Indexing valid time databases via B+-trees

We present an approach, named MAP21, which uses standard B+-trees to provide efficient indexing of valid time ranges. The MAP21 approach is based on mapping one dimensional ranges to one dimensional points where the lexicographical order among the ranges is preserved. The proposed approach may employ more than one tree, each indexing a disjoint subset of the indexed data. When compared to the Time Index and the R*-tree we show that MAP21's performance is comparable to or better than those, depending on the type of query. In terms of storage, MAP21's structure was less than 10 percent larger than the R*-tree's and much smaller than the Time Index's. The main contribution of this paper though, is to show that standard B+-trees, available in virtually any DBMS, can be used to provide an efficient temporal index.