Michael Gruberbauer

Hi, I am using HSQLDB as an in-memory DB to provide calculated statistical aggregate results for subsets of a large data set. Per request, I need to select and aggregate over around 5000 to 20000 rows out of tens of millions with about 10 columns (mostly double values) and numerically combine some values calculated from these column values. Data selection is fast due to indexes, but during calculation, my performance becomes exponentially worse the more requests I fire to the DB, even though this...