Dynamic Load System Identification Dataset for the Solana Containerised Testbed v0.8.0

This dataset contains controlled dynamic-load system-identification experiments for the Solana Containerised Testbed v0.8.0.

The dataset supports initial throughput-channel system identification for the relation:

u_cmd(t) -> u_ach(t)

where u_cmd(t) is the commanded transaction load and u_ach(t) is the achieved throughput.

The dataset includes five completed experiment classes:

M0: steady-state baselines;
M1: upward step-response experiments;
M2: step-down and recovery experiments;
M3: multi-step load profiles;
M4: stepwise sine-approximation / initial frequency-response experiment.

The confirmed safe identification range is:

lambda = 32..128

The dataset includes raw CSV telemetry, commanded and applied load profiles, per-run summaries, Kubernetes Job and Pod captures, preflight reports, aggregate dynamic-identification summaries, an M4 sine-response derived time series, artefact inventory, and SHA-256 checksums.

The aggregate v0.8.0 dynamic-identification summary reports that all M0-M4 experiment classes completed successfully, observed err_per_sec remained zero, throughput tracking was close to unity gain, step-like transitions settled within approximately one 5-second sample interval, and the M4 sine-approximation run showed near-unity low-frequency gain and high correlation between commanded and achieved throughput.

The dataset is intended for throughput-channel modelling and initial transfer-function/PID/MPC research preparation. It is not yet sufficient for latency-constrained MPC because lat_p99 is missing in the current collect_csv outputs.

Files:

solana-containerised-testbed-v0.8.0-dynamic-id-dataset.tar.gz
solana-containerised-testbed-v0.8.0-dynamic-id-dataset.tar.gz.sha256

Checksum:

dde1ec655b974dbeafce987b8616600b30ad7623659f7714f27c757d832f1d9b

Related software:

Solana Containerised Testbed v0.8.0: Dynamic Load System Identification
Software DOI: 10.5281/zenodo.20828274