Next to engines designed for purely streaming data, like Twitter (Apache) Storm, approaches like Apache Flink or Spark Streaming operate on micro-batches interpreted as streaming data. This suggest to review how well traditional NoSQL systems can be used to process continuous queries over streaming data where the high-rate stream is materialized by storing and removing tuples in the NoSQL engines. Specifically, sliding-window-based analytical queries over structured, relational data, will be considered and the devised concepts should be integrated as an extension to the Antidote datastore.
The thesis will be co-adviced by Prof. Michel and Annette Bieniusa.
Prerequisites: Good programming skills (implementation language is Erlang or Java); preferably also Lectures “Distributed Data Management”