Data Transfer Sensitive and Privacy Preserving Scheduling Strategies in Hybrid Cloud Environments

Within this thesis we present an extensive analysis of data transfer aspects forBusiness Process Management System (BPMS) in a cloud based environment. The results of this analysis are used to develop a data transfer sensitive and privacy preserving process scheduling strategy and evaluate this novel strategy against a baseline approach in a testbed. Most of the state of the art resource provisioning and cloud based process schedul- ing algorithms omit the aspect of data transfer, as they often only deal with software systems, which are only deployed within one cloud instance. The network communication within the single entities in the cloud is very efficient and the data sent among the different services are often too small to affect the execution within a BPMS. With the rise of Big Data applications and the trend to create hybrid cloud solutions, the data transfer aspects become more important than in the past. The most important data trans- fer aspects are data transfer costs and data transfer duration, which directly affect the workflow execution, but there are also non-functional data transfer aspects, like privacy issues. Hybrid cloud instances often pose different levels of security and some of these security levels might be insufficient for single services or data transmissions, which poses additional requirements to the scheduling approach. To confront the problems posed by increasing data transmission sizes and hybrid cloud platforms, we extended the Vienna Platform for Elastic Processes (ViePEP) to support hybrid clouds. Further we revised the previously data agnostic process schedul- ing algorithm to respect several data transfer aspects. The evaluation has shown, that the extended ViePEP is capable of managing service executions in hybrid clouds under given privacy Service Level Agreement (SLA)s. Additionally the performance evalua- tion has shown, that our improved scheduling approach yields a cost reduction of about 30 % for the total execution costs and a cost reduction of more than 50 % for external leasing and data transfer costs in comparison with the baseline approach.
 @mastersthesis{hochreiner2014wu,
  author = {Hochreiner, Christoph},
  title = {Data Transfer Sensitive and Privacy Preserving Scheduling Strategies in Hybrid Cloud Environments},
  school = {WU Wien},
  year = {2014}
}