direkt zum Inhalt springen

direkt zum Hauptnavigationsmenü

Sie sind hier

TU Berlin

Page Content

Master Thesis Kaikai Yang

Scalable Stream Processing with PACTs

A growing number of interet-capable devices can produce vast amounts of data in the form of streams, e.g. video and audio. This data poses new challenges for all parts of the underlying infrastructure, especially when it needs to processed under real-time constraints.

Frameworks for massively-parallel data processing in large compute clusters have become popular in both industry and acedemia. These frameworks, such as Hadoop, usually process data in a batch-job fashion, an approach which unfortunately does not address the requirements of streamed data, where results are often required within a very short timespan, e.g. less than a second.

The scope of this thesis is the design and implementation of a programming model for parallel, distributed stream processing based on Stratosphere's PACTs model. The main task is the definition of window semantics for the existing PACTs operators and their adaptation.

Prerequisite to work on this topic are profound knowledge of the Java programming language, interest in current research topics, as well as the willingness to familiarize with an existing system.

Zusatzinformationen / Extras

Quick Access:

Schnellnavigation zur Seite über Nummerneingabe

Auxiliary Functions