I keep going back to it. You also have some wild disparity in the other direction, estimated 500k rows, actual 55. I think with such completely out of wack statistics,...

By the way, being able to use 2017 plan comparison for these things was a life saver for quickly going through the plan.

It looks like straight up differences in row counts. The fast one is estimating 1 row, the slow one 278.402. In one operator I saw estimated number of rows 514,908...

Sounds like sampled stats might be resulting in better plans than full scan stats (based on the index rebuild). It's certainly possible. However, it could be a number of other...

Actually, you can't really do what you want. 10 rows from this table and 10 rows from that table, but, the 10 rows from the first table have a relationship...

You have to have a way to identify the information as belonging to one company or another. Then, using that method, you can export the data (assuming that's what you...

I don't know about the RBAC, but, if someone can view the DMVs, they can see parameter values passed for queries that are ad hoc. That can be a violation...

By design, SQL Server will take every bit of memory that you offer it. It's supposed to. It will grow into the memory allocation limit that you set and then...

It could be tons of things. Limiting the selection list allows the Optimizer to do simplification, eliminating unneeded tables within the views (you haven't specified if the views are of...