Reverse-engineering Distributed Applications to Understand Their Behaviour Reverse-engineering Distributed Applications to Understand Their Behaviour

Understanding the behaviour of distributed applications is a very challenging task, due to their complexity. To manage this complexity, the top-down use of suitable abstraction hierarchies is frequently proposed. Given the size of distributed applications, manually deriving such abstraction hierarchies seems not realistic. This paper discusses tools that automate the derivation of suitable abstraction hierarchies and reports on initial experience with these tools. These abstraction hierarchies enable a top{down approach to the behaviour understanding task, keeping the overall amount of information manageable. We modiied an existing prototype visualization tool to provide abstract visualizations of an execution. A user can navigate through the abstraction hierarchies derived with our tools, displaying an execution at various levels of abstraction. Examples of such abstract visualizations for the execution of one speciic distributed application are given. In general , the abstractions derived represent meaningful parts of the application: they can be interpreted in terms of the application domain.