Select Count take too long to be executed - sql

Question

I'm executing two queries. One that return the number of records and one that display the results. But how come the query that return the total of records takes 11 sec to be excecuted while the one displaying the result take less than 1 sec to be executed ? It should be the opposite, no ?
SQL that return the total of records is executed after 11 sec
(SELECT count(*) as id
FROM (
SELECT DISTINCT title, version
FROM book AS b
WHERE b.title IS NOT NULL AND NOT EXISTS (SELECT * FROM user AS u WHERE u.column1 = b.column1)
UNION ALL
SELECT DISTINCT title, version
FROM book2 AS b2
WHERE b.title IS NOT NULL AND EXISTS (SELECT * FROM user AS u WHERE u.column2 = b.column1)
) c )
SQL that display the result is executed in less than 1 sec
SELECT DISTINCT title, version
FROM book AS b
WHERE b.title IS NOT NULL AND NOT EXISTS (SELECT * FROM user AS u WHERE u.column1 = b.column1)
UNION ALL
SELECT DISTINCT title, version
FROM book2 AS b2
WHERE b.title IS NOT NULL AND EXISTS (SELECT * FROM user AS u WHERE u.column2 = b.column1)

Answer 1

You could try this:
SELECT COUNT(1)
FROM
(
SELECT DISTINCT title, version
FROM book b
LEFT JOIN u ON b.column1 = u.column2
LEFT JOIN u2 ON b.column1 = u2.column1
WHERE u2.column1 IS NULL
AND b.title IS NOT NULL
) sub

Answer 2

There are chances that you can improve your query here is one of the example and you can check other as well shared by friends
SELECT COUNT(1)
FROM book AS b
LEFT OUTER JOIN User U ON u.Column1 = b.column1
LEFT OUTER JOIN user u1 ON u1.Column2 = b.column1
WHERE b.title IS NOT NULL
AND u.Id IS NULL -- with assumption u might have another column id
AND u1.Id IS NULL -- with assumption u1 might have aother column or
GROUP BY title,Version

Answer 3

You are comparing apples and oranges. The count(*) version has to process all the data before it can return anything.
The more detailed version can start returning rows as they become available -- which is presumably when the first subquery starts returning rows.
I don't think either of these queries are what you really need, but this question does not provide that information.
Also, are book and version not unique? I wonder if this does what you want:
SELECT COUNT(*)
FROM book b
WHERE b.title IS NOT NULL AND
( NOT EXISTS (SELECT 1 FROM user u WHERE u.column1 = b.column1) OR
EXISTS (SELECT 1 FROM user u WHERE u.column2 = b.column1)
);
You can also use COUNT(DISTINCT):
SELECT COUNT(DISTINCT b.title + ':' + b.version)
. . .

Answer 4

(SELECT **count(title)** as id
FROM (
SELECT DISTINCT title, version
FROM book AS b
WHERE b.title IS NOT NULL AND NOT EXISTS (SELECT * FROM user AS u WHERE u.column1 = b.column1)
UNION ALL
SELECT DISTINCT title, version
FROM book2 AS b2
WHERE b.title IS NOT NULL AND EXISTS (SELECT * FROM user AS u WHERE u.column2 = b.column1)
) c )
"*" is never advisable as it accounts for all unnecessary details too.