Earth System Science (ESS) is a comprehensive interdisciplinary discipline, which originates from the study of global climate change and benefits from the progress of remote sensing technology. Now, ESS has entered the era of big data and artificial intelligence technology has played a key role in solving the frontier problems of ESS. Scientific data sharing is essential for the prosperity of science development and utilization of data value. After long-term exploration and practice, sound data management policies and mechanisms, continuous data sharing service system, and diversified scientific data integration modes have been established around the world. Innovative development of data sharing is going on thanks to the progress of sharing theory and ideas, such as the popular "findable, accessible, interoperable, and reusable" FAIR principle and data publishing. China has promulgated policies and regulations at the national level, focusing on promoting the development of national scientific data center, collection and management of scientific data resource from national science projects, as well as data publishing and protection. Combing the experience abroad and the actual situation in China, researchers have built the distinctive classification scheme for ESS data resources and major breakthroughs continuously appear in metadata management, distributed interoperability, big data analysis, scientific data sharing services, and other professional technologies, covering the whole life cycle from data collection, integration, analysis to open and sharing. Taking the National Earth System Science Data Center as an example, we summarize the progress of data sharing services and key technologies, and introduce the practice and achievements in China. At present, the national data sharing work in the field of ESS has contributed towards the formation of mature and stable operation mechanism, established a formal standard framework for multi-source distributed scientific data integration, developed multi-scale geoscience database covering diverse disciplines and themes, and built up distributed service networks and systems suitable for massive heterogeneous data sharing. This work not only promotes the development of Geoscience, but also fastens the dissemination and promotion of data sharing theory. However, issues such as isolated data islands, low-level generalization of service systems, and weak accords with international standards still hinder the advance of data sharing. In the future, with personalized needs for data sharing activated, customized "data + knowledge" services are expected to become the prevailing modes, which will bring new opportunities and challenges to data sharing.