I know there is support for receiving multiple rows with one query but at what number of rows does this start to fall apart?
For example, say we hash some time ordered series data to avoid hotspotting. Since all of the rowkeys are hashed we can no longer perform a scan over a time range. I think, theoretically, we can calculate the hash for each data point and retrieve the rows that way. If there are a dozen data points I don't think there would be a problem but how does that change when we up that to a few dozen... a few hundred... and few thousand? I'm not quite sure of the performance implications of using such a method so I'm not quite sure at what point does this method of retrieving rows becomes impractical.
