EWnFM: An Environment States Oriented Web Service Non-Functional Property Model

A proper model of Web service non-functional properties is the key foundation to the evaluation of non-functional properties of Adaptive Service Based Software (ASBS) systems. As the environment in which a Web service is deployed may keep changing, environmental factors would affect the non-function...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Entropy (Basel, Switzerland) Switzerland), 2015, Vol.17 (2), p.509-527
Hauptverfasser: Zhang, Yin, Ge, Liang, Gao, Kening, Zhang, Bin, Xue, Zhuyin
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:A proper model of Web service non-functional properties is the key foundation to the evaluation of non-functional properties of Adaptive Service Based Software (ASBS) systems. As the environment in which a Web service is deployed may keep changing, environmental factors would affect the non-functional properties of a Web service a lot. However, available non-functional property models usually ignore the impact of environmental factors, leading to insufficient modeling power of non-functional properties, limited effect of system wide non-functional property evaluation based on these models, and the inability to support environment states oriented specifications of ASBS. This paper propose an environment states oriented Web service non-functional property model. By considering the differences of a non-functional property under different environment states, environment states of a Web service is analyzed using a Dirichlet process based method. With such a foundation, an environment states oriented Web service non-functional property model is introduced, together with the parameter estimation methods based on historical monitor data. Experiment results have shown that compared to the evaluated methods, our model could generate data that are much close to real monitored data.
ISSN:1099-4300
1099-4300
DOI:10.3390/e17020509