on 04-14-2014 4:31 PM
Hi Guys,
Good day!
I am currently having a performance issue regarding with extraction of a job.
My job extracts an average of 215,000 row count for every run weekly. The extraction takes up to 3 to 4 hours long compared to the usual duration of 2 hours. As per checking of the trace log, the long run is occurring in the Data Flow.
Data flow structure:
SQL (Database Type Oracle 10g) -> Table (Microsoft SQL Server 2008)
The current configuration of the Source and Target in the designer is as follows:
SQL
Database type: Oracle 10g
Array fetch size: 1000
Table Target
Database type: Microsoft SQL Server 2008
Rows per Commit: 1000
Number of loaders: 1
With this, I would like to ask on how can I optimize the performance of the job where I can reduce the extraction time.
Thanks,
Hi Neil,
I suggest you to contact system administrator.Some times virtual memory occupied more memory than there will be performance issue.
I recommend analyse and increase virtual memory size.
Regards
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
Is this dataflow getting full pushdown? If not, then you can try making it full push down. This shall drastically improve the performance.
By the way, are you using any complex logic in this dataflow? How it is taking 2 hours? Share more details so that we can exactly tell you where is the problem.
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
Hi Neil,
Please refer the below guide
http://help.sap.com/businessobject/product_guides/sbods42/en/ds_42_perf_opt_en.pdf
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
Hi,
Increase Rows per Commit and Number of loaders
Regards,
Manoj.
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
User | Count |
---|---|
85 | |
10 | |
10 | |
9 | |
7 | |
6 | |
6 | |
5 | |
4 | |
4 |
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.