The Introduction of QNAT and Webstress
Transcription
The Introduction of QNAT and Webstress
Queueing Network Analysis Tool & Webserver Stress Tool Advisor: Professor Bai Student: Yung-Sen Cheng Date: September 25, 2006 Computer System Laboratory, Institute of Electronic Engineering, FJU 1 Outline The Introduction of QNAT The Installation of QNAT How to use QNAT The Introduction of Webstress The Installation and Use of Webstress References Conclusion Demo Computer System Laboratory, Institute of Electronic Engineering, FJU 2 The Introduction of QNAT Queueing Network Analysis Tool Easy to setup parameters High-performance analysis and simulation Based on Mathematica Computer System Laboratory, Institute of Electronic Engineering, FJU 3 The Installation of Mathematica 4.0 Computer System Laboratory, Institute of Electronic Engineering, FJU 4 Ask for the License ID and Password Computer System Laboratory, Institute of Electronic Engineering, FJU 5 MATHPASS.EXE Æ generate ID & PW Computer System Laboratory, Institute of Electronic Engineering, FJU 6 Key in the License ID and Password Computer System Laboratory, Institute of Electronic Engineering, FJU 7 The Installation of QNAT Computer System Laboratory, Institute of Electronic Engineering, FJU 8 The Pre-Setup of QNAT Computer System Laboratory, Institute of Electronic Engineering, FJU 9 The Finish of Installation Computer System Laboratory, Institute of Electronic Engineering, FJU 10 How to Use QNAT α λ μ1 μd 1−α μ2 Computer System Laboratory, Institute of Electronic Engineering, FJU 11 Queueing Network Analysis Tool Computer System Laboratory, Institute of Electronic Engineering, FJU 12 The Blocking Nodes in the Network M/M/1/N M/M/1 Computer System Laboratory, Institute of Electronic Engineering, FJU 13 Computer System Laboratory, Institute of Electronic Engineering, FJU 14 The Type of Network Computer System Laboratory, Institute of Electronic Engineering, FJU 15 The Type of Blocking Transfer Æ The blocked customer waits at the server from where it has just finished service and blocks it till the destination becomes unblocking. RS-RD Æ The destination node is chosen according to the routing probabilities after every service completion. RS-FD Æ The customer attempts to join the same destination node after subsequent service completions. Rejection Æ The blocked customer leaves the network. Computer System Laboratory, Institute of Electronic Engineering, FJU 16 The Number of nodes α λ 1 μd 1−α 2 μ1 3 μ2 Computer System Laboratory, Institute of Electronic Engineering, FJU 17 Click Mouse Computer System Laboratory, Institute of Electronic Engineering, FJU 18 The Property of a node λ μ The Size of Buffer Computer System Laboratory, Institute of Electronic Engineering, FJU 19 The Property of a node (cont.) μ λ μ μ The Number of Servers = 3 Computer System Laboratory, Institute of Electronic Engineering, FJU 20 Computer System Laboratory, Institute of Electronic Engineering, FJU 21 The Result of Analysis of a node E(n) E(t) ρ Computer System Laboratory, Institute of Electronic Engineering, FJU 22 The Performance of the System Computer System Laboratory, Institute of Electronic Engineering, FJU 23 Collect and Plot 2.0 1.8 Service Rate = 20 Service Rate = 10 Response Time (sec) 1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0 1 2 3 4 5 6 7 8 Arrival Rate (Requests/sec) 9 10 Computer System Laboratory, Institute of Electronic Engineering, FJU 24 The Introduction of Webstress Webserver Stress Tool Performance Tests Load Tests Stress Tests Ramp Tests Computer System Laboratory, Institute of Electronic Engineering, FJU 25 Download Webstress http://cs.ee.fju.edu.tw Computer System Laboratory, Institute of Electronic Engineering, FJU 26 How to Use Webstress Testing Time λ Computer System Laboratory, Institute of Electronic Engineering, FJU 27 Input the URL of Test The Number of URLs Input the URL of Test Computer System Laboratory, Institute of Electronic Engineering, FJU 28 Response Time Computer System Laboratory, Institute of Electronic Engineering, FJU 29 Save the Result to File Computer System Laboratory, Institute of Electronic Engineering, FJU 30 The Result of Measurement the Webpage of Multiplication Operation Click Times and Errors (per URL) 110 g b c d e f b c d e f g Average Request Time [ms] 100 90 Req-Times: Errors: 80 70 60 50 40 30 20 10 0 0 500 1,000 Time Since Start of Test [s] Errors [%] 0 1,500 Test Ty pe: TIME (run test f or 30 minutes) User Simulation: 10 simultaneous users - 1 seconds between clicks Computer System Laboratory, Institute of Electronic Engineering, FJU 31 References Chia-Ya Chen, “The Performance Equalizing Analysis and Measurement for Multiple Grouping of the Web Cluster System,” Master Thesis, Department of Electronic Engineering, Fu Jen Catholic University, Taipei, Taiwan, July 2004. Yu-Nien Yang, “Analysis and Measurement of the Equivalent Model of Serial and Parallel Queues for a Web Cluster with a Low Rejection Rate,” Master Thesis, Department of Electronic Engineering, Fu Jen Catholic University, Taipei, Taiwan, June 2006. D. Manjunath, D.M. Bhaskar, Hema Tahilramani, Sanjay K. Bose, and M.N. Umesh, “QNAT: A GRAPHICAL TOOL FOR THE ANALYSIS OF QUEUEING NETWORKS,” Proceedings of IEEE Region 10 International Conference on Global Connectivity in Energy, Computer, Communication and Control, 1998. Hema Tahilramani Kaur, D. Manjunath, and Sanjay K. Bose, “The Queueing Network Analysis Tool (QNAT),” Proceedings of IEEE 8th International Symposium on Modeling, Analysis and Simulation of Computer and Telecommunication Systems, 2000. http://download2.paessler.com/download/webstressmanual.pdf Computer System Laboratory, Institute of Electronic Engineering, FJU 32 Conclusion The most important is the correctness of your model. Collect and plot the results, and then verify them. Compare results of simulation and measurement with theorem. Computer System Laboratory, Institute of Electronic Engineering, FJU 33 Demo - Measure the Response Time of Embedded Home Server Computer System Laboratory, Institute of Electronic Engineering, FJU 34