Trivial Medium Complex
No. of Physical I/Os 10 20 40
Total CPU Seconds 840 560 252
No. of Transactions Executed 10500 5600 2100
9.
Find the service demands at the CPU and disk for trivial, medium, and complex transactions. Assume that the arrival rates of medium and complex transactions are fixed at 1.6 tps and 0.6 tps, respectively. Using the QN solvers provided with the book, generate a graph with three average response time curves (one for each workload) as a function of the arrival rate of trivial transactions
Repeat the item above for a situation where a new disk, identical to the existing disk, is installed in the database server in such a way that the I/O demand is balanced between the two disks.
< Day Day Up >
• Table of Contents
Performance by Design: Computer Capacity Planning by Example
By Daniel A. Menascé, Virgilio A.F. Almeida, Lawrence W. Dowdy
Publisher: Prentice Hall PTR Pub Date: January 05, 2004
ISBN: 0-13-090673-5 Pages: 552
Individual organizations and society as a whole could face major breakdowns if IT systems do not meet their Quality of Service (QoS) requirements on performance, availability, security, and maintainability. Corporations stand to lose valuable income, and public interests could be put at great risk. System designers and analysts usually do not take QoS requirements into account when designing and/or analyzing computer systems, due mainly to their lack of awareness about the issues that affect
performance and the lack of a framework to reason about performance. This book describes how to map real-life systems (e.g., databases, data centers, e-commerce applications) into analytic performance models. The authors elaborate upon these models, and use them to help the reader thoroughly analyze and better understand potential performance issues.
< Day Day Up > < Day Day Up >
Bibliography
[1] C. Rose, "A measurement procedure for queuing network models of computer systems," ACM
Computing Surveys, vol. 10, no. 3, September 1978.
[2] D. Ferrari, G. Serazzi, and A. Zeigner, Measurement and Tuning of Computer Systems , Prentice Hall, Upper Saddle River, New Jersey, 1983.
[3] P. Heidelberger and S. Lavenberg, "Computer performance methodology," IEEE Transactions on
Computers, vol. C-33, no. 12, December 1984.
[4] M. Kienzle, "Measurements of computers systems for queuing network models," Technical Report
CSRG-86, Department of Computer Science, University of Toronto, Canada, October 1977.
[5] D. M. Levine, P. P. Ramsey, and R. K. Smidt, Applied Statistics for Engineers and Scientists: Using
Microsoft Excel & MINITAB, Prentice Hall, Upper Saddle River, New Jersey, 2001.
[6] D. Lavery, "The design of a hardware monitor for the Cedar supercomputer," CSRD Report no. 866 , University of Illinois at Urbana-Champaign, May 1989.
[7] I. Borovits and S. Neuman, Computer System Performance Evaluation , Lexington Books, Lexington, Massachusetts, 1979.
[8] J. Cady and B. Howarth, Computer System Performance Management and Capacity Planning , Prentice Hall, Brookvale, New South Wales, Australia, 1990.
[9] W. Highleyman, Performance Analysis of Transaction Processing Systems , Prentice Hall, Upper Saddle River, New Jersey, 1989.
[10] J. Cooper, "Capacity planning methodology," IBM Systems Journal, vol. 19, no. 1, 1980.
[11] J. P. Buzen, "A queuing network model of MVS," ACM Computing Surveys, vol. 10, no. 3, September 1978.
[12] J. Silvester and A. Thomasian, "Performance modeling of a large scale multiprogrammed computer using BEST/1," Proceedings of the International Conference on Computer Capacity Management , Chicago, 1981.
• Table of Contents
Performance by Design: Computer Capacity Planning by Example
By Daniel A. Menascé, Virgilio A.F. Almeida, Lawrence W. Dowdy
Publisher: Prentice Hall PTR Pub Date: January 05, 2004
ISBN: 0-13-090673-5 Pages: 552
Individual organizations and society as a whole could face major breakdowns if IT systems do not meet their Quality of Service (QoS) requirements on performance, availability, security, and maintainability. Corporations stand to lose valuable income, and public interests could be put at great risk. System designers and analysts usually do not take QoS requirements into account when designing and/or analyzing computer systems, due mainly to their lack of awareness about the issues that affect
performance and the lack of a framework to reason about performance. This book describes how to map real-life systems (e.g., databases, data centers, e-commerce applications) into analytic performance models. The authors elaborate upon these models, and use them to help the reader thoroughly analyze and better understand potential performance issues.
< Day Day Up >
Chapter 6. Case Study II: A Web Server
Section 6.1. Introduction Section 6.2. The Web Server
Section 6.3. Preliminary Analysis of the Workload Section 6.4. Building a Performance Model Section 6.5. Using the Model
Section 6.6. Secure Downloads
Section 6.7. Experimental Comparison of Two Servers Section 6.8. Concluding Remarks
Section 6.9. Exercises Bibliography
< Day Day Up >
• Table of Contents
Performance by Design: Computer Capacity Planning by Example
By Daniel A. Menascé, Virgilio A.F. Almeida, Lawrence W. Dowdy
Publisher: Prentice Hall PTR Pub Date: January 05, 2004
ISBN: 0-13-090673-5 Pages: 552
Individual organizations and society as a whole could face major breakdowns if IT systems do not meet their Quality of Service (QoS) requirements on performance, availability, security, and maintainability. Corporations stand to lose valuable income, and public interests could be put at great risk. System designers and analysts usually do not take QoS requirements into account when designing and/or analyzing computer systems, due mainly to their lack of awareness about the issues that affect
performance and the lack of a framework to reason about performance. This book describes how to map real-life systems (e.g., databases, data centers, e-commerce applications) into analytic performance models. The authors elaborate upon these models, and use them to help the reader thoroughly analyze and better understand potential performance issues.
< Day Day Up > < Day Day Up >
6.1 Introduction
The case study presented in this chapter introduces several important concepts in performance engineering, including the determination of confidence intervals, the computation of service demands from the results of experiments, the usage of linear regression, and comparison of alternatives through analytic modeling and through experimentation. The examples discussed in this chapter are supported by the MS Excel workbooks WSData.XLS, ClosedQN-Chap6.XLS, ClosedQN-Secure.XLS, and
ServerComparison.XLS.
• Table of Contents
Performance by Design: Computer Capacity Planning by Example
By Daniel A. Menascé, Virgilio A.F. Almeida, Lawrence W. Dowdy
Publisher: Prentice Hall PTR Pub Date: January 05, 2004
ISBN: 0-13-090673-5 Pages: 552
Individual organizations and society as a whole could face major breakdowns if IT systems do not meet their Quality of Service (QoS) requirements on performance, availability, security, and maintainability. Corporations stand to lose valuable income, and public interests could be put at great risk. System designers and analysts usually do not take QoS requirements into account when designing and/or analyzing computer systems, due mainly to their lack of awareness about the issues that affect
performance and the lack of a framework to reason about performance. This book describes how to map real-life systems (e.g., databases, data centers, e-commerce applications) into analytic performance models. The authors elaborate upon these models, and use them to help the reader thoroughly analyze and better understand potential performance issues.
< Day Day Up >
6.2 The Web Server
Consider a large software company that uses an internal Web server to allow its programmers, testers, and documentation personnel to download two types of files: 1) PDF files containing documents and manuals, and 2) ZIP files containing software files (e.g., source code and executables). The Web server has one CPU and four identical disks. PDF files are stored on disks 1 and 2 in such a way that access to these files is evenly distributed between these two disks. Similarly, ZIP files are stored on disks 3 and 4 in a way that balances the load on these two disks. The main question of interest in this case study is: "What is the maximum number of concurrent PDF and ZIP file downloads that can be in progress in order to satisfy a certain prespecified SLA?"
The Web log contains one entry for each downloaded file, including its type and size. The worksheet Log
of the MS Excel WSData.XLS workbook includes 1,000 entries for file downloads captured over 200 seconds during a peak hour. A sample of the first six entries in this worksheet is given below:
File Type Size (KB) Elapsed Time (sec) PDF 303 1.43 ZIP 1233 5.81 ZIP 1077 5.08 PDF 315 1.48 ZIP 1240 5.84 PDF 413 1.95 . . . . . . . . .
The elapsed time column is the total time spent at the server to download the associated file. This time can be recorded in the HTTP log. For example, the elapsed time is captured in Microsoft's Internet Information Server (IIS) by selecting the "Time Taken" field in the Extended Logging Option.
< Day Day Up >
• Table of Contents
Performance by Design: Computer Capacity Planning by Example
By Daniel A. Menascé, Virgilio A.F. Almeida, Lawrence W. Dowdy
Publisher: Prentice Hall PTR Pub Date: January 05, 2004
ISBN: 0-13-090673-5 Pages: 552
Individual organizations and society as a whole could face major breakdowns if IT systems do not meet their Quality of Service (QoS) requirements on performance, availability, security, and maintainability. Corporations stand to lose valuable income, and public interests could be put at great risk. System designers and analysts usually do not take QoS requirements into account when designing and/or analyzing computer systems, due mainly to their lack of awareness about the issues that affect
performance and the lack of a framework to reason about performance. This book describes how to map real-life systems (e.g., databases, data centers, e-commerce applications) into analytic performance models. The authors elaborate upon these models, and use them to help the reader thoroughly analyze and better understand potential performance issues.
< Day Day Up > < Day Day Up >
6.3 Preliminary Analysis of the Workload
In order to obtain a first-cut analysis of the workload of this Web server, the entries in the Log worksheet are sorted by file type, then by file size, using Excel's Data Sort facility. The result is recorded in the
SortedLog worksheet of WSData.XLS. Then, the Descriptive Statistics facility of Excel (see Tools Data Analysis) is separately applied to the set of PDF and ZIP file entries to obtain basic statistics for these types of files.
PDF Statistics: Figure 6.1 shows the basic statistics for the 411 PDF files in the log (i.e., 41.1% of all downloaded files). As seen, the average size of a PDF file retrieved during the interval is 377.6 KB. The sample standard deviation is 43.1 KB. From these statistics, the coefficient of variation (i.e., the ratio between the standard deviation and the mean), CPDF, is computed as
This CPDF is relatively small and indicates that the set of all PDF files can be modeled as a single class in the ensuing performance model. [Note: A good rule-of-thumb is that if the coefficient of variation in a data set is less than 0.25, it is safe to assume the data set forms a single class. For higher values of the coefficient of variation, further data clustering (e.g., see Chapter 4) may be required.]