Abstract: The current "data swamp" problem has become increasingly prominent. This article combined the application of railway big data and proposed a new top-level design from the storage architecture of the railway data service platform. The design included three parts: storage component partition management module, full-link storage component monitoring module, and non-inductive storage component access middleware. The storage component partition management module flexibly stored the data in cost-effective and highly reliable storage components according to the data type, frequency of use, and application scenarios of the data. The full-link storage component monitoring module provided the data flow process throughout the life cycle, real-time monitoring of the storage status of various types of data in storage components. Non-sense storage component access middleware provided non-sense, cross-component data access services. Finally, combined with the application examples of the railway data service platform, the article made a more intuitive description of the data storage, management and access in the platform, so that the "data swamp" problem could be solved in the railway data service platform.