Efficient Indexing Of Temporal Databases Via B+-Trees

Temporal databases has been the object of research for more than two decades , and a wealth of material has been published. However, the majority of papers have dealt with modelling and query languages issues, relatively little research has addressed implementation issues. In this dissertation we investigate the topic of indexing temporal databases, focusing on valid time databases (where valid time is the dimension indexed) and bitemporal databases (where both valid time and transaction time are indexed). As the rst contribution of this dissertation, we propose a simple, yet eeect-ive, B +-tree based approach for indexing ranges, particularly valid time ranges. We call this approach MAP21. Our research shows that MAP21 has good performance while being feasibly implementable on top of facilities provided by existing commercial database management systems. The second contribution we present is a set of indexing structures for bitemporal databases, which also use the B +-tree as a basic framework. The main idea behind the design of such structure is the trade-oo between query processing time and size of the index. This trade-oo is analysed via analytic performance studies.