[glogla]

You are correct about clustered indexes (1) if and only if the table clustering is selected well. Which is not always the case - especially with Azure SQL which doesn't allow for non-clustered tables or tables clustered by anything other than primary key.

Table clustering is mostly betting on "I think most access is going to happen using this columns" which might be good choice or it might not, but it's not something that should happen implicitly or by the only choice at all.

I think Markus (2) explains this much better than I do, so I'll just link his text.

(1): which is pretty stupid name, btw - Oracle call it "index organized tables" which is much clearer way to describe the concept.

(2): http://use-the-index-luke.com/sql/clustering/index-organized...