Thread: Indexes and sorting

Are indexes useful for speeding up ORDER BY clauses?  Example:

    CREATE TABLE t
    (
      a INT,
      b INT,
      c INT,
      d INT
    );

    SELECT * FROM t WHERE a = 1 AND b = 2 AND c = 3 ORDER BY b;

Let's say the table just has one index:

    CREATE INDEX b_idx ON t (b);

In this case, obviously the b_idx will be used and no sorting after the fact
will be required.  Now let's add an index:

    CREATE INDEX key_idx ON t (a, b, c);

On the same query, now the key_idx will be used and there'll be a sort
wrapped around it all.  The question is, is the b_idx useful at all anymore?
Can it be used to speed up the sort step?  If so, how?  If not, why not?

-John

On Fri, 6 Feb 2004, John Siracusa wrote:

> Are indexes useful for speeding up ORDER BY clauses?  Example:
>
>     CREATE TABLE t
>     (
>       a INT,
>       b INT,
>       c INT,
>       d INT
>     );
>
>     SELECT * FROM t WHERE a = 1 AND b = 2 AND c = 3 ORDER BY b;
>
> Let's say the table just has one index:
>
>     CREATE INDEX b_idx ON t (b);
>
> In this case, obviously the b_idx will be used and no sorting after the fact
> will be required.  Now let's add an index:
>
>     CREATE INDEX key_idx ON t (a, b, c);

If you're really doing the above alot, you probably really want (b,a,c)
which can probably avoid the sort as well (unless of course you're also
doing frequent sorts on a, etc...)

> On the same query, now the key_idx will be used and there'll be a sort
> wrapped around it all.  The question is, is the b_idx useful at all anymore?

Yes.  Queries searching on just b won't use key_idx.

> Can it be used to speed up the sort step?  If so, how?  If not, why not?

Not really at least right now.

John Siracusa wrote:
>     SELECT * FROM t WHERE a = 1 AND b = 2 AND c = 3 ORDER BY b;
>
> Let's say the table just has one index:
>
>     CREATE INDEX b_idx ON t (b);
>
> In this case, obviously the b_idx will be used and no sorting after the fact
> will be required.  Now let's add an index:
>
>     CREATE INDEX key_idx ON t (a, b, c);
>
> On the same query, now the key_idx will be used and there'll be a sort
> wrapped around it all.  The question is, is the b_idx useful at all anymore?
> Can it be used to speed up the sort step?  If so, how?  If not, why not?

I believe that if you want to do ORDER BY b, then INDEX ON t(b) is
needed. Or ON t(b,c). If you want to ORDER BY a, then either INDEX ON
t(a) or ON t(a,b,c) or ON t(a,c) etc. is needed. See the manual on
CREATE INDEX for more details.

--
dave