High-Level Views of Distributed Executions: Convex Abstract Events

作者:Thomas Kunz

摘要

Due to the complexity of distributed applications, understanding their behaviour is a challenging task. To remedy this problem, graphical visualizations of distributed executions in the form of process-time diagrams are frequently employed. However, such process-time diagrams do not scale for long-running and complex distributed applications. To reduce the display complexity, abstract graphical views of an execution are frequently suggested. One commonly used abstraction is to group primitive events into abstract events. This paper discusses some of the problems encountered when analyzing executions at abstract levels and introduces the concept of convex abstract events. Such abstract events can be used to reason about executions at higher levels, facilitating program development, debugging, and maintenance. We discuss some fundamental aspects, such as the precedence relation and its efficient detection, for abstract events. The paper also presents a graphical representation for convex abstract events which can easily be included in process-time diagrams. Poet, our visualization tool, was enhanced with a facility to display abstract events. Using a non-trivial distributed application, examples of the resulting abstract execution visualizations are discussed.

论文关键词:Program understanding, distributed debugging, visualization, Hermes, program behavior, abstraction

论文评审过程:

论文官网地址:https://doi.org/10.1023/A:1008685117587