Main Content

Data-driven Cloud Computing

A cloud "raining" a binary data stream.

Cloud computing is an approach for using hardware/software resources over the Internet. To keep the implementation details of the resource supply away from users and to offer them an adequate level of abstraction in dealing with resources, the resources are often virtualised and provided by an appropriate cloud/web service API.

We have investigated several issues of cloud computing, such as the analysis and verification of cloud access control policies, cloud map-reduce for systems biology applications, utility-driven resource allocation to virtual machines, increasing virtual machine security, grid/cloud meta-scheduling combined with peer-to-peer technologies, cloud/web service orchestration, and platform-as-a-service clouds.

Furthermore, we looked at machine learning methods to solve a variety of problems in cloud computing infrastructures in a data-driven manner without following predefined rules. Apart from supporting tasks requiring classification, regression, and prediction, they can be used for anomaly detection, decision making, and behavior modelling in a situational manner with the goal of improving adaptability, performance, resilience, and security. In particular, we developed machine learning approaches for predicting overload and underload of cloud servers for live migration of virtual machines, and for predicting resource demand in cloud infrastructures to improve resource allocation decisions.

Selected Publications

  • Dorian Minarolli, Artan Mazrekaj, Bernd Freisleben:
    Tackling Uncertainty in Long-term Predictions for Host Overload and Underload Detection in Cloud Computing. Journal of Cloud Computing, 6: 4, 2017
  • Meryeme Ayache, Mohammed Erradi, Ahmed Khoumsi, Bernd Freisleben:
    Analysis and Verification of XACML Policies in a Medical Cloud Environment. Scalable Computing - Practice and Experience 17(3): 189-206, 2016
  • Pablo Graubner, Lars Baumgärtner, Patrick Heckmann, Marcel Müller, Bernd Freisleben:
    Dynalize: Dynamic Analysis of Mobile Apps in a Platform-as-a-Service Cloud. 8th IEEE International Conference on Cloud Computing, New York City, NY, USA, 925-932, IEEE, 2015
  • Dorian Minarolli, Bernd Freisleben:
    Cross-Correlation Prediction of Resource Demand for Virtual Machine Resource Allocation in Clouds. 6th International Conference on Computational Intelligence, Communication Systems, and Networks, Tetova, 2014, pp. 119-124, IEEE, 2014
  • Dorian Minarolli, Bernd Freisleben:
    Distributed Resource Allocation to Virtual Machines via Artificial Neural Networks. 22nd Euromicro International Conference on Parallel, Distributed, and Network-Based Processing, Torino, Italy, 490-499, IEEE, 2014
  • Dorian Minarolli, Bernd Freisleben:
    Virtual Machine Resource Allocation in Cloud Computing via Multi-Agent Fuzzy Control. 2013 International Conference on Cloud and Green Computing, Karlsruhe, Germany, 188-194, IEEE, 2013
  • Tolga Dalman, Tim Dörnemann, Ernst Juhnke, Michael Weitzel, Wolfgang Wiechert, Katharina Nöh, Bernd Freisleben:
    Cloud MapReduce for Monte Carlo Bootstrap Applied to Metabolic Flux Analysis. Future Generation Computer Systems 29(2): 582-590, 2013
  • Roland Schwarzkopf, Matthias Schmidt, Christian Strack, Simon Martin, Bernd Freisleben:
    Increasing Virtual Machine Security in Cloud Environments. Journal of Cloud Computing, 1: 12, 2012
  • Dorian Minarolli, Bernd Freisleben:
    Utility-driven Allocation of Multiple Types of Resources to Virtual Machines in Clouds. 13th IEEE Conference on Commerce and Enterprise Computing, Luxembourg-Kirchberg, Luxembourg, 137-144, IEEE, 2011
    (Best Paper Award)
  • Tim Dörnemann, Markus Mathes, Roland Schwarzkopf, Ernst Juhnke, Bernd Freisleben:
    DAVO: A Domain-Adaptable, Visual BPEL4WS Orchestrator. IEEE 23rd International Conference on Advanced Information Networking and Applications, Bradford, United Kingdom, 121-128, IEEE, 2009
    (Highly Commmended Paper Award)
  • Michael Heidt, Tim Dörnemann, Kay Dörnemann, Bernd Freisleben:
    Omnivore: Integration of Grid Meta-Scheduling and Peer-to-Peer Technologies. 8th IEEE International Symposium on Cluster Computing and the Grid, Lyon, France, 316-323, IEEE, 2008
    (Best Paper Award)

Further information