Compressed StreamCube: Implementation of Compressed Data Cube in DSMS

Using data cube to analysis historical fact data online more faster than Ad-Hoc queries, but it need very large external storage. In DSMS (Data Stream Management System), due to capacity of memory is much smaller than disk, we meet even more problem in analyzing stream data by in-memory StreamCube. So, we compress StreamCube to gain more information about stream data in certain storage. We implement two compressed structures of StreamCube, StreamDwarf and StreamQCTree, base on state-of-the-art compressed data cube, Dwarf and QCtree. For example, we make a query cost model of StreamQCTree, and get that if we choose the proper cells to materialized, queries using a partial materialized cube costs a little more than the case of a fully materialized cube, the query response time is still kept close to StreamQCTree. Experiments show that compressed StreamCube performs well in tradeoff between storage space and queries response time.