The spatial index for an SDE geodatabase feature class makes queries with spatial filters faster. Determining the optimum spatial index is an iterative process and depends on individual feature geometries that may vary from dataset to dataset. This document demonstrates how to determine a good starting point for a spatial index for a feature class whose feature geometries are relatively uniformly sized. This document uses SDE command line and SQL*Plus. The same process can be used with SQL Server Query Analyzer.
sdelayer -o si_stats -l <layer_name>,shape -i <instance> -u <username> -p <password>
select avg((((emaxx - eminx)+(emaxy-eminy))/2)*3) average_edge from<F_table>;
<F_table> is F<layerID>