本文介绍使用索引的with ignore_dup_row选项来在线删除表的重复数据行。

创建非唯一、聚簇索引，带选项 : with ignore_dup_row
在索引的创建语法中，unique 和  with ignore_dup_row 是互斥的，nonclustered 和  with ignore_dup_row 是互斥的；也就是说，使用with ignore_dup_row选项时索引必须是非唯一、聚簇的。
通过索引删除重复数据后，再把新建的非唯一、聚簇索引删除。

下面演示创建过程及注意问题：

--创建表
create table DeleteDuplicateRow(id int
, name varchar(30)
, begin_time datetime
)
go
create index idx_DeleteDuplicateRow on DeleteDuplicateRow(id , name)
go

--插入测试数据
insert into DeleteDuplicateRow values(1,'www.dbainfo.net','2014-5-5 14:30')
go
insert into DeleteDuplicateRow values(2,'Sybase数据库技术,数据库恢复','2014-8-17 1:20')
go
insert into DeleteDuplicateRow values(3,'分享Sybase数据库知识','2014-9-11 8:54')
go

1> select * from DeleteDuplicateRow
2> go
 id          name                           begin_time                     
 ----------- ------------------------------ -------------------------------
           1 www.dbainfo.net                            May  5 2014  2:30PM
           2 Sybase数据库技术,数据庠            Aug 17 2014  1:20AM
           3 分享Sybase数据库知识                Sep 11 2014  8:54AM

(3 rows affected)

--从自身循环插入2次
1> insert into DeleteDuplicateRow
2> select * from DeleteDuplicateRow
3> go 2

--总行数：12
1> select count(*) from DeleteDuplicateRow
2> go

--因为ASE要求一张表上最多只能有一个聚簇索引，所以先检查表上是否存在聚簇索引：
1> sp_helpindex DeleteDuplicateRow
2> go
Object has the following indexes
 
 index_name             index_keys index_description index_max_rows_per_page index_fillfactor index_reservepagegap index_created       index_local 
 ---------------------- ---------- ----------------- ----------------------- ---------------- -------------------- ------------------- ------------
 idx_DeleteDuplicateRow  id, name  nonclustered                            0                0                    0 Dec  9 2014 11:10PM Global Index

(1 row affected)
 index_ptn_name                   index_ptn_seg
 -------------------------------- -------------
 idx_DeleteDuplicateRow_766902983 default      

(1 row affected)
--发现表上只有一个非唯一、非聚簇索引：idx_DeleteDuplicateRow，可以不管这个索引。

--如果使用with ignore_dup_row时不加clustered会报错：
1> create index clu_DeleteDuplicateRow on DeleteDuplicateRow(id , name , begin_time) with ignore_dup_row
2> go
Msg 1916, Level 16, State 3:
Server 'CENTOS5', Line 1:
CREATE INDEX options nonclustered and ignore_dup_row are mutually exclusive.

--创建表DeleteDuplicateRow上所有列的非唯一、聚簇索引：clu_DeleteDuplicateRow
1> create clustered index clu_DeleteDuplicateRow on DeleteDuplicateRow(id , name , begin_time) with ignore_dup_row
2> go
Warning: deleted duplicate row.  Primary key is '1, "www.dbainfo.net", "May  5 2014  2:30:00:000PM"'
Warning: deleted duplicate row.  Primary key is '1, "www.dbainfo.net", "May  5 2014  2:30:00:000PM"'
Warning: deleted duplicate row.  Primary key is '1, "www.dbainfo.net", "May  5 2014  2:30:00:000PM"'
Warning: deleted duplicate row.  Primary key is '2, "Sybase数据库技术,数据庢, "Aug 17 2014  1:20:00:000AM"'
Warning: deleted duplicate row.  Primary key is '2, "Sybase数据库技术,数据庢, "Aug 17 2014  1:20:00:000AM"'
Warning: deleted duplicate row.  Primary key is '2, "Sybase数据库技术,数据庢, "Aug 17 2014  1:20:00:000AM"'
Warning: deleted duplicate row.  Primary key is '3, "分享Sybase数据库知识", "Sep 11 2014  8:54:00:000AM"'
Warning: deleted duplicate row.  Primary key is '3, "分享Sybase数据库知识", "Sep 11 2014  8:54:00:000AM"'
Warning: deleted duplicate row.  Primary key is '3, "分享Sybase数据库知识", "Sep 11 2014  8:54:00:000AM"'
Non-clustered index (index id = 2) is being rebuilt.

--查看表DeleteDuplicateRow上的索引情况，包含：非唯一、非聚簇索引，非唯一、聚簇索引
1> sp_helpindex DeleteDuplicateRow
2> go
Object has the following indexes
 
 index_name             index_keys            index_description                index_max_rows_per_page index_fillfactor index_reservepagegap index_created       index_local 
 ---------------------- --------------------- -------------------------------- ----------------------- ---------------- -------------------- ------------------- ------------
 clu_DeleteDuplicateRow  id, name, begin_time clustered, ignore duplicate rows                       0                0                    0 Dec  9 2014 11:46PM Global Index
 idx_DeleteDuplicateRow  id, name             nonclustered                                           0                0                    0 Dec  9 2014 11:10PM Global Index

(2 rows affected)
 index_ptn_name                    index_ptn_seg
 --------------------------------- -------------
 clu_DeleteDuplicateRow_1230904636 default      
 idx_DeleteDuplicateRow_1230904636 default      

(2 rows affected)
(return status = 0)

--删除刚才创建的非唯一、聚簇索引：clu_DeleteDuplicateRow
1>drop index DeleteDuplicateRow.clu_DeleteDuplicateRow
2>go

--查看表DeleteDuplicateRow上的索引情况
1> sp_helpindex DeleteDuplicateRow
2> go
Object has the following indexes
 
 index_name             index_keys index_description index_max_rows_per_page index_fillfactor index_reservepagegap index_created       index_local 
 ---------------------- ---------- ----------------- ----------------------- ---------------- -------------------- ------------------- ------------
 idx_DeleteDuplicateRow  id, name  nonclustered                            0                0                    0 Dec  9 2014 11:10PM Global Index

(1 row affected)
 index_ptn_name                    index_ptn_seg
 --------------------------------- -------------
 idx_DeleteDuplicateRow_1230904636 default      

(1 row affected)
(return status = 0)

--最后验证一下删除重复行之后的数据
1> select * from DeleteDuplicateRow
2> go
 id          name                           begin_time                     
 ----------- ------------------------------ -------------------------------
           1 www.dbainfo.net                            May  5 2014  2:30PM
           2 Sybase数据库技术,数据庠            Aug 17 2014  1:20AM
           3 分享Sybase数据库知识                Sep 11 2014  8:54AM

(3 rows affected)
--EOF--

相关文章

• 2012-03-21 -- Sybase去除重复数据的一种方法 (0)

以后花时间整理一下去除重复数据的不同方法。

今天，仅仅介绍如何使用union操作符来去除重复值。

在select命令中union all不删除重复行，而union会删除重复行。

这样的话，将包含重复行的记录通过union操作符插入到一张临时表中，然后再倒回来就行了。

比如：对表tmp有5条记录，其中有2条重复记录。

1> select *  from tmp
2> go
 id          name
 ----------- --------------------------------------------------------
           1 sysobjects
           2 sysindexes
           3 syscolumns
           1 sysobjects
           3 syscolumns
(5 rows affected)

通过union操作符，将表tmp的记录插入到另外一张临时表tmp1中。

select * into tmp1 from tmp

union

select * from tmp where 1=2

go

1>select * into tmp1 from tmp
2>union
3>select * from tmp where 1=2
4>go
(3 rows affected)
1>select * from tmp1
2>go
 id          name
----------- --------------------------------------------------------
           1 sysobjects
           2 sysindexes
           3 syscolumns
(3 rows affected)

再将临时表tmp1的数据倒回tmp就完成去除重复数据的工作。

相关文章

• 2019-04-12 -- Sybase ASE及SQL Anywhere数据库修复专家 (0)
• 2013-09-16 -- ASE从表中随机选择数据的三种方法 (0)
• 2013-09-08 -- 查看ASE v15正在执行的SQL语句内容 (0)
• 2013-05-23 -- 修改ASE的服务器名称 (0)
• 2012-04-11 -- ASE修改列的默认值属性 (0)
• 2019-04-11 -- SQL Server 数据库修复专家SQLRescue (0)
• 2015-01-16 -- master数据库配置区域(configuration area)损坏的一种修复方法 (1)
• 2014-12-09 -- ASE使用with ignore_dup_row删除重复数据 (1)
• 2014-10-10 -- ASE对象名称长度限制 (0)
• 2014-06-09 -- 统计用户数据库内所有表的行数 (0)