solutions and show the benefits of this integration in a hybrid cloud test-bed.
|Item Type:||IMT PhD Thesis|
|Subjects:||Q Science > QA Mathematics > QA75 Electronic computers. Computer science|
|PhD Course:||Computer Science and Engineering|
|Date Deposited:||10 Apr 2015 08:30|
Actions (login required)