SaifAlHarthi

What if the tables were partitioned, do you calculate local statistics or you aggregate the stats and expose it to the optimizer?

Yes, that code was painful indeed (was trying to port it at one time). On another note are there any plans on adding advisory tools for database tuning (such as AutoAdmin from MSR) or adding ML to self-tune depending on workload?

Great article!. While reading it I see some similarities with "Plan Stitching" from MSR as shown here https://www.microsoft.com/en-us/research/uploads/prod/2018/0.... It is interesting to see what the performance differences between the two. Also, I am curious of why build a query optimizer from scratch when you can adopt or port something like Orca?