Spatial database access methods

In the discussion of research issues in spatial databases (SIGMOD Record vol. 19, no. 4, Dec 1990) we stated the need for a robust framework for analytical comparison of a broad range of spatial access methods. The utility of such a comparison, even of very closely related access methods, was shown in [FALO87]. A necessary precondition for a meaningful analytical comparison is the existence of strong analytical results for individual access methods. In the following paper, Salzberg and Lomet take the worst case analytical results on fan-out and average storage utilization they obtained for their hB-tree [LOME89,LOME90] and extend the analysis to another robust method, Z-order encoding [OREN84]. We think this paper is a start on the comparative assessment of access methods based on analytical results. We hope to see future work extend the framework beyond worst case analysis, and to other access methods as well.