This concise book offers plenty of concrete suggestions for improving real-world programs, but it also helps you think about performance as you design and test Java software. The authors first look at what "performance" really means. Beyond raw computational speed, this term can mean reducing RAM footprint, creating more responsive interfaces, and adding better scalability to programs so that they can handle more users and data.

A strength of this title is that the authors share their expertise, garnered from optimizing the Swing classes for Sun. (The tips for creating custom models and renderers in Swing will help your tables and other controls work with large datasets efficiently.) This text shows you how to benchmark and profile Java code and where to look for problem hot spots--and, once these are solved, where to go next for even better performance. You learn how to reduce object creation and class loading, which can allow your programs to load faster and consume less memory.

Also notable here is a solution for letting multiple Java programs share a single JVM for reduced memory overhead. The authors do a good job of dismantling the notion that using the Java Native Interface (JNI) is a quick fix for better performance. (The benchmarked code suggests that in almost every case native code is likely to be slower.) They also show how memory leaks still can exist in Java, and how to find them. The book closes with a guide to Java garbage collection and the latest on Sun's HotSpot Virtual Machine.

If anything, this book will convince you that good performance in Java isn't accidental; it takes planning, expertise, and plenty of testing. Also, it will get you thinking about performance in new ways with excellent strategies and tips that can help you write faster and more efficient Java code. --Richard Dragan

Topics covered:

• Measuring Java performance (computational speed, RAM footprint, startup time, scalability, perceived performance)
• Engineering performance into the software design process
• Benchmarking techniques (micro- and macro-benchmarks)
• Profiling
• Techniques to improve I/O performance
• Reducing RAM footprint (including minimizing object creation and detecting memory leaks)
• Controlling class loading (eager class loading and patterns for reducing the number of classes)
• Running multiple Java programs within the same virtual machine
• Optimizing String objects
• Object mutability and performance
• Choosing the right algorithm and data structure
• The Java Collection classes and arrays compared
• Four patterns for the Java Native Interface (JNI)
• Why native code is often slower
• Swing custom models and renderers for improved performance and scalability
• JDK 1.3 and Swing Timers for more responsive user interfaces
• Deployment and performance hints
• Tutorial on Java garbage collection
• The Java HotSpot Virtual Machine