oracle比varchar大的,同样的数据，为什么varchar2(4000)比varchar2(1000)产生的redo大？

如题，在网上看的文章，做了下测试，不知道是为什么，求高手能解释一下，我看网上大多数人都说varchar2的长度不影响性能，但是从实验上看肯定是有影响的。

1、创建2张表，t1用varchar2(4000)，t2用varchar2(1000)

create table t1(x number, x2 varchar2(4000),x3 varchar2(4000))  SEGMENT CREATION IMMEDIATE;

create table t2(x number, x2 varchar2(1000),x3 varchar2(1000))  SEGMENT CREATION IMMEDIATE;

2、分别向表中插入相同的数据

insert into t1 select level, rpad(' ', 100, ' '),rpad('a',100,'a') from dual connect by level <= 1e5;

insert into t2 select level, rpad(' ', 100, ' '),rpad('a',100,'a') from dual connect by level <= 1e5;

3、开启autotrace

SQL> insert into t1 select level, rpad(' ', 100, ' '),rpad('a',100,'a') from dual connect by level <= 1e5;

100000 rows created.

Execution Plan

----------------------------------------------------------

Plan hash value: 1236776825

------------------------------------------------------------------------------

| Id  | Operation              | Name | Rows  | Cost (%CPU)| Time     |

------------------------------------------------------------------------------

|   0 | INSERT STATEMENT              |      |           1 |           2   (0)| 00:00:01 |

|   1 |  LOAD TABLE CONVENTIONAL      | T1   |             |          |             |

|*  2 |   CONNECT BY WITHOUT FILTERING|      |             |          |             |

|   3 |    FAST DUAL              |      |           1 |           2   (0)| 00:00:01 |

------------------------------------------------------------------------------

Predicate Information (identified by operation id):

---------------------------------------------------

2 - filter(LEVEL<=1e5)

Statistics

----------------------------------------------------------

48  recursive calls

112998  db block gets

5588  consistent gets

0  physical reads

45758768  redo size

838  bytes sent via SQL*Net to client

860  bytes received via SQL*Net from client

3  SQL*Net roundtrips to/from client

2  sorts (memory)

0  sorts (disk)

100000  rows processed

=================================================================

SQL> insert into t2 select level, rpad(' ', 100, ' '),rpad('a',100,'a') from dual connect by level <= 1e5;

100000 rows created.

Execution Plan

----------------------------------------------------------

Plan hash value: 1236776825

------------------------------------------------------------------------------

| Id  | Operation              | Name | Rows  | Cost (%CPU)| Time     |

------------------------------------------------------------------------------

|   0 | INSERT STATEMENT              |      |           1 |           2   (0)| 00:00:01 |

|   1 |  LOAD TABLE CONVENTIONAL      | T2   |             |          |             |

|*  2 |   CONNECT BY WITHOUT FILTERING|      |             |          |             |

|   3 |    FAST DUAL              |      |           1 |           2   (0)| 00:00:01 |

------------------------------------------------------------------------------

Predicate Information (identified by operation id):

---------------------------------------------------

2 - filter(LEVEL<=1e5)

Statistics

----------------------------------------------------------

43  recursive calls

16829  db block gets

6660  consistent gets

0  physical reads

23691740  redo size

840  bytes sent via SQL*Net to client

860  bytes received via SQL*Net from client

3  SQL*Net roundtrips to/from client

2  sorts (memory)

0  sorts (disk)

100000  rows processed

=====================================================================

4、最终两张表的block都是2944，一样大

SEGMENT_NAME           BLOCKS

------------------------------ ----------

T1             2944

T2             2944

5、从上面的测试看，最终结果是两张表一样大，但是insert的时候，t1产生的db block gets比t2要多很多，而且产生的redo比t2产生的redo多了将近一倍，这是为什么呢？