Abstract: The effective manipulation methods of massive memory data in large scale e-trading oriented real-time database are studied, using linear pool to organize the memory data for physical storage, based on the update-in-place and append-only characteristics of trade transaction updating memory data. T-tree index is used for the logical manipulation of record-number-fixed data and entangled bidirectional linked-list index is adapted for the logical manipulation of dynamic data. The algorithm of physical memory allocation and buffer adjustment is simplified. The efficiency of data updating and index reconstruction is increased and the time complexity of memory data querying and accessing reduced. The results show that the time complexities of the two query methods, based on T-tree index and linked-list index, are not more than O(lnN) and O(ln(N+1)-1) respectively, and they fit for the need of large scale e-trading.