Getting a SQL Server RowCount Without doing a Table Scan

来源:互联网 发布:日语单词背诵软件 编辑:程序博客网 时间:2024/06/03 12:53

Written By: Andy Novick -- 8/21/2006


Problem
Sometimes there is a need to get record counts from every table in your database.  One way of doing this is to do a SELECT count(*) on all of your tables, but this could create a lot of overhead especially for large databases and large tables. If you don't require an exact answer, it isn't necessary to use a SELECT count(*) query on the rows in a table to get the row count.

Solution
Thanks to Andrew Novick at Novick Software here is the answer.  SQL Server keeps the row count in sysindexes and it can be retrieved there. The key is to select the correct record from sysindexes. Sysindexes is a system table that exists in every database.

SQL Server maintains at least one row in sysindexes for every user table. A few of the most important columns are:
Column         Data Type         Description
id         int         ID of the table referred to by this row 
indid         int         See the text that follows...
rowcnt         bigint         Number of rows in the index

The indid column tells us what part of the table structure this row of sysindexes is referring to: 
indid value         Description
0         Table data when there is no clustered index
1         Refers to the clustered index
2 - 254         Non-clustered indexes
255         Text or Image data pages

A table will only have an entry in sysindexes with an indid value of for 0 or 1, never both.  That's the entry that we're interested in because its rowcnt field gives us the number of rows in the table.  Here is a query that shows the table, index and indid from the pubs database:
USE pubs
GO

SELECT so.[name] as [table name]
     , CASE WHEN si.indid between 1 and 254
            THEN si.[name] ELSE NULL END
            AS [Index Name]
     , si.indid
     FROM sysindexes si
          INNER JOIN sysobjects so
             ON si.id = so.id
     WHERE si.indid < 2
       AND so.type = 'U' -- Only User Tables
       AND so.[name] != 'dtproperties'
     ORDER BY so.[name]

Here are the results:
table name         Index Name         indid 
authors         UPKCL_auidind         1
discounts         NULL         0
employee         employee_ind         1
jobs         PK__jobs__117F9D94         1
pub_info         UPKCL_pubinfo         1
publishers         UPKCL_pubind         1
roysched         NULL         0
sales         UPKCL_sales         1
stores         UPK_storeid         1
titleauthor         UPKCL_taind         1
titles         UPKCL_titleidind         1

As you can see from the results, most of the indexes are clustered (indid=1) but a few tables such as discounts don't have a clustered index (indid=0).

I started this tip with "If you don't need an exact answer..."  That's because there are times when rowcnt is not the exact number of records in the table. 

This can be corrected by updating statistics on the table with:
dbcc updateusage
go

Here's the CREATE FUNCTION script for udf_Tbl_RowCOUNT
CREATE FUNCTION dbo.udf_Tbl_RowCOUNT (

        @sTableName sysname  -- Table to retrieve Row Count
        )

    RETURNS INT -- Row count of the table, NULL if not found.

/*
* Returns the row count for a table by examining sysindexes.
* This function must be run in the same database as the table.
*
* Common Usage:  
SELECT dbo.udf_Tbl_RowCOUNT ('')

* Test  
PRINT 'Test 1 Bad table ' + CASE WHEN SELECT
       dbo.udf_Tbl_RowCOUNT ('foobar') is NULL
        THEN 'Worked' ELSE 'Error' END
       
* © Copyright 2002 Andrew Novick http://www.NovickSoftware.com
* You may use this function in any of your SQL Server databases
* including databases that you sell, so long as they contain
* other unrelated database objects. You may not publish this
* UDF either in print or electronically.
***************************************************************/

AS BEGIN
   
    DECLARE @nRowCount INT -- the rows
    DECLARE @nObjectID int -- Object ID

    SET @nObjectID = OBJECT_ID(@sTableName)

    -- Object might not be found
    IF @nObjectID is null RETURN NULL

    SELECT TOP 1 @nRowCount = rows
        FROM sysindexes
        WHERE id = @nObjectID AND indid < 2

    RETURN @nRowCount
END
GO

GRANT  EXECUTE  ON [dbo].[udf_Tbl_RowCOUNT]  TO PUBLIC
GO


Let's use it:
use pubs -- assuming the UDF was created in pubs
go

SELECT [name]
     , dbo.udf_Tbl_RowCOUNT ([name]) as [Row Count]
    FROM sysobjects
    WHERE type='U' and name != 'dtproperties'
    ORDER BY [name]
GO

Here are the results:
name         Row Count
authors         24
discounts         3
employee         43
jobs          14
pub_info         8
publishers         8
roysched         86
sales         21
stores         6
titleauthor         25
titles         18
Next Steps