On BigQuery's roadmap / in beta is a recommendation of optimal # of slots given your team's usage. But it doesn't address an issue of reducing # of slots overall — essentially "flattening" slots usage and spreading it throughout the day.
We want to answer the question, “What would be an optimal usage of slots (optimal time in the day) for queries executed regularly?”
Could be interesting to know (A) how to reduce slots by slot schedules and (B) whether to purchase more slots and what the associated time savings would be
1. Take a profile of a company’s slots usage — hills (high slot usage) and dips (low slot usage) — to get monthly recurring savings on slots.
2. To have this, you can have an algorithm which predicts your usage of slots (or tries to) and filters out things that happen on-and-off. Jobs that that occur regularly would be our sweet spot in terms of optimisation.
3. Our value is telling users where to you move these queries around so that you can have a flatter daily slot usage pattern.
Paired with BigQuery's existing recommender, this would increase savings on slots.