[英]What could be the workaround to avoid the MERGE issue i.e. The target of a MERGE statement cannot be a remote table?
Copying data from one table to another both on different servers but similar structures.将数据从一个表复制到另一个在不同服务器上但结构相似的表。
Ended up on this.到此为止。
declare @ClassIds table (OldClassId int, NewClassId int);
merge into newDB.dbo.tblClasses as target
using
(
select
Id = Id * (-1),
[Name]
from
oldDB.dbo.tblClasses
)
as source on source.Id = target.Id
when not matched by target then
insert ([Name])
values (source.[Name])
output source.Id * (-1), inserted.Id -- ← the trick is here
into @ClassIds (OldClassId, NewClassId);
insert into newDB.dbo.tblStudents
select
s.Id,
s.[Name],
ClassId = ids.NewClassId
from
oldDB.dbo.tblStudents s
inner join @ClassIds ids on ids.OldClassId = s.ClassId;
but error:但错误:
The target of a MERGE statement cannot be a remote table, a remote view, or a view over remote tables. MERGE 语句的目标不能是远程表、远程视图或远程表上的视图。
Workaround could be reversing ie target and server but that's not ideal in my situation.解决方法可能是反转即目标和服务器,但这在我的情况下并不理想。
What should I do?我该怎么办?
Original question:原问题:
Reason to do this:这样做的原因:
the reason is I am copying the parent-child data and in the target the references to parent would be lost since the primary keys are auto generated hence in target a new record in parent would generate new Id but child would have the old parent id as of the source hence lost.原因是我正在复制父子数据,并且在目标中对父级的引用将丢失,因为主键是自动生成的,因此在目标中父级中的新记录将生成新的 Id 但子级将具有旧的父级 ID来源因此丢失。 So to avoid that the merge would make sure tyo update the child record with new parent ids.
因此,为了避免合并将确保使用新的父 ID 更新子记录。
edit:编辑:
the newDB is on the different server ie [192.168.xxx.xxx].newDB.dbo.tblStudents
newDB 在不同的服务器上,即
[192.168.xxx.xxx].newDB.dbo.tblStudents
If you are not able to change the remote DB structure, I would suggest to build the ClassId mapping table right in the target Class table:如果您无法更改远程数据库结构,我建议在目标类表中构建 ClassId 映射表:
drop table if exists #ClassIdMap;
create table #ClassIdMap (SourceClassId int, TargetClassId int);
declare @Prefix varchar(10) = 'MyClassId=';
insert into targetServer.targetDb.dbo.Classes
([Name])
select
-- insert the source class id values with some unique prefix
[Name] = concat(@Prefix, Id)
from
sourceServer.sourceDb.dbo.Classes;
-- then create the ClassId mapping table
-- getting the SourceClassId by from the target Name column
insert #ClassIdMap (
SourceClassId,
TargetClassId)
select
SourceClassId = replace([Name], @Prefix, ''),
TargetClassId = Id
from
targetServer.targetDb.dbo.Class
where
[Name] like @Prefix + '%';
-- replace the source Ids with the Name values
update target set
[Name] = source.[Name]
from
targetServer.targetDb.dbo.Class target
inner join #ClassIdMap map on map.TargetClassId = target.Id
inner join sourceServer.sourceDb.dbo.Classes source on source.Id = map.SourceClassId;
-- and use the ClassId mapping table
-- to insert Students into correct classes
insert into targetServer.targetDb.dbo.Students (
[Name] ,
ClassId )
select
s.[Name],
ClassId = map.TargetClassId
from
sourceServer.sourceDb.dbo.Students s
inner join #ClassIdMap map on map.SourceClassId = s.ClassId;
The problem or risk with this script is that it is not idempotent — being executed twice it creates the duplicates.此脚本的问题或风险在于它不是幂等的——执行两次会创建重复项。 To eliminate this risk, it is necessary to somehow remember on the source side what has already been inserted.
为了消除这种风险,有必要以某种方式在源端记住已经插入的内容。
声明:本站的技术帖子网页,遵循CC BY-SA 4.0协议,如果您需要转载,请注明本站网址或者原文地址。任何问题请咨询:yoyou2525@163.com.