Based on the reductionism perspective， the current data governance models have many limitations， including potential constraints， unilateral dominance， and the absence of consensus， which makes it difficult to accommodate the extensive application of big data. The current synergy of multiple subjects needed for data governance can be regarded as an ecosystem. The corresponding concept to this is ecological governance， in which multiple subjects， based on systematic reflection and holistic consideration， aim to build the data governance along the path of "crowd-sourcing problems-- complementing capabilities--sharing benefits" through jointly bearing risks， transferring values and equitable interaction. This path makes up for the lack of adaptability and integrity of the reductionism model in data governance practice.