Identifying bottleneck onset in a private Solana testbed

Abstract

This paper examines the onset of bottleneck behaviour in a private Solana testbed under synthetic transaction load. The study focuses on end-to-end observable metrics, including commanded load, achieved throughput, latency behaviour and the saturation ratio between sustained output and offered input. A grey-box interpretation is used to distinguish four empirical operating regimes: low load, mid load, knee and high load. The results show that while the system remains close to proportional scaling at lower and moderate transaction rates, efficiency begins to degrade near the knee region. At higher load, throughput no longer tracks the commanded rate proportionally, indicating saturation-dominated behaviour. The proposed approach demonstrates that bottleneck onset can be identified without full internal visibility into validator mechanisms, providing a practical and reproducible basis for benchmarking, diagnostics and capacity planning in private Solana-based blockchain deployments.