Bachelor Thesis BCLR-2015-24

BibliographySiegert, Valentin: Performance Profiling of a Parallelization Framework for Complex-Event-Processing Operators.
University of Stuttgart, Faculty of Computer Science, Electrical Engineering, and Information Technology, Bachelor Thesis No. 24 (2015).
63 pages, english.
CR-SchemaC.2.4 (Distributed Systems)
C.4 (Performance of Systems)
Abstract

The Internet of Things (IoT) is an upcoming technological complex of themes, whereby the recognition of complex events, happened in the real world, is no more neglectable for any new application. Complex Event Processing (CEP) builds a middleware to generate due a operator network complex events for a application out of simple input events from e.g. some sensors and thereby it should stay in a real-time latency budget at all. Since we concentrate on parallelized CEP operators and more especially on the framework PACE, the latency of one operator can be influenced by several factors, e.g. adapting the parallelization degree and or or using batch scheduling. To adjust the different factors more perfectly, we need to know the latency of the splitting and the merging before the specific situation comes up in the real process, whereby the prediction of the splitting’s latency is a more interesting question. By analyzing the latency of the splitting in the framework PACE we determine that the most influencing parameter of the latency is the number of opened selections and give a solving approach by predicting the latency of the splitting with regression learning and some time series predictions. Finally we will see that our predictions deliver nice results without outliers and that our approach is simple enough to not generate a higher runtime or rather a higher latency by asking for and calculating the prediction.

Full text and
other links
PDF (1374011 Bytes)
Department(s)University of Stuttgart, Institute of Parallel and Distributed Systems, Distributed Systems
Superviser(s)Rothermel, Prof. Kurt; Mayer, Ruben
Entry dateSeptember 25, 2018
   Publ. Department   Publ. Institute   Publ. Computer Science