Refine
Document type
Language
- English (4)
Is part of the Bibliography
- No (4)
Keywords
- Analysis (1)
- Apache Kafka (1)
- ClickHouse (1)
- Event stream processing (1)
- GQL (1)
- Graph analytics (1)
- Graph database (1)
- Path semantics (1)
- Software-defined networking (1)
- Wikipedia (1)
- Worst-case optimal join (1)
- cyber physical systems (1)
Course of studies
- INM - Informatik (1)
Compared to relational databases, graph database systems provide a novel way of processing and analyzing highly interconnected data. Due to their unique properties, graph databases embody an interesting area of research in academic circles. For this reason, this work is fundamentally concerned with examining the state of the industry and current challenges. In this regard, we revisit the basic concepts and highlight the tremendous heterogeneity of available systems using the example of differing path semantics. Based on this insight, we explore algorithmic advancements for graph query processing regarding path finding and worst-case optimal joins. Subsequently, we discuss issues regarding performance and support for graph analytics. Finally, we provide an overview of GQL, a joint standardization effort towards unification of property graph databases.
Wikipedia is the largest free encyclopedia and one of the most popular websites worldwide. Analyzing user activity within this encyclopedic ecosystem represents unique opportunities for academic research and analysis. For this reason, this work is fundamentally concerned with obtaining and processing real-time article edit streams from Wikipedia. In this regard, we leverage the Wikimedia EventStreams API and propose a general-purpose event pipeline allowing for further processing of observed page edits. In the suggested pipeline, events are ingested and transported via an Apache Kafka cluster and inserted into a ClickHouse database for storage and analysis. Finally, we confirm the viability of our design by exploring several exemplary analytical use cases.
Traditional networking severely limits the dynamic requirements of newly emerging
use cases like the Internet of Things. For this reason, this thesis is fundamentally
concerned with the proposal of a software-defined networking approach for industrial
IoT networks. We revisit the core concepts, examine related work and subsequently
present an SDN-focused edge architecture for the KOSMoS research project. In this regard, we aim to improve network flexibility, scalability, maintainability and security.
Finally, the developed concept is implemented as an emulated proof of concept in order to assess its feasibility. For our prototype, we present an intent-driven approach that automatically compiles and deploys network configuration policies based on business goals submitted as description templates.