Execution patterns for visualizing web services

Abstract

Web Services are well on their way to becoming the Lingua Franca for distributed computing. Although tools for building and monitoring web services applications are more powerful and easier to use than ever, they do not yet fully address the horizontal complexity of mature applications built as large nets of interconnected web services. We present a pattern-based visualization that enables business owners, application designers, programmers, and operations staff to quickly understand the behavior of complex web services applications. We describe a novel pattern extraction algorithm that captures important trends from web services execution traces. We demonstrate a new way to visualize these patterns that shows the behavior of web services applications at different levels of abstraction. Finally, we explain how this can help developers with performance analysis by showing both the averages and variations in the data contained in each pattern.