An improved AG-Tree based on column store

Data storage has become a hot study issue in data warehouse field. A good index structure can promote the system performance evidently, and hence the study on how to store data in data warehouse is a useful and significant work. In this paper we firstly analyzed the features of queries in the data warehouse. Based on these features, we then explored some problems within the AG-Tree based on the row store. Then, we proposed the AG-Tree based on column store which combine the idea of column store and AG-Tree. It can be used to support the partial dimensional range query, all dimensional range queries and the aggregate query efficiently. And it can avoid accessing irrelevant data, while the access may decrease the I/O frequencies. Formal definitions of multi-dimensional space and range query are given in our work. It is shown that the system based on this index structure can give a better performance by the time and space complexity analysis.