Network-Controlled Dataset of Blockchain Benchmarks for Evaluating Repeatability and Performance Predictability

Dataset

Public research dataset associated with the ACM Distributed Ledger Technologies: Research and Practice journal article on topology-aware blockchain benchmarking.

Blockchain benchmarkingRepeatabilityPerformance predictabilityNetwork topologyReproducible research

Open dataset on Zenodo Open GitHub repository Associated ACM DLT article

Dataset overview

This dataset preserves aggregate statistics from repeated blockchain benchmark executions under controlled network conditions. It is available through its archival Zenodo record and a documented public GitHub repository.

Experimental matrix

The dataset contains 300 aggregate configuration rows and 50 columns. It covers five blockchain systems, five network topologies, six workloads, and deployments with 10 and 40 nodes.

Blockchains: Algorand, Diem, Ethereum, Quorum, and Solana.
Topologies: fat-tree, full mesh, hypercube, scale-free, and torus.
Workloads: DDoS, FIFA, GAFAM, Gaming, PayPal, and VISA.
Measurements: throughput, latency, energy, and committed transactions.

Repeatability and predictability

Each row summarizes repeated benchmark executions through counts, means, extrema, quartiles, standard deviations, absolute ranges, and relative-dispersion measures. The resulting evidence supports analysis of variability, repeatability, and performance predictability under controlled network conditions.

Reproducibility resources

The GitHub repository provides the released CSV, a data dictionary, methodology and provenance notes, citation metadata, and the CC BY 4.0 license. The experiments were conducted with Lilith, the topology-aware blockchain benchmarking framework.