I havent used SQL server much since the 2008 days but it used to happen quite often that the estimated and actual execution plan differed a lot. I think the est

I havent used SQL server much since the 2008 days but it used to happen quite often that the estimated and actual execution plan differed a lot. I think the estimated part also includes how many rows get returned from a subquery and that informs which joining algorithm is used. I second your longing for something more detailed than explain query plan in Sqlite though. In fact I complained about it here and some sqlite developer pointed me to these resources: https://news.ycombinator.com/item?id=30405275 https://news.ycombinator.com/item?id=30405275 I've never actually tried yet. Queries are still too fast to do all that:)