Question

select a,b,max(c) from myTable
where a>b
group by a,b

my first instinct is to open a clustered index for [a,b,c]

and here is the question :

suppose someone is running another simple select :

select a from myTable.

will the [a] column will be seeked throught the [a,b,c] index? or it wont use index at all…? (my guess it will use the [a,b,c] index)

but i guess it wont be fast as a single [a] index…

so what should i do ?

create [a,b,c] index
or
create [a,b,c] + [a] index

Answer 1

Your first thought is correct. The query:

SELECT a
FROM myTable

will be able to use the (a,b,c) index effectively. No need to add the (a) index.

---

As to whether this index should be clustered or not, that’s a whole other discussion. Since SQL-Server allows only one clustered index per table, the choice should be made wisely. See the MSDN articles: Clustered Index Design Guidelines and Creating Clustered Indexes.