一次sqlldr性能测试案例

$ sqlldr

SQL*Loader: Release 10.2.0.1.0 - Production on Tue May 19 16:06:21 2009

Copyright (c) 1982, 2005, Oracle. All rights reserved.

Usage: SQLLDR keyword=value [,keyword=value,...]

Valid Keywords:

userid -- ORACLE username/password

control -- control file name

log -- log file name

bad -- bad file name

data -- data file name

discard -- discard file name

discardmax -- number of discards to allow (Default all)

skip -- number of logical records to skip (Default 0)

load -- number of logical records to load (Default all)

errors -- number of errors to allow (Default 50)

rows -- number of rows in conventional path bind array or between direct path data saves

(Default: Conventional path 64, Direct path all)

bindsize -- size of conventional path bind array in bytes (Default 256000)

silent -- suppress messages during run (header,feedback,errors,discards,partitions)

direct -- use direct path (Default FALSE)

parfile -- parameter file: name of file that contains parameter specifications

parallel -- do parallel load (Default FALSE)

file -- file to allocate extents from

skip_unusable_indexes -- disallow/allow unusable indexes or index partitions (Default FALSE)

skip_index_maintenance -- do not maintain indexes, mark affected indexes as unusable (Default FALSE)

commit_discontinued -- commit loaded rows when load is discontinued (Default FALSE)

readsize -- size of read buffer (Default 1048576)

external_table -- use external table for load; NOT_USED, GENERATE_ONLY, EXECUTE (Default NOT_USED)

columnarrayrows -- number of rows for direct path column array (Default 5000)

streamsize -- size of direct path stream buffer in bytes (Default 256000)

multithreading -- use multithreading in direct path

resumable -- enable or disable resumable for current session (Default FALSE)

resumable_name -- text string to help identify resumable statement

resumable_timeout -- wait time (in seconds) for RESUMABLE (Default 7200)

date_cache -- size (in entries) of date conversion cache (Default 1000)

PLEASE NOTE: Command-line parameters may be specified either by

position or by keywords. An example of the former case is 'sqlldr

scott/tiger foo'; an example of the latter is 'sqlldr control=foo

userid=scott/tiger'. One may specify parameters by position before

but not after parameters specified by keywords. For example,

'sqlldr scott/tiger control=foo logfile=log' is allowed, but

'sqlldr scott/tiger control=foo log' is not, even though the

position of the parameter 'log' is correct.

0)
控制文件

install.ctl(含序列,不能设direct=true):
load data
infile '/orahome/wangnc/INSTALLDETAIL.dat'
append
into table sys_login_records_detail_1
fields terminated by ','
(
LOGIN_DATE DATE "yyyy-mm-dd" TERMINATED BY whitespace,
FILLER_1 FILLER,
GAME_ID,
FILLER_2 FILLER,
IP,
MAC,
NETBAR_ID,
PROVINCE,
CITY,
LOGIN_TIMES,
id POSITION(1:1) "sys_login_records_detail_seq_1.nextval"
)

install4.ctl(不含序列,可设direct=true):
Load data
infile '/orahome/wangnc/INSTALLDETAIL.dat'
append
into table sys_login_records_detail_1
fields terminated by ','
(
LOGIN_DATE DATE "yyyy-mm-dd" TERMINATED BY whitespace,
FILLER_1 FILLER,
GAME_ID,
FILLER_2 FILLER,
IP,
MAC,
NETBAR_ID,
PROVINCE,
CITY,
LOGIN_TIMES
)

1)
常规导入

sqlldr userid='emoa_sqlldr/"xxx"@db_esuite' \
control=/orahome/wangnc/install.ctl \
log=/orahome/wangnc/install1.log \
bad=/orahome/wangnc/install1.bad
##Space allocated for bind array: 132352 bytes(64 rows)
##Elapsed time was: 00:01:45.85

2) 1000条提交一次

sqlldr userid='emoa_sqlldr/"xxx"@db_esuite' \
control=/orahome/wangnc/install.ctl \
log=/orahome/wangnc/install2.log \
bad=/orahome/wangnc/install2.bad \
rows=1000 readsize=2068000 bindsize=2068000
##Space allocated for bind array: 2068000 bytes(1000 rows)
##Elapsed time was: 00:00:22.80

3) 10000条提交一次

sqlldr userid='emoa_sqlldr/"xxx"@db_esuite' \
control=/orahome/wangnc/install.ctl \
log=/orahome/wangnc/install3.log \
bad=/orahome/wangnc/install3.bad \
rows=10000 readsize=20680000 bindsize=20680000
##Space allocated for bind array: 20680000 bytes(10000 rows)
##Elapsed time was: 00:00:20.25

4)
设置direct=true,
含序列

sqlldr userid='emoa_sqlldr/"xxx"@db_esuite' \
control=/orahome/wangnc/install.ctl \
log=/orahome/wangnc/install4.log \
bad=/orahome/wangnc/install4.bad \
direct=true
##ORA-01400: cannot insert NULL into ("EMOA"."sys_login_records_detail_1"."ID")

5)
设置direct=true,
去掉序列字段,
最快的方式

#SQL> alter table emoa.sys_login_records_detail_1 drop column id;
sqlldr userid='emoa_sqlldr/"xxx"@db_esuite' \
control=/orahome/wangnc/install4.ctl
\
log=/orahome/wangnc/install4.log \
bad=/orahome/wangnc/install4.bad \
direct=true
##Elapsed time was: 00:00:10.98
##但是如果有重复数据,
会把唯一索引置为unusable,
要确保数据源已剔重

6)
设置readsize和bindsize到最大值,
高效又稳妥的方式,
最终采纳方案

sqlldr userid='emoa_sqlldr/"xxx"@db_esuite' \
control=/orahome/wangnc/install4.ctl \
log=/orahome/wangnc/install6.log \
bad=/orahome/wangnc/install6.bad \
rows=100160 readsize=20971520 bindsize=20971520 PARALLEL=TRUE
##Space allocated for bind array: 20970240 bytes(10160 rows)
##Elapsed time was: 00:00:14.36

--End--