根据多字段去重

有重复数据主要有一下几种情况：

1.存在两条完全相同的纪录
      这是最简单的一种情况，用关键字distinct就可以去掉
			example： select distinct * from table(表名) where (条件)

2.存在部分字段相同的纪录（有主键id即唯一键）
    如果是这种情况的话用distinct是过滤不了的，这就要用到主键id的唯一性特点及group by分组
	  example: select * from table where id in (select max(id) from table group by [去除重复的字段名列表,....])

eg:

SELECT count(1)
FROM
TCHotelLog.dbo.ELongRequestErrorLog erel WITH (NOLOCK)
LEFT JOIN TCHotelResource.dbo.HotelInfo hi WITH (NOLOCK)
ON hi.Id = erel.HotelId
LEFT JOIN TCHotelResource.dbo.RoomType rt WITH (NOLOCK)
ON rt.Id = erel.RoomTypeId
LEFT JOIN TCHotelResource.dbo.HotelPolicy hp WITH (NOLOCK)
ON hp.Id = erel.PolicyId
WHERE
erel.ErrorTime >= '2014-6-13'
AND erel.RefId IN (9516406, 1646048)
AND erel.Id IN (SELECT max(Id)
FROM
TCHotelLog.dbo.ELongRequestErrorLog WITH (NOLOCK)
GROUP BY
MemberId
, PolicyId)

3.没有唯一键ID
   这种情况我觉得最复杂，目前我只会一种方法，有那位知道其他方法的可以留言，交流一下：
   example:select identity(int1,1) as id,* into newtable(临时表) from table
	select * from newtable where id in (select max(id) from newtable group by [去除重复的字段名列表,....])
           drop table newtable




这个也可以

WITH cte
AS ( SELECT ROW_NUMBER() OVER ( PARTITION BY t1.ParentId ORDER BY t1.ParentId ) rowNum ,
*
FROM [TCHotelResource].dbo.HotelDianping t1
WHERE t1.HotelId = 135
AND t1.OrderId != ''
AND t1.OrderId = '110513000016'
)
SELECT *
FROM cte
WHERE rowNum = 1