如图所示，这将返回正确的计数，而使用元数据表的方法将只返回估计值。

    CREATE PROCEDURE ListTableRowCounts 
    AS 
    BEGIN 
        SET NOCOUNT ON 

        CREATE TABLE #TableCounts
        ( 
            TableName VARCHAR(500), 
            CountOf INT 
        ) 

        INSERT #TableCounts
            EXEC sp_msForEachTable 
                'SELECT PARSENAME(''?'', 1), 
                COUNT(*) FROM ? WITH (NOLOCK)' 

        SELECT TableName , CountOf 
            FROM #TableCounts
            ORDER BY TableName 

        DROP TABLE #TableCounts
    END
    GO

要在SQL Management Studio中获取该信息，右键单击数据库，然后选择Reports—> Standard Reports—> Disk Usage by Table。

USE DatabaseName
CREATE TABLE #counts
(
    table_name varchar(255),
    row_count int
)

EXEC sp_MSForEachTable @command1='INSERT #counts (table_name, row_count) SELECT ''?'', COUNT(*) FROM ?'
SELECT table_name, row_count FROM #counts ORDER BY table_name, row_count DESC
DROP TABLE #counts

接受的答案在Azure SQL上不适合我，这里有一个，它非常快，完全符合我的要求:

select t.name, s.row_count
from sys.tables t
join sys.dm_db_partition_stats s
  ON t.object_id = s.object_id
    and t.type_desc = 'USER_TABLE'
    and t.name not like '%dss%'
    and s.index_id = 1
order by s.row_count desc

SELECT 
    T.NAME AS 'TABLE NAME',
    P.[ROWS] AS 'NO OF ROWS'
FROM SYS.TABLES T 
INNER JOIN  SYS.PARTITIONS P ON T.OBJECT_ID=P.OBJECT_ID;

select T.object_id, T.name, I.indid, I.rows 
  from Sys.tables T 
  left join Sys.sysindexes I 
    on (I.id = T.object_id and (indid =1 or indid =0 ))
 where T.type='U'

这里indid=1表示群集索引，而indid=0表示堆索引