Model for Performance Analysis of Distributed Stream Processing Applications

Logo poskytovatele

Varování

Publikace nespadá pod Filozofickou fakultu, ale pod Fakultu informatiky. Oficiální stránka publikace je na webu muni.cz.
Autoři

NÁLEPA Filip BATKO Michal ZEZULA Pavel

Rok publikování 2015
Druh Článek ve sborníku
Konference Database and Expert Systems Applications
Fakulta / Pracoviště MU

Fakulta informatiky

Citace
Doi http://dx.doi.org/10.1007/978-3-319-22852-5_42
Obor Informatika
Klíčová slova Stream processing; Performance analysis; Data stream model
Popis Nowadays, a lot of data is produced every second and it needs to be processed immediately. Processing such unbounded streams of data is often applied in a distributed environment in order to achieve high throughput. There is a challenge to predict the performance-related characteristics of such applications. Knowledge of these properties is essential for decisions about the amount of needed computational resources, how the computations should be spread in the distributed environment, etc. In this paper, we propose a model to represent such streaming applications with the respect to their performance related properties. We present a conversion of the model to Colored Petri Nets (CPNs) which is used for performance analysis of the original application. The behavior of the proposed model and its conversion to the CPNs is validated through experiments. Our prediction was able to achieve nearly 100 % precise maximum delays of real stream processing applications.
Související projekty:

Používáte starou verzi internetového prohlížeče. Doporučujeme aktualizovat Váš prohlížeč na nejnovější verzi.