Latency performance modeling and analysis for hyperledger fabric blockchain network

Highlights：

• We present an analytical model to calculate the average transaction latency in single and multiple channels Fabric blockchains. Besides, we give a quantitative latency analysis of two typical consensus mechanisms in Fabric blockchain, i.e. Solo and PBFT consensus.

• We conduct a series of experiments to evaluate the validity of our proposed analytical model, and the results show that our model predicts the experimental results with an accuracy error of less than 6.1%.

• We discuss the general implications of our results for latency performance in a configurable Fabric blockchain system. Additionally, based on the analytical model, we identify performance bottlenecks and provide feasible insights.

摘要

•Based on the transaction flow of the Fabric blockchain, we decompose the transaction latency into three phases and each with several substeps for fine-grained analysis, and we indicate the configuration parameters that may affect the latency of each phase.•We present an analytical model to calculate the average transaction latency in single and multiple channels Fabric blockchains. Besides, we give a quantitative latency analysis of two typical consensus mechanisms in Fabric blockchain, i.e. Solo and PBFT consensus.•We conduct a series of experiments to evaluate the validity of our proposed analytical model, and the results show that our model predicts the experimental results with an accuracy error of less than 6.1%.•We discuss the general implications of our results for latency performance in a configurable Fabric blockchain system. Additionally, based on the analytical model, we identify performance bottlenecks and provide feasible insights.