Transactional feature analysis and synthetic workload generator for UTXO-based DLTs

Accumulation of blockchain data has led to a significant body of studies pertaining to block generation performance, financial assets, user interactions, etc. At the same time, surprisingly little attention has been dedicated to analyzing system characteristics of transactional workloads and to generating synthetic workloads. We argue that such analysis is imperative for multiple design and assessment goals. In this paper, we provide a comprehensive analysis of transaction features in the workload of three UTXO-based blockchain systems. The analysis covers 15 months and includes metrics such as input and output count, transaction size, temporal features, as well as their correlations and evolution over time. A few of the findings are quite surprising and have never been observed in the past. For example, we observe a shift in the distribution of output count over time. We also provide the first synthetic workload generator for UTXO transactions and release it as open source.