Availability Measurement and Modeling for An Application Server

Authors:
Dong Tang;Dileep Kumar;Sreeram Duvur;Oystein Torbjornsen
Affiliations:
Sun Microsystems, Inc., Santa Clara, CA;Sun Microsystems, Inc., Santa Clara, CA;Sun Microsystems, Inc., Santa Clara, CA;Sun Microsystems, Inc., Santa Clara, CA
Venue:
DSN '04 Proceedings of the 2004 International Conference on Dependable Systems and Networks
Year:
2004

Citing 0
Cited 3

Simulation-based analysis of middleware service impact on system reliability: Experiment on Java application server

Journal of Systems and Software
MEMS: a method for evaluating middleware architectures

QoSA'06 Proceedings of the Second international conference on Quality of Software Architectures
Hierarchical availability analysis of multi-tiered Web applications

Software Quality Control

Quantified Score

Hi-index	0.00

Visualization

Abstract

Application Server is a standard middlewareplatform for deploying web-based business applicationswhich typically require the underlying platform to deliverhigh system availability and to minimize loss oftransactions. This paper presents a measurement-basedavailability modeling and analysis for a fault tolerantApplication Server system - Sun Java System ApplicationServer, Enterprise Edition 7. The study applieshierarchical Markov reward modeling techniques on thetarget software system. The model parameters areconservatively estimated from lab or field measurements.The uncertainty analysis method is used on the model toobtain average system availability and confidenceintervals by randomly sampling from possible ranges ofparameters that cannot be accurately measured in limitedtime frames or may vary widely in customer sites. Asdemonstrated in this paper, the combined use of labmeasurement, analytical modeling, and uncertaintyanalysis is a useful evaluation approach which canprovide a conservative availability assessment at statedconfidence levels for a new software product.