I haven't calculated the break-even point for bigquery vs scratchdb. It would be impossible to do so, as I'd need to know the number of rows bigquery will scan

I haven't calculated the break-even point for bigquery vs scratchdb. It would be impossible to do so, as I'd need to know the number of rows bigquery will scan in order to do an estimation in advance. Also what is a "slot-hour"? That is why I chose pricing uses units of "GB" and "hours" for storage and compute - those are things you can more easily observe. It is a good question, though, and perhaps I can do an experiment and write a blog post using example data showing the differences. I might be surprised at how efficient bigquery is!