Real-time applications are increasingly being built on top of open Java™ and Linux® environments. However, diagnosing real-time compliance for these applications against the time unreliability of underlying Java Virtual Machine (JVM) or operating system presents unique challenges. Many applications design their own instrumentation and application trace files that can be later analyzed for non-compliance issues. However, these files can become large and can be hard to analyze.

IBM TuningFork Visualization Tool for Real-Time Systems is a visualization and performance analysis tool that processes large event trace files that can be generated from IBM's Real-time JVM and Real-time Linux. It is an online, scriptable data visualization and analysis tool that supports the development and continuous monitoring of real-time systems. Support for C++ and recent versions of Linux is also included.

TuningFork allows various client programs to design custom data formats, new visualization and analysis components, and new export formats. TuningFork views allow the visualization of data in time scales from microseconds to minutes, enabling rapid understanding and analysis of system behavior. Although TuningFork was originally designed and tested for use with a particular real-time Java Virtual Machine, it was designed for extensibility by using the Eclipse plug-in architecture. Because the format is generic, instrumentation in other Java Virtual Machines or other run-time environments is possible.

How does it work?

By gathering information in a compact but extensible format from various components through existing tracing facilities and additional tracing facilities included with this technology, data exploration and performance analysis is possible. TuningFork enhances this task by coupling a powerful data-processing mechanism with a friendly and high-performing user interface.

This technology supports views for standard data exploration, such as time series and histograms, as well as views tailored for examination of timing behavior. Features such as online viewing of data and time synchronization of multiple views are designed with real-time monitoring needs in mind.

David F. Bacon, Ph.D., is a research staff member at IBM's T. J. Watson Research Center. There he leads the Metronome project, which produced the first hard, real-time, garbage-collected system. His algorithms are included in most compilers and run-time systems for modern object-oriented languages. His recent work focuses on high-level, real-time programming; embedded systems; data visualization; and programming language design. Dr. Bacon holds six patents, is a member of ACM and IEEE, and is on the governing boards of ACM SIGPLAN and SIGBED.

Perry Cheng, Ph.D., is a research staff member in the Dynamic Optimization Group at IBM's T. J. Watson Research Center. Dr. Cheng was a key member of IBM's Jalapeño project, and he has served on the progam committees of POPL, LCTES, ISMM, PASTE, SPACE, and JTRES.

Daniel Frampton, a graduate student, worked as a research intern with the Metronome project at IBM's T. J. Watson Research Center in 2005 and 2006. He has made significant contributions to the Jikes RVM project, is jointly responsible for the current design of the Memory Management Toolkit (MMTk), and has been part of the Jikes RVM core team since 2004.

David Grove, Ph.D., is a research staff member in the Dynamic Optimization Group at IBM's T. J. Watson Research Center. Dr. Grove was a key member of IBM's Jalapeño project; he holds two patents; and he is co-author of over 30 refereed journal and conference publications.