優化代碼運行速度更快，效率更高

我試圖找到一種方法來優化代碼以加快運行速度。我有很多選擇語句（大約30多個選擇查詢），然後我使用UNION ALL來連接所有這些選擇語句。我知道這是一個可怕的方式。我該如何改進它？這裏的鏈接，如果你想看到的結果：http://sqlfiddle.com/#!2/db3e7b/1 優化代碼運行速度更快，效率更高

這裏是我的代碼示例：

SELECT 'east/flw' AS LOCATION, 
    ROUND(sum(CASE WHEN t2.locid = '2815' THEN t2.value ELSE 0 END), 2) AS Reading, 
    ROUND(sum(CASE WHEN t2.locid = '2620' THEN t2.value ELSE 0 END), 2) AS Flw, 
    ROUND(sum(CASE WHEN t2.locid = '2618' THEN t2.value ELSE 0 END), 2) AS Prs, 
    ROUND(sum(CASE WHEN t2.locid = '2595' THEN t2.value ELSE 0 END), 2) AS Lvl 

FROM table2 t2 

INNER JOIN table1 t1 
    ON t1.id = t2.locid 
WHERE t2.t_stamp = (SELECT MAX(t2.t_stamp) 
        FROM table2 t2 
        WHERE t1.id = t2.locid) 
GROUP BY 'east/flw' 

UNION ALL 

SELECT 'west/flw' AS LOCATION, 
    ROUND(sum(CASE WHEN t2.locid = '2715' THEN t2.value ELSE 0 END), 2) AS Reading, 
    ROUND(sum(CASE WHEN t2.locid = '2720' THEN t2.value ELSE 0 END), 2) AS Flw, 
    ROUND(sum(CASE WHEN t2.locid = '2718' THEN t2.value ELSE 0 END), 2) AS Prs, 
    ROUND(sum(CASE WHEN t2.locid = '2795' THEN t2.value ELSE 0 END), 2) AS Lvl 

FROM table2 t2 

INNER JOIN table1 t1 
    ON t1.id = t2.locid 
WHERE t2.t_stamp = (SELECT MAX(t2.t_stamp) 
       FROM table2 t2 
       WHERE t1.id = t2.locid) 

GROUP BY 'west/flw' 
UNION ALL 
. 
. 
. 
UNION ALL 
. 
. 
.

來源

2013-11-22 sleepsleepsleep90731

表格說明??解釋命令？索引？ PK和FK？ – jcho360

該表位於提供的鏈接中。並且我無法對我的實際數據庫上的表進行任何更改 – sleepsleepsleep90731

如果無法修改表格，我不會在sqlfiddle中查看任何PK，FK，UK，index，etx，您希望如何加快速度一個簡單的查詢？ – jcho360

我會試着扔掉所有UNION ALL和讀表一次，
這樣：

SELECT 
    CASE 
     WHEN t2.locid IN ('2815', '2620', '2618', '2595')   
     THEN 'east/flw' 
     WHEN t2.locid IN ('2715', '2720', '2718', '2795') 
     THEN 'west/flw' 
     ....... 
     ....... 
     ....... 
     ....... 
    END AS LOCATION, 

    ROUND(sum(CASE WHEN t2.locid IN ('2815', '2715', .............) 
       THEN t2.value ELSE 0 END), 2) AS Reading, 
    ROUND(sum(CASE WHEN t2.locid IN ('2620', '2720', ...............) 
       THEN t2.value ELSE 0 END), 2) AS Flw, 
    ROUND(sum(CASE WHEN t2.locid IN ('2618', '2718', ...............) 
       THEN t2.value ELSE 0 END), 2) AS Prs, 
    ROUND(sum(CASE WHEN t2.locid IN ('2595', '2795', ...............) 
       THEN t2.value ELSE 0 END), 2) AS Lvl 
FROM table2 t2 
INNER JOIN table1 t1 
    ON t1.id = t2.locid 
WHERE t2.t_stamp = (SELECT MAX(t2.t_stamp) 
        FROM table2 t2 
        WHERE t1.id = t2.locid) 
GROUP BY location

來源

2013-11-22 23:44:55 krokodilko

優化代碼運行速度更快，效率更高

回答

相關問題