Doctor's Theses (authored and supervised):
"Cloud-Based Elasticity for Business Processes and Data Storage";
Supervisor, Reviewer: S. Schulte, J. Spillner, I. Weber;
Institute of Information Systems Engineering, Distributed Systems Group,
oral examination: 2020-12-11.
Over the last decade, the concept of cloud computing found widespread adoption as a way to provide software services to customers. Elasticity, i.e., the possibility to rapidly adapt the amount of utilized computational resources to changing workloads, is often named as one of the major reasons for using cloud computing. Specialized resource provisioning approaches help to improve elasticity. This thesis focuses on the resource-efficient execution of business processes on elastic cloud-based computational resources and the redundant and cost-efficient storage of data on elastic cloud storage services.
Nowadays, business processes and Business Process Management Systems are an integral part of almost every organization. Business Process Management Systems are often relying on heavyweight Virtual Machines as cloud-based execution environment or are not utilizing elastic cloud resources at all, leading to limited elasticity. In this thesis, we propose a novel approach by using lightweight containers as cloud-based execution environment, which decreases the required resource consumption and increases the elasticity in comparison to Virtual Machines. We develop an elastic Business Process Management System
and an accompanying resource provisioning approach that optimizes the
execution of concurrent business processes for resource efficiency while processspecific Service Level Agreements are fulfilled. By thoroughly evaluating the approach, we show that a cost saving of more than 20% can be achieved.
Due to the high elasticity, availability, durability, and low IT maintenance cost, cloud storage services have gained popularity in recent years. However, the decision which cloud storage service is the most suitable one is not trivial and relying on only one cloud storage service involves the risk of vendor lock-in. One solution to avoid this, is the redundant usage of different cloud storage services.
In this thesis, we propose three different resource provisioning approaches that dynamically store data in a cost-efficient way on multiple cloud storage services, by considering customer-defined Service Level Agreements and monitored data access patterns. Furthermore, specialized long-term storage services are considered
for not or rarely used data. In the evaluation, we show that our approach has the potential to achieve a cost saving of 30% to 50%, depending on the baseline.
Cloud Computing / Elasticity / Elastic Processes / Elastic Process Execution / Business Processes / Cloud Storage / Optimization for cost efficiency
Created from the Publication Database of the Vienna University of Technology.