A visualization model for concurrent systems

Abstract

Concurrent systems maintain a distributed state and thus require coordination and synchronization between components to ensure consistency. To provide a coherent design approach to concurrent systems, recent work has employed an object-based methodology which emphasizes interactions through well-defined interfaces. The Actor model has provided formal reasoning about distributed object systems. Nonetheless, due to the complex interactions among components and the high volume of observable information produced, understanding and reasoning about concurrent algorithms in terms of simple interactions is a difficult task. Coordination patterns, which abstract over simple interactions, are not biased by low-level event orderings and are the appropriate mechanism for reasoning about concurrent algorithms. We outline a methodology for visualizing coordination patterns in concurrent algorithms which emphasizes observable interactions and causal connections between objects. We introduce visualizers as a linguistic mechanism for mapping coordination patterns to visualization. Visualizers are specified separately from algorithm code and thus respect code integrity. Moreover, visualizers may be implemented strictly in terms of object interfaces and thus preserve object encapsulation.