Toward Distributed, Pluggable Tools and Data: Re-Engineering a Data Analysis Architecture

An existing system for data analysis and visualization had the need to evolve to accommodate new analysis techniques and new application domains. However, its architecture was a significant barrier to realizing this need. We performed an architectural analysis of the system, which led us to re-engineer the system to use an architecture based on components and frameworks. What results is an architecture that supports plugins for new data analysis tools, new visualization techniques, and new data types and sources. This paper presents and evaluates the initial and the re-engineered architecture.