Experiments with various data sets showed that any multi-dimensional range query on the cubetrees is very fast. For the 3GB data set of the demo, check below, the query response time on a cold start of a cubetree is anywhere between one thousandth to one third of a second per thousand aggregate points retrieved. On a hot tree, follow up queries, the time is anywhere from one half to one tenth of the cold start times.
But the best feature of the cubetree organization is its ability to do very efficient bulk incremental updates. The reduction of the cube update problem to sorting and merge-packing provides a scalable and industrial strength solution. We believe that this is the first one.