Techniques are disclosed hereafter for efficiently enforcing a “WITHOUT OVERLAP” range constraint by confirming primary-key integrity for a new or modified row (the “target row”) by checking just two neighboring index entries, using a new “two-sided halted range scan” of the same key index on entities which have range-endpoints data as part of their primary key. Techniques are described for reducing search time and resources in situations where a query specifies an entity and a point within a non-overlapping range. Techniques are also described for optimized handling of queries that do not specify a primary key but have both a range condition and a filter on a non-key column.