Font Size: a A A

Research Of Web Server Parameter Optimization Based On Hybrid Binary Particle Swarm Optimization Algorithm

Posted on:2018-04-16Degree:MasterType:Thesis
Country:ChinaCandidate:J Y ChenFull Text:PDF
GTID:2348330536978217Subject:Engineering
Abstract/Summary:PDF Full Text Request
With the development of information technology,the amount of network users has increased dramatically,and the Web server has been widely used in large software systems.Usually,the server needs to be configured with several parameters related to performance before running.By adjusting the configuration parameters of the Web server,the performance of the Web server can be improved greatly.However,even for a simple Web server,its configuration parameters are as many as hundreds.If these configuration parameters are adjusted by manual,it will cost too much time,so the research of Web server parameter optimization is of great significance.Traditional research of Web server parameter optimization is usually carried out in the simulation condition and often not satisfactory in the real world.These researches can only optimize 1 to 2 configuration parameters and need to spend a lot of time.In order to obtain the reasonable configuration parameters of Web server more conveniently and quickly,a parameter optimization method based on hybrid binary particle swarm for Web server based on J2 EE multi-tier architecture is proposed.This method can find the optimal or near optimal configuration parameters of the Web server automatically,so that the Web server can display the best overall performance.This paper mainly includes the following contents:(1)To automatic optimize parameters and obtain a realistic optimization effect,an experimental platform is setted up.In this paper,we use Tomcat and MySQL to build a Web server which is based on J2 EE multi-tier architecture,and deploy a small online shopping site on this server.Jmeter is selected to be performance test software.To connect Web server and test software,a control program realized the optimization algorithm with Java is designed.The experiment platform is made up of Web server,test software and control program,which can optimize configuration parameters of Web server automatically.(2)To optimize multiple parameters of the Web server,a Web server parameter optimization method based on the binary particle swarm optimization(BPSO)algorithm is proposed.We deep study the basic principle and workflow of BPSO algorithm,and apply it to the Web server parameter optimization problems.According to the documentation of Tomcat,MySQL and JVM,encode their main tunable configuration parameters.And Math.random()method is used generate random intial population of BPSO algorithm.Then,BPSO algorithm will calculate the fitness value and search for global optimal solution iteratively.Finally,we give the experimental results.The experimental results show that BPSO algorithm can solve the optimization problem of Web server parameters well,and it can improve the performance of Web server atmost 51%.(3)To obtain the reasonable configuration parameters of the Web server more quickly,a Web server parameter optimization method based on the hybrid binary particle swarm optimization algorithm is proposed.To improve the common BPSO algorithm,we propose a hybrid binary particle swarm optimization algorithm.This hybrid algorithm add experience factor and the nonlinear inertia weight factor which has control factor to improve searching efficiency and accuracy.Bying adding hill-climbing algorithm,the hybrid algorithm overcome the problem that the common BPSO trapp into local optimal value easily in the later time.Then,we use this algorithm to solve the Web server parameter optimization problem and comparing the experiment results with the results of common BPSO algorithm.The comparison results show that the method can reduce the experimental time greatly,and can obtain a better configuration parameter combination for the Web server.
Keywords/Search Tags:Web server, parameter optimization, Binary Particle Swarm Optimization(BPSO), empirical factor, mountain climbing algorithm
PDF Full Text Request
Related items