Heterogeneous Multi-Model Streaming Data Management (PhD Thesis, ongoing)
Data-driven systems today are increasingly reliant on real-time, dynamic data, or data in-motion, to stay current and make informed decisions. This is evident across a wide range of applications, from analyzing sensor data in high-tech sports like Formula 1 and low-tech sports like soccer, to reacting to burst-heavy financial market data and monitoring live IoT and smart devices.
The Challenge of Dynamic Data
Unlike traditional systems that handle static data and can rely on rigid structures and formats for optimization, dynamic data streams present a unique set of challenges. These include:
- Inconsistent data: Data points may be missing, delayed, or out of order.
- Irregular loads: Data arrives in unpredictable bursts.
- Heterogeneous formats: The data can come from many different sources and lack a uniform structure.
Traditionally, data engineers manage these challenges, but modern data landscapes have also shifted toward architectures with overlapping data models. This adds complexity, as these new models define how data is stored and how operations interact with it.
Project Goals and Contributions
The objective of this PhD project is to address the complexities of managing dynamic data by providing similar guarantees to those found in traditional database management. The project will focus on developing solutions for handling heterogeneous, in-motion data that may or may not conform to a specific data model.
The target contributions can be categorized into two areas:
- Storing: Developing methods to efficiently store and manage dynamic data streams, accounting for their unique characteristics.
- Processing: Creating solutions to process this data reliably, ensuring consistency and integrity despite its dynamic nature.
Staff
Examiners
Prof. Dr. Heiko Schuldt, Prof. Dr. Isabel Wagner
Start Date
2022-08-16