cancel
Showing results for 
Search instead for 
Did you mean: 

Parallelization of 0CO_OM_NWA_2 extractor

Former Member
0 Kudos
189

Each week, I load labor cost data from this extractor to my Cost Cube. Each week, I pull 2 million rows. Each week, I have to wait for 5 hours before the extractor starts to pull data. Then, BW processes the data very efficiently, under 90 minutes. I've read most of the notes I can find regarding the performance of this extractor and have implemented most of the changes.

My question is: Can I establish separate initializations for this extractor, each with a range of Network Activities, and then use parallel extractors (one for each initialization range) to pull and process the data faster?

For that matter, can I apply the same approach to my CCA_9 extractor?

Regards,

George.

Accepted Solutions (0)

Answers (5)

Answers (5)

Former Member
0 Kudos

See Note:382329 and create missing index on table COEP.

Former Member
0 Kudos

Why don't you extract the data more frequently, lets say daily to get smaller pieces of data to be extracted?

Former Member
0 Kudos

Not familiar with the extractor to know how complex it might be and the number of tables/views it might be using.

Have you looked at adding index(es) to any of the R3 tables involved to sppeed the extraction?

Former Member
0 Kudos

Make use of disjoint selections for 0FISCPER like

0FISCPER = 2003001 to 2003016

0FISCPER = 2004001 to 2004016

0FISCPER = 2005001 to 2005016

Thanks

Former Member
0 Kudos

Hi George,

if you use (generally speaking) a datasource in delta mode you are able to perform different (not overlapping) inits in parallel, but only ONE delta channel will be managed by your system after that...so, several (parallel) inits don't give the chance to manage several (parallel or not) delta loads for the same datasource/BW destination...

Bye,

Roberto