Article Preview
Top1. Introduction
Globalizations of businesses through internet and service-oriented computing have provided the flexible opportunity to the users and enterprises to develop and deliver their services over network. These services, popularly termed as web services are becoming a promising technology to deliver interoperable services. They follow service-oriented architecture with some specific roles (Jones, 2005). The web services serve as: (i) service provider, (ii) requester, and (iii) broker. However, with the rapid growth of internet users, it is a good practice to serve consumers of web services without any server load errors. The inclusion of load balancing technique of clustering web server enhances the web server capacity to increase the server-side performances. However, clustering web server may impute some impact on the performance and reliability aspect of hosting web services. As such, evaluating such a system is an important concern from the perspective of end users. The software testers are the key persons who vigorously test the service-oriented computing based system. The concerned person should have a familiar knowledge about the system and the inputs to the operating environment to carry out the system testing effectively (Whittaker, 2000). A highly important factor is to verify that the system is scalable, stable and reliable against massive growth of web service consumers. As such, rigorous testing and statistical analysis are considered to be a good candidature for evaluating the overall quality of service-oriented system (Kalita et al., 2011). However, a data sample needs to be collected which describes the behavior of the running applications (Kallepalli et al., 2001). This work is an extended part of our study which is discussed elsewhere (Bora et al., 2013, Bora et al., 2015). In our previous work, we had deployed the architecture of client, broker and child web services in non-cluster-based web server and carried out the study over that. However, this paper presents the system metrics by deploying the web services in cluster-based load balancing web server. The experiment is carried out by gradually increasing the load over the cluster-based load balancing web server. This study considers such an architecture since now-a-days the software industries deploy multiple web services with different roles for computational implementation towards a goal. In this paper, the Simple Object Access Protocol (SOAP) based web services are deployed in load balancing cluster-based Tomcat web server. The novelty of the study is the highlights of the performance and reliability metrics of multi web services execution by using different cluster sizes of load balancing web server. A novel methodology is deployed which mimics the system’s behavior against different stresses of users if deployed in single server machine with different cluster sizes of web server. The system mimic can be defined as the way the web server machine with web service or other web architectures will perform as closely as possible like live server over internet. It gives an idea about the real-time assessment of web service-oriented system. The paper is also trying to cover the performance and reliability aspects of load balancing cluster-based web server for massive requests while executing multiple web services that can play the individual role as client, broker and service provider. The stability of cluster-based web server is better than non-cluster-based web server (Bezboruah et al., 2015), but what will be the quality aspects of cluster-based web server beyond and within its capacity limit? A detailed discussion on an experimental arrangement, testing results and statistical analysis is presented here.