[Zurück]

M. Halper:
"Classification and Monitoring of Incidents in Cloud-based Big Data Analytics";
Betreuer/in(nen): H. Truong; Institute of Information Systems Engineering, Distributed Systems Group, 2018; Abschlussprüfung: 26.02.2018.

Cloud Computing delivers the resources for highly distributed and resource intensive Big Data applications that analyses the massive amount of data produced by multimedia, social media, operation of enterprises, trading of enterprises and scientific research. In the data transformation process various stakeholders add their knowledge and expertise. The problem is that each of them has its own methodologies and routines for incident management but none of them sees the process as a whole. A failure of single elements in the data transformation process can lead to a domino effect that impacts the whole process. This thesis aims to deliver groundwork regarding an incident management process by applying an end-to-end approach that considers the specific needs of Cloud-based Big Data analytics. We present a stakeholder analysis including roles and actors. We describe the necessary generic use cases that need to be implemented. An incident survey details the characteristics of incidents of Cloud-based Big Data analytics. We develop a classification of cloud hosted Big Data analytic incidents from the survey. Metrics are added to describe the important elements in the data transformation process. We introduce a generic architecture of an incident management system that can be adapted to the specific needs of Cloud-based Big Data analytics applications. By delivering common building blocks like stakeholders, use cases, classifications and software component diagrams, the essentials of existing architectures are captured and put into the perspective of the Cloud-based Big Data analytics application. We deliver a prototype for the evaluation of the architecture and we illustrate with a test scenario the viability of the generic architecture.