找出索引或结构以加速查询 - MS SQL Server 2008

时间:2013-10-10 14:23:23

标签: sql sql-server-2008 group-by

我需要帮助找出某些索引或某些不同的结构来加速我的查询,我先尝试分组,然后根据以下内容选择分组CTE的结果... http://codeidol.com/sql/sql-performance-tuning/GROUP-BY/Optimal-GROUP-BY-Clauses/

我最后只是擦了擦,然后以下是基本查询,花了2分钟,查询730,000条记录并且排序成本为91%,我假设它主要是针对分组。我真的需要这个至少1分钟。

查询:

SELECT
        tblFLS.fldFN,
        tblFLS.fldRN,
        tblFC.fldFC,
        tblFC.fldCity,
        tblFC.fldState,
        tblFLS.fldIN,
        tblUL.fldPRK,
        tblUL.fldLN,
        tblUA.fldANA,
        tblUA.fldMTH,
        (CASE
            WHEN (ISNULL([fldDTT], '') = '') THEN [fldDTT2]
            ELSE [fldDTT]
        END) AS fldDate,
        tblUA.fldPG
    FROM tblUA
    INNER JOIN tblUL ON tblUA.fldULKey = tblUL.fldULID
    INNER JOIN tblFLS ON tblUL.fldFileKey = tblFLS.fldFileID
    INNER JOIN tblFC ON tblFLS.fldFacKey = tblFC.fldFacID
    GROUP BY
        tblFLS.fldFN,
        tblFLS.fldRN,
        tblFC.fldFC,
        tblFC.fldCity,
        tblFC.fldState,
        tblFLS.fldIN,
        tblUL.fldPRK tblUL.fldLN,
        tblUA.fldANA,
        tblUA.fldMTH, 
        (CASE
            WHEN (ISNULL([fldDDT], '') = '') THEN [fldDDT2]
            ELSE [fldDDT]
        END),
        tblUA.fldPG

表格结构:

tblUA -- 850,000 rows, clustered Index on fldUAID

fldULK... INT
fldUAID... INT
fldALY... NVARCHAR(10)
fldAK... INT
fldMTH... NVARCHAR(15)
fldMK... INT
fldMMK... TINYINT
fldUNI... NVARCHAR(10)
fld7... INT
fld8... TINYINT
fld9...TINYINT
fld10...TINYINT

tblUL -- 200,000 rows, clustered Index on fldULID

fldULID... INT
fldFK... INT
fldPRK... INT
fld2... INT
fld3... INT
fldLN... NVARCHAR(15)
fldDTT... DATETIME
fldDTT2... DATETIME

tblFLS -- 70,000 rows, clustered Index on Primary Key, Non-Unique Non-Clustered On: fldDV, fldFK, fldIN, fldMK

    fldFN... NVARCHAR(255)
    fldRN... NVARCHAR(255)
    fldIN... NVARCHAR(100)
    fldDV... NVARCHAR(5)
    fldM... NVARCHAR(100)
    fldMK... INT
    fldINC... NVARCHAR(10)
    fldIDN... NVARCHAR(30)
    fldDTA... NVARCHAR(50)

    tblFC -- 10,000 records, clustered Index on Primary Key, Non-Unique Non-Clustered on fldFCID

    fldFCID... INT
    fldFC..NVARCHAR(100)
    fldADDR... NVARCHAR(50)
    fldADDR2... NVARCHAR(50)
    fldCity... NVARCHAR(50)
    fldState... NVARCHAR(10)

Execution Plan

1 个答案:

答案 0 :(得分:0)

如果您经常运行查询,则将按部分预先计算为帮助程序表

此外,您可以使用ISNULL([fldDDT], [fldDDT2])

简化计算