海量数据迁移之通过rowid切分大表
在之前的章节中，讨论过了通过 分区+并行等方式来进行超大的表的切分，通过这种方式能够极大的提高数据的平均分布，但是不是最完美的。
比如在数据量再提高几个层次，我们假设这个表目前有1T的大小。有10个分区，最大的分区有400G,那么如果我们想尽可能的平均的导出数据，使用并行就不一定能够那么奏效了。
比方说我们要求每个dump文件控制在200M总有，那样的话400G的分区就需要800个并行才能完成，在实际的数据库维护中，我们知道默认的并行数只有64个，提高几倍，也不可能超过800
所以在数据量极大的情况下，如果资源紧张，可能生成的dump就会比较大。
我们考虑使用rowid来满足我们的需求。
我们可以根据需要来指定需要生成几个dump文件。比如表subscriber有600M,那么如果按照200M为一个单位，我们需要生成3个dump文件。
如果想数据足够平均，就需要在rowid上做点功夫。
我们先设定一个参数文件，如下的格式。
可以看到表memo数据量极大，按照200M一个单位，最大的分区（P9_A3000_E5）需要800个并行。
表ICE_AGREEMENT比较小，不是分区表，我们以x来临时作为分区表的代名，在处理的时候可以方便的甄别
MEMO P9_A3000_E0 156
MEMO P9_A3000_E1 170
MEMO P9_A3000_E2 190
MEMO P9_A3000_E3 200
MEMO P9_A3000_E4 180
MEMO P9_A3000_E5 800
MEMO PMAXVALUE_AMAXVALUE_EMAXVALUE 1
ICE_AGREEMENT x 36
CRIBER_HISTORY x 11
可以使用如下的脚本来完成rowid的切分。
创新互联是专业的建安网站建设公司，建安接单;提供成都网站设计、网站建设,网页设计,网站设计,建网站,PHP网站建设等专业做网站服务;采用PHP框架,可快速的进行建安网站开发网页制作和功能扩展;专业做搜索引擎喜爱的网站,专业的做网站团队,希望更多企业前来合作!
#### $1 dba conn details
#### $2 table owner
#### $3 table_name
#### $4 subobject_name
#### $5 parallel_no
function normal_split
{
sqlplus -s $1 <<1eof
set linesize 200
set pages 0
set feedback off
spool list/rowid_range_$3_x.lst
select rownum || ', ' ||' rowid between '||
chr(39)||dbms_rowid.rowid_create( 1, DOI, lo_fno, lo_block, 0 ) ||chr(39)|| ' and ' ||
chr(39)||dbms_rowid.rowid_create( 1, DOI, hi_fno, hi_block, 1000000 )||chr(39) data
from (
SELECT DISTINCT DOI, grp,
first_value(relative_fno) over (partition BY DOI,grp order by relative_fno, block_id rows BETWEEN unbounded preceding AND unbounded following) lo_fno,
first_value(block_id ) over (partition BY DOI,grp order by relative_fno, block_id rows BETWEEN unbounded preceding AND unbounded following) lo_block,
last_value(relative_fno) over (partition BY DOI,grp order by relative_fno, block_id rows BETWEEN unbounded preceding AND unbounded following) hi_fno,
last_value(block_id+blocks-1) over (partition BY DOI,grp order by relative_fno, block_id rows BETWEEN unbounded preceding AND unbounded following) hi_block,
SUM(blocks) over (partition BY DOI,grp) sum_blocks,SUBOBJECT_NAME
FROM(
SELECT obj.OBJECT_ID,
obj.SUBOBJECT_NAME,
obj.DATA_OBJECT_ID as DOI,
ext.relative_fno,
ext.block_id,
( SUM(blocks) over () ) SUM,
(SUM(blocks) over (ORDER BY DATA_OBJECT_ID,relative_fno, block_id)-0.01 ) sum_fno ,
TRUNC( (SUM(blocks) over (ORDER BY DATA_OBJECT_ID,relative_fno, block_id)-0.01) / (SUM(blocks) over ()/ $5 ) ) grp,
ext.blocks
FROM dba_extents ext, dba_objects obj
WHERE ext.segment_name = UPPER('$3')
AND ext.owner = UPPER('$2')
AND obj.owner = ext.owner
AND obj.object_name = ext.segment_name
AND obj.DATA_OBJECT_ID IS NOT NULL
ORDER BY DATA_OBJECT_ID, relative_fno, block_id
) order by DOI,grp
);
spool off;
EOF
}
function partition_split
{
sqlplus -s $1 <<1eof
set linesize 200
set pages 0
set feedback off
spool list/rowid_range_$3_$4.lst
select rownum || ', ' ||' rowid between '||
chr(39)||dbms_rowid.rowid_create( 1, DOI, lo_fno, lo_block, 0 ) ||chr(39)|| ' and ' ||
chr(39)||dbms_rowid.rowid_create( 1, DOI, hi_fno, hi_block, 1000000 )||chr(39) data
from (
SELECT DISTINCT DOI, grp,
first_value(relative_fno) over (partition BY DOI,grp order by relative_fno, block_id rows BETWEEN unbounded preceding AND unbounded following) lo_fno,
first_value(block_id ) over (partition BY DOI,grp order by relative_fno, block_id rows BETWEEN unbounded preceding AND unbounded following) lo_block,
last_value(relative_fno) over (partition BY DOI,grp order by relative_fno, block_id rows BETWEEN unbounded preceding AND unbounded following) hi_fno,
last_value(block_id+blocks-1) over (partition BY DOI,grp order by relative_fno, block_id rows BETWEEN unbounded preceding AND unbounded following) hi_block,
SUM(blocks) over (partition BY DOI,grp) sum_blocks,SUBOBJECT_NAME
FROM(
SELECT obj.OBJECT_ID,
obj.SUBOBJECT_NAME,
obj.DATA_OBJECT_ID as DOI,
ext.relative_fno,
ext.block_id,
( SUM(blocks) over () ) SUM,
(SUM(blocks) over (ORDER BY DATA_OBJECT_ID,relative_fno, block_id)-0.01 ) sum_fno ,
TRUNC( (SUM(blocks) over (ORDER BY DATA_OBJECT_ID,relative_fno, block_id)-0.01) / (SUM(blocks) over ()/ $5 ) ) grp,
ext.blocks
FROM dba_extents ext, dba_objects obj
WHERE ext.segment_name = UPPER('$3')
AND ext.owner = UPPER('$2')
AND obj.owner = ext.owner
AND obj.object_name = ext.segment_name
AND obj.DATA_OBJECT_ID IS NOT NULL
AND obj.subobject_name=UPPER('$4')
ORDER BY DATA_OBJECT_ID, relative_fno, block_id
) order by DOI,grp
);
spool off
EOF
}
sub_partition_name=$4
if [[ $sub_partition_name = 'x' ]]
then
normal_split $1 $2 $3 x $5
else
partition_split $1 $2 $3 $4 $5
fi
脚本比较长，需要的参数有5个，因为访问dba_extents,dba_objects需要一定的权限，可以使用dba权限的账号即可。
第2个参数是表的owner，第3个参数是表名，第4个参数是分区表名（如果是分区表就是分区表名，如果不是就填x）,第5个参数就是期望使用的并行度，能够在一定程度上加快速度
简单演示一下，可以通过下面的方式来运行脚本，我们指定生成10个dump这个表不是分区表。
ksh gen_rowid.sh n1/n1 prdowner subscriber_history x 10
1, where rowid between 'AAB4VPAAJAAD7qAAAA' and 'AAB4VPAAJAAD/R/EJA'
2, where rowid between 'AAB4VPAAJAAD/SAAAA' and 'AAB4VPAAKAABV5/EJA'
3, where rowid between 'AAB4VPAAKAABV6AAAA' and 'AAB4VPAALAAE/p/EJA'
4, where rowid between 'AAB4VPAALAAE/qAAAA' and 'AAB4VPAAMAAFFh/EJA'
5, where rowid between 'AAB4VPAAMAAFFiAAAA' and 'AAB4VPAAyAACuh/EJA'
6, where rowid between 'AAB4VPAAyAACuiAAAA' and 'AAB4VPAAzAACe5/EJA'
7, where rowid between 'AAB4VPAAzAACe6AAAA' and 'AAB4VPAA1AACZR/EJA'
8, where rowid between 'AAB4VPAA1AACZSAAAA' and 'AAB4VPAA2AACWR/EJA'
9, where rowid between 'AAB4VPAA2AACWSAAAA' and 'AAB4VPAA4AACP5/EJA'
10, where rowid between 'AAB4VPAA4AACQCAAAA' and 'AAB4VPAA5AACHx/EJA'
然后我们来看看数据是否足够平均。
可以类似下面的方式验证，我们抽第1,2,10个。
SQL> select count(*)from subscriber_history where rowid between 'AAB4VPAAJAAD7qAAAA' and 'AAB4VPAAJAAD/R/EJA'
2 ;
COUNT(*)
----------
328759
SQL> select count(*)from subscriber_history where rowid between 'AAB4VPAAJAAD/SAAAA' and 'AAB4VPAAKAABV5/EJA'
2 /
COUNT(*)
----------
318021
SQL> select count(*)from subscriber_history where rowid between 'AAB4VPAA4AACQCAAAA' and 'AAB4VPAA5AACHx/EJA';
COUNT(*)
----------
332638
可以看到数据还是很平均的，达到了我们的期望。
本文题目：海量数据迁移之通过rowid切分大表
文章起源：http://myzitong.com/article/ieooss.html