试想一下,如果一个文件5G、10G甚至更大。如何将它迅速地加载到数据库指定的表呢?我们看看Sybase IQ是如何迅速地将表的数据加载到数据库的。
数据文件格式:
1440,2011-01-09 00:00:00,1,珠海,1,C网,8612345678222,221943,1,12175,1,12,14426467,1191632,9,
1440,2011-01-09 00:00:00,1,珠海,1,C网,8612345678222,968852,1,82077,1,7,2430696,134921,5,
1440,2011-01-09 00:00:00,1,珠海,1,C网,8612345678222,936862,3,10847,1,5,4585323,362630,5,
1440,2011-01-09 00:00:00,1,珠海,1,C网,8612345678222,308796,2,5614,2,12,14401931,1202200,11,
1440,2011-01-09 00:00:00,1,珠海,2,固网,861234567000,11058523,6,984391,3,19,12789576,1113565,19,
据以“,”分割。并以“,”结尾。
commit;
lock table iqloadtest in write mode wait '00:05:00';
Set temporary option conversion_error='off';
Set temporary option escape_character='on';
Set temporary option load_memory_mb=256;
Set temporary option timestamp_format='YYYY-MM-DD HH:mm:ss';--解决时间格式问题
LOAD TABLE iqloadtest
(
col1 ',',
col2 ',',
col3 ',',
col4 ',',
col5 ',',
col6 ',',
col7 ',',
col8 ',',
col9 ',',
col10 ',',
col11 ',',
col12 ',',
col13 ',',
col14 ',',
col15 ',\x0a',--最后一个字段顺便过滤掉换行符0a
filler(1)--在此例子中加了此条语句最后一行数据不被加载
)
FROM '/home/test/test.load'
ESCAPES OFF
QUOTES OFF
IGNORE CONSTRAINT ALL 0
MESSAGE LOG '/home/test/iqMsg.log'
ROW LOG '/home/test/iqRow.log'
ONLY LOG ALL
Notify 50000
WITH CHECKPOINT ON;
commit;
结论:如果行结束符后面带了分隔符的话必须用此种方式。否则无法加载。
如果行结束符后面不带分隔符的话可以用filler(1)这种方式
IQLoad 会遇到的问题。
1、 ODBC 与 JDBC 的默认设置不一样。 JDBC 需要加上
Set temporary option escape_character='on';
2、当数据中存在null时的解决方法
commit;
lock table iqloadtest in write mode wait '00:05:00';
Set temporary option conversion_error='off';
Set temporary option escape_character='on';
Set temporary option load_memory_mb=256;
Set temporary option timestamp_format='YYYY-MM-DD HH:mm:ss';
LOAD TABLE iqloadtest
(
字段1 '分隔符' null('NULL'),
字段2 '分隔符' null('NULL'),
col15 '分隔符\x0a' null('NULL'),
filler(1)--在此例子中加了此条语句最后一行数据不被加载
)
FROM '/home/test/test.load'
ESCAPES OFF
QUOTES OFF
IGNORE CONSTRAINT ALL 0
MESSAGE LOG '/home/test/iqMsg.log'
ROW LOG '/home/test/iqRow.log'
ONLY LOG ALL
Notify 50000
WITH CHECKPOINT ON;
commit;