Author: Andreas Dittrich Ph.D. Thesis, submitted to the Faculty of Mathematics and Natural Sciences, Humboldt University of Berlin, December 9, 2014. Oral defense March 20, 2015. Grade: magna cum laudae. Reviewers: Prof. Dr. Miroslaw Malek, Lugano, Switzerland Prof. Dr. Alexander Reinefeld, Berlin, Germany Prof. Dr. Jörg Kaiser, Magdeburg, Germany Download: published version, meta information

Dependability of service provision is one of the primary goals in modern networks. Since providers and clients are part of a connecting Information and Communications Technology (ICT) infrastructure, service dependability varies with the position of actors as the ICT devices needed for service provision change. We present two approaches to quantify user-perceived service dependability. The first is a model-driven approach to calculate instantaneous service availability. Using input models of the service, the infrastructure and a mapping between the two to describe actors of service communication, availability models are automatically created by a series of model to model transformations. The feasibility of the approach is demonstrated using exemplary services in the network of University of Lugano, Switzerland. The second approach aims at the responsiveness of the service discovery layer, the probability to find service instances within a deadline even in the presence of faults, and is the main part of this thesis. We present a hierarchy of stochastic models to calculate user-perceived responsiveness based on monitoring data from the routing layer. Extensive series of experiments have been run on the Distributed Embedded Systems (DES) wireless testbed at Freie Universität Berlin. They serve both to demonstrate the shortcomings of current discovery protocols in modern dynamic networks and to validate the presented stochastic models. Both approaches demonstrate that the dependability of service provision indeed differs considerably depending on the position of service clients and providers, even in highly reliable wired networks. The two approaches enable optimization of service networks with respect to known or predicted usage patterns. Furthermore, they anticipate novel service dependability models which combine service discovery, timeliness, placement and usage, areas that until now have been treated to a large extent separately.

	Authors: Andreas Dittrich Daniel Solis Herrera Pablo Coto Miroslaw Malek 20th IEEE Pacific Rim International Symposium on Dependable Computing, PRDC 2014, Singapore, November 18-21, 2014 Download: accepted version, final published version

Service Discovery (SD) is an integral part of service networks. Before a service can be used, it needs to be discovered successfully. Thus, a comprehensive service dependability analysis needs to consider the dependability of the SD process. As a time-critical operation, an important property of SD is responsiveness: the probability of successful discovery within a deadline, even in the presence of faults. This is especially true for dynamic networks with complex fault behavior such as wireless networks. We present results of a comprehensive responsiveness evaluation of decentralized SD, specifically active SD using the Zeroconf protocol. The ExCovery experiment framework has been employed in the Distributed Embedded System (DES) wireless testbed at Freie Universität Berlin. We present and discuss the experiment results and show how SD responsiveness is affected by the position and number of requesters and providers as well as the load in the network. Results clearly demonstrate that in all but the most favorable conditions, the configurations of current SD protocols struggle to achieve a high responsiveness. We further discuss results reflecting the long-term behavior of the testbed and how its varying reliability impacts SD responsiveness.

	Authors: Andreas Dittrich Stefan Wanja Miroslaw Malek 28th IEEE International Symposium on Parallel Distributed Processing, Workshops and PhD Forum, IPDPSW 2014, Phoenix, AZ, USA, May 19-23, 2014 Download: accepted version, final published version GitHub Project Page and Source Repository

Experiments are a fundamental part of science. They are needed when the system under evaluation is too complex to be analytically described and they serve to empirically validate hypotheses. This work presents the experimentation framework ExCovery for dependability analysis of distributed processes. It provides concepts that cover the description, execution, measurement and storage of experiments. These concepts foster transparency and repeatability of experiments for further sharing and comparison. ExCovery has been tried and refined in a manifold of dependability related experiments during the last two years. A case study is provided to describe service discovery as experiment process. A working prototype for IP networks runs on the Distributed Embedded System (DES) wireless testbed at the Freie Universität Berlin.

	Authors: Andreas Dittrich Björn Lichtblau Rafael Rezende Miroslaw Malek 17th International GI/ITG Conference on “Measurement, Modelling and Evaluation of Computing Systems” and “Dependability and Fault-Tolerance”, MMB & DFT 2014, Bamberg, Germany, March 17-19, 2014 Download: accepted version, the final publication is available at link.springer.com

In service networks, discovery plays a crucial role as a layer where providers can be published and enumerated. This work focuses on the responsiveness of the discovery layer, the probability to operate successfully within a deadline, even in the presence of faults. It proposes a hierarchy of stochastic models for decentralized discovery and uses it to describe the discovery of a single service using three popular protocols. A methodology to use the model hierarchy in wireless mesh networks is introduced. Given a pair requester and provider, a discovery protocol and a deadline, it generates specific model instances and calculates responsiveness. Furthermore, this paper introduces a new metric, the expected responsiveness distance der, to estimate the maximum distance from a provider where requesters can still discover it with a required responsiveness. Using monitoring data from the DES testbed at Freie Universität Berlin, it is shown how responsiveness and der of the protocols change depending on the position of nodes and the link qualities in the network.

	Authors: Rafael Rezende Andreas Dittrich Miroslaw Malek 19th IEEE Pacific Rim International Symposium on Dependable Computing (PRDC), Vancouver, BC, Canada, December 2-4, 2013 Download: accepted version, final published version

Today’s businesses rely ever more on dependable service provision deployed on information and communications technology (ICT) infrastructures. Service dependability is highly influenced by the individual infrastructure component properties. Combining these properties for consistent dependability analysis is challenging as every service requester might use a different set of components during service usage, constituting the user-perceived view on a service. This paper presents a methodology to evaluate user-perceived instantaneous service availability. It uses three input models: (1) The ICT infrastructure, with failure rates, repair rates and deployment times of all components, (2) an abstract description of complex hierarchical services, (3) a mapping that contains concrete ICT components for the service pair requester and provider, as well as their existing replicas, and a duration of usage. The methodology then automatically generates an availability model from those parts of the ICT infrastructure needed during provision for the specified pair. To calculate instantaneous availability, the age of the ICT components, the order and time of their usage during service provision are taken into account. The methodology supports generation of different availability models, we demonstrate this by providing reliability block diagrams and fault-trees. We demonstrate the feasibility of the methodology by applying it to parts of the network infrastructure of University of Lugano, Switzerland.