Trial lecture
«Consensus under crash and Byzantine failures»
Time and place: February 16, 2024 11:15 AM, Lille auditorium, Kristen Nygaards hus
Main research findings
The Internet of Things (IoT) connects billions of devices with varying capacities that produce constant streams of data. Distributed stream processing systems (DSPS) are used to analyze these data streams and the processing is often performed close to the sources to optimize DSPS performance. In the highly dynamic environments created by IoT, (1) DSPSs need to adapt through updating their distribution in the form of operator migration and (2) it is very challenging to evaluate these systems through repeatable and reproducible experiments. One major contribution of this dissertation is the development of tools to simplify the evaluation of DSPSs, for example, defining experiments easily, running them on different DSPS, making fairer comparisons, and ensuring experiments can be repeated. The second main contribution is related to operator migration. This dissertation provides a novel conceptual model of operator migration in a comprehensive tutorial. Furthermore, two novel operator migration mechanisms are developed to work in geographically distributed environments. This dissertation utilizes the developed performance evaluation tools to compare the operator migration mechanisms with state-of-the-art operator migration mechanisms and demonstrate the advantages of the new mechanisms.
Adjudication committee:
-
Professor Dr. rer. nat. Ruben Mayer, University of Bayreuth, Germany
-
Professor Dr. rer. nat. Sukanya Bhowmik, University of Potsdam, Germany
-
Professor Martin Giese, Department of Informatics, University of Oslo, Norway
Supervisors:
- Professor Thomas Peter Plagemann, Department of Informatics, University of Oslo, Norway
- Dr. Stein Kristiansen, Department of Informatics, University of Oslo, Norway
Chair of defence:
Associate Professor Ragnhild Kobro Runde
Contact information at Department: Pernille Adine Nordby