Software Performance and Scalability (eBook)

Henry H. Liu, PhD, is Software Developer at BMC Software. Previously, he worked as a physicist in the national labs of China, France, Germany, and the United States. He also worked at Intel and Amdocs as a software performance engineer prior to joining BMC. He was an Alexander von Humboldt Research Fellow from 1990-1992 in Germany. He was awarded the Best Paper Award at CMG's 2004 conference in the category of software performance engineering. He is a certified Sun Enterprise Architect, IBM XML Developer, and Microsoft .NET Developer. Dr. Liu is most interested in applying his previous scientific research disciplines to solving software performance and scalability challenges.

Preface.

Acknowledgments.

Introduction: Performance versus Scalability.

PART 1 THE BASICS.

1 Hardware Platform.

1.1 Turning Machine.

1.2 von Neumann Machine.

1.3 Zuse Machine.

1.4 Intel Machine.

1.5 Sun Machine.

1.6 System Under Test.

1.7 Odds Against Turing.

1.8 Sizing Hardware.

1.9 Summary.

Recommended Reading.

Exercises.

2 Software Platform.

2.1 Software Stack.

2.2 APIs.

2.3 Multithreading.

2.4 Categorizing Software.

2.5 Enterprise Computing.

2.6 Summary.

Recommended Reading.

Exercises.

3 Testing Software Performance and Scalability.

3.1 Scope of Software Performance and Scalability Testing.

3.2 Software Development Performance.

3.3 Defining Software Performance.

3.4 Stochastic Nature of Software Performance Measurements.

3.5 Amdahl's Law.

3.6 Software Performance and Scalability Factors.

3.7 System Performance Counters.

3.8 Software Performance Data Principles.

3.9 Summary.

Recommended Reading.

Exercises.

PART 2 APPLYING QUEUING THEORY.

4 Introduction to Queuing Theory.

4.1 Queuing Concepts and Metrics.

4.2 Introduction to Probability Theory.

4.3 Applying Probability Theory to Queuing Systems.

4.4 Queuing Models for Networked Queuing Systems.

4.5 Summary.

Recommended Reading.

Exercises.

5 Case Study I: Queuing Theory to SOA.

5.1 Introduction to SOA.

5.2 XML Web Services.

5.3 The Analytical Model.

5.4 Service Demand.

5.5 MedRec Application.

5.6 MedRec Deployment and Test Scenario.

5.7 Test Results.

5.8 Comparing the Model with the Measurements.

5.9 Validity of the SOA Performance Model.

5.10 Summary.

Recommended Reading.

Exercises.

6 Caser Study II: Queuing Theory Applied to Optimizing and Tuning Software Performance and Scalability.

6.1 Analyzing Software Performance and Scalability.

6.2 Effective Optimization and Tuning Techniques.

6.3 Balanced Queuing System.

6.4 Summary.

Recommended Reading.

Exercises.

PART 3 APPLYING API PROFILING.

7 Defining API Profiling Framework.

7.1 Defense Lines Against Software Performance and Scalability Defects.

7.2 Software Program Execution Stack.

7.3 The PerfBasic API Profiling Framework.

7.4 Summary.

Exercises.

8 Enabling API Profiling Framework.

8.1 Overall Structure.

8.2 Global Parameters.

8.3 Main Logic.

8.4 Processing Files.

8.5 Enabling Profiling.

8.6 Processing Inner Classes.

8.7 Processing Comments.

8.8 Processing Methods Begin.

8.9 Processing Return Statements.

8.10 Processing Method End.

8.11 Processing Main Method.

8.12 Test Program.

8.13 Summary.

Recommended Reading.

Exercises.

9. Implementing API Profiling Framework.

9.1 Graphics Tool--dot.

9.2 Graphics Tool--ILOG.

9.3 Graphics Resolution.

9.4 Implementation.

9.5 Summary.

Exercises.

10 Case Study: Applying API Profiting to Solving Software Performance and Scalability Challenges.

10.1 Enabling API Profiling.

10.2 API Profiling with Standard Logs.

10.3 API Profiling with Custom Logs.

10.4 API Profiling with Combo Logs.

10.5 Applying API Profiling to Solving Performance and Scalability Problems.

10.6 Summary.

Exercises.

APPENDIX A STOCHASTIC EQUILIBRIUM AND ERGODICITY.

A.1 Basic Concepts.

A.2 Classification of Random Processes.

A.3 Discrete-Time Markov Chains.

A.4 Continuous-Time Markov Chains.

A.5 Stochastic Equilibrium and Ergodicity.

A.6 Birth-Death Chains.

APPENDIX B MEMORYLESS PROPERTY OF THE EXPONENTIAL DISTRIBUTION.

APPENDIX C M/M/1 QUEUES AT STEADY STATE.

C.1 Review of Birth-Death Chains.

C.2 Utilization and Throughput.

C.3 Average Queue Length in the System.

C.4 Average System Time.

C.5 Average Wait Time.