Thanks for the workaround. Yeah, what you have in the code I submitted is an incomplete example of what I am porting. I have cases in which some of our services may need to operate on upwards of 25K docs at a time. In HS 5.5 I found that fetching that many docs in one shot to be extremely slow (this is all likelihood an artifact of Hibernate and not HS). So, instead I have a method implemented that returns the whole list, but pages in a chunk at a time internally to build the result set (limited to 25K). That sped things up dramatically. Once I saw HS 6 had this scrolling support, it occurred to me I can do that same thing without the downsides of paging, so I started implementing that when I ran into this issue. In the end, the final implementation will limit the max documents I scroll through. Having said all of this I should go back and test the performance of the one shot query again to see if I am still seeing the same behavior. Thanks again.