There is a big flaw in the article, you take two different statements with the same result but only by accident. The subselect ist quite the same as an outer join. The second statement uses an inner join which is in this example the same because ther is no employe without manager. But if there would be an employe without manager the statments would have different results. so it depends on what you try to accomplish. If you have two tables that could be inner joined, the inner join will be better than a subselect. But if you have to do an outer join, than the subselect will be mostly equal to the join. A third scenario could bethat you need an outer join AND more than one column from the subselect (multiple subselects), than the outer join will be much better than the subselect.

So there is a big difference in which scenario you are using subselects. in some scenarios there will be virtually no difference between both approches.

And if you are using oracle databases than all of this is the exact opposite than in mssql. 😉 I tried this example on my oracle db and found out that an subselect has only half the costs than the join even if i do multiple subselects.