can I get count() and rows from one sql query in sql server?

Question

I'd like to get the total count of results and top n rows of some query - is it possible in one statement?

I'd expect the results as:

count(..) column1        column2
125         some_value   some_value
125         some_value   some_value

Thank you in advance!

Answer 1

Like this:

SELECT TOP 100 --optional
    MC.Cnt, M.Column1, M.Column2
FROM
    myTable M
    CROSS JOIN
    (SELECT COUNT(*) AS Cnt FROM myTable) MC

Edit: After downvote and COUNT/OVER answer. A comparison on 2 tables of mine

You can see a huge difference between my CROSS JOIN/simple aggregate and a COUNT/empty ORDER BY clause

SELECT COUNT(*) OVER() AS C, key1col, key2col
FROM myTable

(24717 row(s) affected)

Table 'Worktable'. Scan count 3, logical reads 49865, physical reads 0, read-ahead reads 0, lob logical reads 0, lob physical reads 0, lob read-ahead reads 0.
Table 'myTable'. Scan count 1, logical reads 77, physical reads 0, read-ahead reads 0, lob logical reads 0, lob physical reads 0, lob read-ahead reads 0.

StmtText
  |--Nested Loops(Inner Join)
       |--Table Spool
       |    |--Segment
       |         |--Index Scan(OBJECT:([MyDB].[dbo].[myTable].[IX_useful]))
       |--Nested Loops(Inner Join, WHERE:((1)))
            |--Compute Scalar(DEFINE:([Expr1003]=CONVERT_IMPLICIT(int,[Expr1005],0)))
            |    |--Stream Aggregate(DEFINE:([Expr1005]=Count(*)))
            |         |--Table Spool
            |--Table Spool

SELECT
    MC.Cnt, M.key1col, M.key2col
FROM
    myTable M
    CROSS JOIN
    (SELECT COUNT(*) AS Cnt FROM myTable) MC

(24717 row(s) affected)

Table 'myTable'. Scan count 2, logical reads 154, physical reads 0, read-ahead reads 0, lob logical reads 0, lob physical reads 0, lob read-ahead reads 0.


StmtText
  |--Nested Loops(Inner Join)
       |--Compute Scalar(DEFINE:([Expr1005]=CONVERT_IMPLICIT(int,[Expr1009],0)))
       |    |--Stream Aggregate(DEFINE:([Expr1009]=Count(*)))
       |         |--Index Scan(OBJECT:([MyDB].[dbo].[myTable].[IX_useful]))
       |--Index Scan(OBJECT:([MyDB].[dbo].[myTable].[IX_useful] AS [M]))

I've repeated this on a table with 570k rows and here is the IO

Table 'Worktable'. Scan count 3, logical reads 1535456, physical reads 0, read-ahead reads 0, lob logical reads 0, lob physical reads 0, lob read-ahead reads 0.
Table 'myTable'. Scan count 1, logical reads 2929, physical reads 0, read-ahead reads 0, lob logical reads 0, lob physical reads 0, lob read-ahead reads 0.


Table 'myTable'. Scan count 34, logical reads 6438, physical reads 0, read-ahead reads 0, lob logical reads 0, lob physical reads 0, lob read-ahead reads 0.

Answer 2

what about

SELECT COUNT(*) OVER() AS C, COLUMN1, COLUMN2
FROM TABLE  

Regarding CROSS JOIN queries
In a heavy INSERT/DELETE environment, the cross join will return incorrect row count.

Try this from multiple connections
connection 1

set nocount on;
drop table dbo.test_table;
GO
create table dbo.test_table
(
    id_field uniqueidentifier not null default(newid()),
    filler char(2000) not null default('a')
);
GO
create unique clustered index idx_id_fld on dbo.test_table(id_field);
GO
while 1 = 1
insert into dbo.test_table default values;

connection 2

select T2.cnt, T1.id_field, T1.filler
from dbo.test_table T1
cross join (select COUNT(*) as cnt from dbo.test_table) T2

select T2.cnt, T1.id_field, T1.filler
from dbo.test_table T1
cross join (select COUNT(*) as cnt from dbo.test_table) T2

select T2.cnt, T1.id_field, T1.filler
from dbo.test_table T1
cross join (select COUNT(*) as cnt from dbo.test_table) T2

Each time, the count of records (@@ROWCOUNT) is different to T2.cnt

In the case of COUNT(*) OVER(), there is only a single table scan and the @@ROWCOUNT is always the same as T2.cnt

Regarding query plans - SQL 2005 SP3 appears to be much weaker at doing COUNT(*) OVER() than SQL 2008 R2. On top of that, it incorrectly reports query costs (I never thought a sub query could cost more than 100% of the entire query).

In a lot of scenarios, the cost of the COUNT(*) OVER() is between 50-75% of the CROSS JOIN

The best case scenario for a cross join would be if there was a very narrow index to do the count on. That way there will be a clustered index scan for the data + an index scan for the count.

As always, it's best to measure, measure, measure and go with the compromise that you're happy to live with.