我想合并两张表，上次数据A和这次数据B一样的标记为N，A表有B表没有，就插入B表，标记为X，A表没有B表有，就标记为Y
新表就是A的表的X+B表的N和Y,并需要有标记

UPDATE tbl_old SET 变动=''
UPDATE tbl_new SET 变动=''
update c set c.变更='N' from tbl_new c,tbl_old d
where c.产品=d.产品 AND c.国家=d.国家 and c.商户=d.商户 and c.批次=d.批次  and (ABS(c.金额-d.金额))<=0.1  AND c.进货日期=d.进货日期 AND c.货币='RMB' AND d.货币='RMB'

update d set d.变更='N' from tbl_new c,tbl_old d
where c.产品=d.产品 AND c.国家=d.国家 and c.商户=d.商户 and c.批次=d.批次  and (ABS(c.金额-d.金额))<=0.1  AND c.进货日期=d.进货日期 AND c.货币='RMB' AND d.货币='RMB'

UPDATE tbl_old SET 变动='X' WHERE 变更<>'N'
UPDATE tbl_new SET 变动='Y' WHERE 变更<>'N'
insert into tbl select * from (select * from tbl_old where 变更='X' union all selelct * from tbl_new) a

数据量有几十万条，就很慢，我用merge好像也是很慢

MERGE INTO tbl_new c
USING tbl_old d
ON c.产品=d.产品 AND c.国家=d.国家 and c.商户=d.商户 and c.批次=d.批次  and (ABS(c.金额-d.金额))<=0.1  AND c.进货日期=d.进货日期 AND c.货币='RMB' AND d.货币='RMB'
WHEN MATCHED
THEN UPDATE SET c.变动='N'
WHEN NOT MATCHED 
THEN INSERT VALUES(d.[   ] ,d.产品,d.国家 ,d.商户,d.批次,d.进货日期,d.货币,d.金额,d.变动);

不知道大神有没有优化的方法，谢谢了

[此贴子已经被作者于2017-11-1 23:34编辑过]