I see no problems of benchmark apart from you'd be avoiding table scans in prod for tables of those sizes - that kind of benchmark kind of pointless. Well, it s

I see no problems of benchmark apart from you'd be avoiding table scans in prod for tables of those sizes - that kind of benchmark kind of pointless. Well, it shows a point, but not something you would want to use anyway. Once data is cached, using indexed lookups are fast, 0.5ms.