如何对比两个数据库差异

构建两个临时表，将两个数据库结构信息导入。

create Table #t1

(

ID Int Identity(1,1) Not Null Primary Key,

tablename nvarchar(50) NULL,

columnName nvarchar(50) NULL,

columnIndex int null,

columnType nvarchar(50) NULL

)

use 数据库1

insert into #t1

create Table #t2

(

ID Int Identity(1,1) Not Null Primary Key,

tablename nvarchar(50) NULL,

columnName nvarchar(50) NULL,

columnIndex int null,

columnType nvarchar(50) NULL

)

//开始比较

use 数据库2

insert into #t2

SELECT

SO.name as '表名',

SC.name as '表列名',

SC.colid as '索引',

ST.name as '类型'

FROM

sysobjects SO, -- 对象表

syscolumns SC, -- 列名表

systypes ST -- 数据类型表

WHERE

SO.id = SC.id

AND SO.xtype = 'U'-- 类型U表示表，V表示视图

AND SO.status >= 0 --加一个条件：SO.status >= 0，否则会将系统的临时表显示出来

AND SC.xtype = ST.xusertype

ORDER BY

SO.name, SC.colorder

go

//查询出 在t1 里有， t2 里没有的字段，查询列出来。

select * from

(

select tablename,columnName,columnType from #t1 where tablename like '%EMS_%'

EXCEPT

select tablename,columnName,columnType from #t2 where tablename like '%EMS_%'

) as c

order by tablename

比较两个数据库中表和字段的差异

-- 比较两个数据库中表的差异

-- u表，p存储过程,v视图

-- INTFSIMSNEW新库，INTFSIMS旧库

SELECT NTABLE = A.NAME, OTABLE = B.NAME

FROM INTFSIMSNEW..SYSOBJECTS A

LEFT JOIN INTFSIMS..SYSOBJECTS B

ON A.NAME = B.NAME

WHERE ISNULL(B.NAME, '') = ''

AND A.XTYPE = 'U'

UNION ALL

SELECT NTABLE = B.NAME, OTABLE = A.NAME

FROM INTFSIMS..SYSOBJECTS A

LEFT JOIN INTFSIMSNEW..SYSOBJECTS B

ON A.NAME = B.NAME

WHERE ISNULL(B.NAME, '') = ''

AND A.XTYPE = 'U'

ORDER BY 1, 2

-- 比较两个数据库中每个表字段的差异

SELECT

表名A = CASE WHEN ISNULL(A.TABLENAME, '') <>'' THEN A.TABLENAME ELSE B.TABLENAME END,

字段名A = A.FIELDNAME,

字段名B = B.FIELDNAME,

顺序= A.FIELDSNO,

说明= CASE WHEN A.FIELDTYPE <>B.FIELDTYPE THEN '类型: ' + A.FIELDTYPE + '-->' + B.FIELDTYPE

WHEN A.FIELDSNO <>B.FIELDSNO THEN '顺序: ' + str(A.FIELDSNO) + '-->' + str(B.FIELDSNO)

WHEN A.LENGTH <>B.LENGTH THEN '长度: ' + str(A.LENGTH) + '-->' + str(B.LENGTH)

WHEN A.LENSEC <>B.LENSEC THEN '小数位: ' + str(A.LENSEC) + '-->' + str(B.LENSEC)

WHEN A.ALLOWNULL <>B.ALLOWNULL THEN '允许空值: ' + str(A.ALLOWNULL) + '-->' + str(B.ALLOWNULL)

END

FROM (SELECT

TABLENAME = B.NAME,

FIELDNAME = A.NAME,

FIELDSNO = A.COLID,

FIELDTYPE = C.NAME,

LENGTH = A.LENGTH,

LENSEC = A.XSCALE,

ALLOWNULL = A.ISNULLABLE

FROM INTFSIMSNEW..SYSCOLUMNS A

LEFT JOIN INTFSIMSNEW..SYSOBJECTS B

ON A.ID = B.ID

LEFT JOIN INTFSIMSNEW..SYSTYPES C

ON A.XUSERTYPE = C.XUSERTYPE

WHERE B.XTYPE = 'U') A

FULL JOIN (SELECT

TABLENAME = B.NAME,

FIELDNAME = A.NAME,

FIELDSNO = A.COLID,

FIELDTYPE = C.NAME,

LENGTH = A.LENGTH,

LENSEC = A.XSCALE,

ALLOWNULL = A.ISNULLABLE

FROM INTFSIMS..SYSCOLUMNS A

LEFT JOIN INTFSIMS..SYSOBJECTS B

ON A.ID = B.ID

LEFT JOIN INTFSIMS..SYSTYPES C

ON A.XUSERTYPE = C.XUSERTYPE

WHERE B.XTYPE = 'U') B

ON A.TABLENAME = B.TABLENAME

AND A.FIELDNAME = B.FIELDNAME

WHERE ISNULL(A.TABLENAME, '') = ''

OR ISNULL(B.TABLENAME, '') = ''

OR A.FIELDTYPE <>B.FIELDTYPE

OR A.FIELDSNO <>B.FIELDSNO

OR A.LENGTH <>B.LENGTH

OR A.LENSEC <>B.LENSEC

OR A.ALLOWNULL <>B.ALLOWNULL

ORDER by 1, 4

如果两个数据库可以建立dblink，那么通过dblink就可以像在一个数据库一样去对比。

如果不能建立dblink，那么就只能导出两行表，通过spool导出就可以。然后通过系统级别的命令去比对。我记得linux是有比较两个文件不同行的功能的。

如果认为两个都导出比较麻烦，那就导出一个，在导入另外一个库（注意变换名字，否则表明重复就麻烦了），然后再对比。

---