Question

I’ve noticed in our project the following use of a Common Table Expression to perform a simple select from a view:

WITH CTE_View_Alias AS
(
  SELECT 
    Id,
    Column1,
    Column2
  FROM VW_View
),
CTE_Foo AS
(
  SELECT
    B.Bar_Id,
    V.Column1
  FROM VW_Bar B
  INNER JOIN CTE_View_Alias V
    ON B.View_Id = V.Id
)

The CTE is only used in one other place, another CTE. The columns in the CTE are a subset of the view’s columns.

Is there a benefit to using this? Why not directly reference the view? Unfortunately, the originator of these sql statements is no longer with the company. To me, this reduces readability but I don’t want to eliminate it if there’s a reason my inexperience prevents me from seeing.

Answer 1

There’s no difference between:

WITH CTE_View_Alias AS (
 SELECT v.id,
        v.column1,
        v.column2
   FROM VW_View v)
SELECT t.*
  FROM CTE_View_Alias t

…and:

SELECT v.id,
       v.column1,
       v.column2
  FROM VW_View v

It’s plausible the view isn’t necessary either, but that’s not the question posed…

There is no performance difference between the previously mentioned options vs a derived table either:

 SELECT x.*
   FROM (SELECT v.id,
                v.column1,
                v.column2
           FROM VW_View v) x

The CTE is syntactic sugar for a derived table, if not using recursive functionality.

Conclusion

You’ll have to ask the author why they chose to use a CTE. It possible there were more complex operations that were removed & the CTE was left as-is in the fear of breaking something unforeseen. Or it’s just functionality they wanted to use…

I always fear that questions like these are too abstracted/simplified to demonstrate decisions.