Question

我有以下查询，想知道是否有更好的方法来调整它？

MERGE INTO target_table TARGET USING
(SELECT DISTINCT g1.column_name
FROM test_gtt1 g1,
test_gtt2 g2
WHERE g1.column_name = g2.column_name
AND g1.diff_id      IS NULL
AND g2.diff_id      IS NULL
UNION ALL
SELECT DISTINCT g1.column_name
FROM test_gtt1 g1,
test_gtt2 g2
WHERE 
g1.column_name  = g2.column_name
AND g1.diff_id = g2.diff_id
AND g1.diff_id IS NOT NULL
AND g2.diff_id IS NOT NULL
) SOURCE ON ( TARGET.column_name = SOURCE.column_name)
WHEN MATCHED THEN
UPDATE SET TARGET.column_name2 = 123;

Answer 1

这取决于数据类型，索引和生成的计划是什么（如果你提供了这个，那将会有所帮助。）

但是工会显然是多余的：

SELECT DISTINCT g1.column_name
FROM test_gtt1 g1,
test_gtt2 g2
WHERE g1.column_name = g2.column_name
AND ((g1.diff_id IS NULL
      AND g2.diff_id IS NULL)
     OR (g1.diff_id IS NOT NULL
         AND g2.diff_id IS NOT NULL))

<强>更新

我（在回复APC的评论时）发现使用....也可以得到相同的结果....

SELECT DISTINCT g1.column_name
FROM test_gtt1 g1,
test_gtt2 g2
WHERE g1.column_name = g2.column_name
AND DECODE(g1.diff_id, NULL, -1, 1)=DECODE(g2.diff_id, NULL, -1, 1);

Answer 2

如果您要将这两个集合与UNION合并，则无需测试diff_id is null和diff_id is not null：

       select * from foo where bar is null
       union all
       select * from foo where bar is not null

与

相同

       select * from foo

有没有更好的方法来调整此查询？

2 个答案: