When a collection of objects is retrieved by ObjectDB (e.g. by a query) ordinary referenced objects and collection of objects that have to be loaded eagerly are loaded in bulk for better performance.
However, inverse (mapped by) references that have to be loaded eagerly currently do not participate in a bulk load, which means that every reference requires running a separate query and doing a separate client-server round trip.
Future versions of ObjectDB should load eager mapped by (inverse) fields in bulk as well.
ObjectDB - Fast Object Database for Java (JPA/JDO)