| sqlldr parallel jobs [message #407164] |
Mon, 08 June 2009 20:41  |
mrpranab
Messages: 32
Registered: March 2005
|
Member |
|
|
Hi,
How to do parallel jobs in sqlldr?
Suppose I have a flat file having 100 millions records. I need to split the jobs into 5 parts having one job insert 20 millions records each in the oracle table.
So basically 5 sqlldr command need to be call in the script and 1st sqlldr command will insert first 20 millions records, then 2nd sqlldr command will insert next 20 millions records from the same file so on and it will insert data in the same table.
What is the approach need to follow for this sqlldr parallel load jobs.
Thanks.
|
|
|
|
| Re: sqlldr parallel jobs [message #407165 is a reply to message #407164] |
Mon, 08 June 2009 20:51  |
 |
BlackSwan
Messages: 26766
Registered: January 2009
Location: SoCal
|
Senior Member |
|
|
>What is the approach need to follow for this sqlldr parallel load jobs.
split the original file into 5 pieces.
launch sqlldr 4 times into background
launch sqlldr as foreground task.
You need to realize that since all 5 tasks will be banging against the same disk volume you will be creating a disk bottleneck.
Since the physical disk drive is the SLOWEST component in the environment by a factor of 100, you will not get your desired/expect performance boost.
If a single sqlldr runs to completion in 1000 seconds, I would expect 5 jobs "in parallel" to complete in 500 seconds.
[Updated on: Mon, 08 June 2009 21:17] Report message to a moderator |
|
|
|
] 
Blog
Search
Help
Members
Register
Login
Home
