Why can't the two be combined with a central system dedicated to storing data, then during operation, nodes are used, and if issues arise, we check the central data for comparison, also periodically comparing the central data with node data, taking the node data as the standard, while the central storage serves only as an auxiliary?