Diploma and Master Theses (authored and supervised):
"ViePEP - Vienna Platform for Elastic Processes";
Supervisor: S. Dustdar, S. Schulte, S. Venugopal;
Institut für Informationssysteme, AB Verteilte Systeme,
final examination: 2013-03-05.
Within this thesis, we propose a novel Business Process Management System for the Cloud. It is able to process several hundred workflows simultaneously while monitoring their executions in order to counteract against potential Service Level Agreement violations through scheduling the queued workflows and acquiring additional computing resources when needed or releasing unneeded ones.
Elastic Processes are a novel paradigm from the field of Cloud computing. It combines the various facets of elasticity that captures process dynamics in the Cloud. Elastic Processes are described in three ways: cost elasticity, resource elasticity and quality elasticity. Nowadays, these elasticities are also relevant for workflows in the Cloud. A workflow consists out of several
individual processes, each requires a different amount of resources, follows different Quality-of-Service attributes and produces a different amount of costs. Since in today´s workflows resource
intensive tasks get more and more common, and the amount of workflows executed in parallel varies over time, the amount of needed resources will also diversify enormous. That is way it is necessary to acquire additional computing resources during the system´s runtime, or release
some whenever they are not needed anymore.
Therefore, we developed ViePEP - the Vienna Platform for Elastic Processes. ViePEP is on the one hand a Business Process Management System for the Cloud, capable to manage and process the execution of several hundred workflows simultaneously. It further monitors their
executions in order to identify potential Service Level Agreement violations in time. And on the other hand, ViePEP is able to counteract against the lack of needed or the excess of used
computing resources. Using a prediction model, ViePEP is not just able to counteract in-time but also predict the future need of resources for the near future in order to acquire additional
resources punctual or release unneeded resources. By evaluating ViePEP with different configured experiments we have shown, that ViePEP is able to automatically process workflows, while
taking care of Service Level Agreements through scheduling their executions and acquiring or releasing computing resources.
Created from the Publication Database of the Vienna University of Technology.