When selecting a group-by attribute that gives a very large number of key values (e.g. startTime), the histogram table will have a very large number of rows and be unreasonably slow (e.g. when changing the sort order). In worst case this could be handled with a hard limit (same as for item lists, see related JMC-4279). More helpful would be to combine 'similar' keys in groups (e.g. 3kB-6kB, see related JMC-3554 (or even 'hat'-'have')).