The Performance Analysis Of Web Services Composition Based On Hierarchical Colored Petri Nets

Due to the Internet’s dynamic openness, the web service composition performance is influenced not only by the BPEL process structure and its branch probability, also influenced by the load of web server, including the web service composition analyzing the performance of the network, the load bring to the web server and the load added to the forming of web server caused by other web service combination, the handling potentials of software and hardware, the scheduling policy vulnerable. At present, the mainly methods, namely queuing Petri nets, stochastic Petri nets,Markov process and Pi calculus analysis methods are used to analyze web service composition performances. But these methods cann’t model all factors above at the same time, and can’t reflect the composition performance thoroughly on the Internet.In [1], we map the various factors which influence the performance of web service composition to the queuing network, and establish the analysis index system and its calculation formula of web service performance, on the basis of these performance analysis index system, we use mathematical software named Mathematica to compute the performance of web service composition and its change rules.In this thesis, to validate the computational analysis result of queuing networks of the literature [1] in by using of a hierarchical colored Petri net(HCPN) and simulation tools CPN Tools. At first, we map all the above factors influence the web service composition performance to the CPN Tools, and put forward a method which can analyze the web service composition performance based on CPN Tools, and the simulation experiment shows web service composition performance and its various rules, simulation results are highly consistent with the results which obtained by queuing network in[1], so as to achieve the two-way mutual verification purposes between queuing network performance analysis model and the CPN Tools performance simulation model. The main contributions are as follows:1. According to different scheduling strategies, considering various influence factors of Web service composition performance, to establish the HCPN model of two scheduling strategies: first-come first-serve( FCFS) and shared services(PS)model.2. Based on this model, the BPEL mapping in structured activities to HCPN, so we can analyze web service composition performances by HCPN.3. Performance analysis of BPEL instance converted by simulation tools CPN Tools.(1) Simulation data collected by the monitor of CPN Tools.(2) Arrival time analysis of Web service composition performance parameters and its variation, that is, not the performance of Web service composition and its variation when the customer arrives to interference customer arrival and presence of interference by changing the customer.(3) To analyze web service composition and its variation by changing the probability of the execution of performance of the branch.(4) To analyze web service composition and its variation by changing the service node and service time parameter.(5) To compare the Results of the two models(FCFS and PS).