毫秒级百万数据分页存储过程[欢迎转载]
数据分页一直是一个经久不衰的话题,在园子里看到了N多高手,都比我厉害,
今天把我使用的分页代码继续分享;请高手手下留情,少拍砖啦...哈哈;
欢迎大家转载!!!!
首先参考园子里的高手方法生成一个百万数据表; 耗时 33 秒;
然后创建分页存储过程;这个是基于SQL 2005 的ROW_NUMBER的, SQL 2000 不适用; SQL 2008 我没有安装,所以也没有测试过!估计应该可以运行.
2种算法可以选择 1. 使用 BETWEEN ; 2. TOP ; 代码中已经注释了.
我测试了一下2种方法基本没有什么区别;代码如下:
USE [DATA_TEMP] GO /****** 对象: StoredProcedure [dbo].[im531_Page] 脚本日期: 11/26/2010 10:52:35 ******/ SET ANSI_NULLS ON GO SET QUOTED_IDENTIFIER ON GO -- ============================================= -- Author: im531 -- Create date: 2010-10-25 -- Description: SQL 2005 ROW_NUMBER 分页 -- Modify Date: 2010-10-26 -- ============================================= CREATE PROCEDURE [dbo].[im531_Page] @TabeName AS NVARCHAR(50), @Fields AS NVARCHAR(1024), @SearchWhere AS NVARCHAR(1024), @OrderFields AS NVARCHAR(1024), @pageNumber AS INT, @page AS INT AS BEGIN SET NOCOUNT ON; DECLARE @sqlType AS TINYINT SET @sqlType = 0 -- 0 BETWEEN 1 TOP DECLARE @sql AS NVARCHAR(MAX) IF @SearchWhere <> '' SET @SearchWhere = ' WHERE ' + @SearchWhere IF @page < 2 SET @page = 1 IF @page = 1 BEGIN SET @sql = 'SELECT TOP ' + CONVERT(NVARCHAR(20),@pageNumber) + ' ' + @Fields + 'FROM [' + @TabeName + '][a] WITH(NOLOCK) ' + @SearchWhere + ' ORDER BY ' + @OrderFields END ELSE BEGIN --临时表 SET @sql = ';WITH [Page_____Table] AS(' + 'SELECT ROW_NUMBER() OVER(ORDER BY ' + @OrderFields + ') AS [RowNow],' + @Fields + 'FROM ['+ @TabeName +' ][a] WITH(NOLOCK) ' + @SearchWhere + ')' --查找当前页面记录 IF @sqlType = 0 BEGIN -- A USE BETWEEN SET @sql = @sql + 'SELECT *' + 'FROM [Page_____Table] WITH(NOLOCK)' + 'WHERE [RowNow] BETWEEN ' + CONVERT(NVARCHAR(20),(@page - 1) * @pageNumber + 1) + ' AND ' + CONVERT(NVARCHAR(20),@page * @pageNumber) + 'ORDER BY [RowNow]' END ELSE BEGIN -- B USE TOP SET @sql = @sql + 'SELECT TOP ' + CONVERT(NVARCHAR(20),@pageNumber) + ' *' + 'FROM [Page_____Table] WITH(NOLOCK)' + 'WHERE [RowNow] > ' + CONVERT(NVARCHAR(20),(@page - 1) * @pageNumber) + 'ORDER BY [RowNow]' END END EXEC(@sql) END
然后测试分页性能;每页显示20条记录, 百万数据最大50000页,当前页采用随机数取值;
DECLARE @d DATETIME SET @d = GETDATE() DECLARE @M AS INT;DECLARE @N AS INT;DECLARE @cPage AS INT SET @M = 50000;SET @N = 1 SET @cPage = CAST((rand()*(@M-@N)+@N) AS INT) EXECUTE [im531_Page] @TabeName = 'Table_1', @Fields = '[a],[b]', @SearchWhere = '[a] > 0', @OrderFields = '[a] DESC', @pageNumber = 20, @page = @cPage SELECT [当前页] = @cPage,[分页执行耗时间(毫秒)] = DATEDIFF(ms,@d,GETDATE())
在我的机器上面测试均为毫秒级!!!!
园子里的高手若有更好的方法欢迎分享一下!!!!!!!!!!!!!!!
--------------------------------------- 我是华丽的分割线 ---------------------------------------
回复10楼园友 ★金★ :
加上数据统计结果后稍慢了一丁点儿,估计在实际运用中可以忽略不计;