Knowledge-driven Exploratory Performance Data Analysis for Execute-Order-Validate Blockchains

Abstract

Exploratory data analysis (EDA) of the performance characteristics of complex IT systems, such as enterprise blockchain solutions, would significantly benefit from explicit representations of, and inference on knowledge about the analyzed system. However, connecting EDA and knowledge representation is not part of the current practice. As a novel approach, this paper presents a generic hierarchical activity ontology, connected to Hyperledger Fabric experiments with end-to-end delay, endorsement delay, ordering delay, and block validation observations. On this basis, we present rules for inferring knowledge-based visualization declarations on this ontology. Lastly, we generate Jupyter notebooks for the inferred sequence of visualizations.