The idea was that instead of loading objects one by one, you would trigger the load of n object (preemptively) So scrollableResults.get() would either: - extract the n following docs and load the n following objects and cache n-1 of them - read the data from the already preemptively loaded objets The object loading could be done by QueryLoader.load(EntityInfo... entityInfos) (if applicable), which is more efficient than the one by one object loading. On Jul 28, 2007, at 14:27, John Griffin wrote: