The current research work over data streams is mainly focused on dealing with the arrival of recent data in memory, neglecting the analysis and management of historical streaming data. An approach is proposed to store and query historical streaming data by using multi-layer recursive sampling method and HDS-Tree structure, which indexes the aggregation of historical streaming data and supports all kinds of aggregation queries over historical streaming data. The time-complexity and the error of aggregation algorithms are also analyzed based on HDS-Tree. The analytical and experimental results show that the approach can be effectively used to store and analyze the historical streaming data.