Question:
I am responsible for a Transactional Data Consolidation Cube. The main problem is with loading. It has only two InfoSources 3EC_CS_1A for year 2003/2004. When I kick off the InfoPackage to load the data, it takes 2-3 hours before I see any movement. We do a full load everytime. Delta was not tried because the data we receive is not so big (one million records or less). We have other cubes that load within minutes with much more data. Does anyone know what's so special about ECCS cubes. Is it R/3 that has do extra work to extract data out of certain areas and causes the delay. Any way we can avoid the delay with out implementing the deltas? Any help or suggestion would be appreciated. Thanks.
BG
Answer:
hi,
U can make it faster by breakign it for smaller periods.
By specifying consolidation group.
Delete the secondary indexes of the cube.
check out if u are using any custom code.
see if thereis any breakpoint in any of your code.
Try to use version based on actual and plan might be u are loading plan and actual at the same time.
I hope u are doing all these which wil improve the loads to go faster.
Answer:
Thanks VAS1,
I cannot use smaller than 1 year's time periods. So I have two extracts one each for the current and last years. I am using all the restriction that you have mentioned but I am pulling both the planing and actual data as this was a requirement.
I was thinking of finding the tables in R/3 where the extract is coming from and ask the DB guys to create indexes etc. but this extract is using a functional module FCIW_BIW_GET_DATA_A. The abap code is hard to read and I have no clue where the data is coming from. Even if I can find out, chances are the data is scattered all over. It's not going to be a couple of tables where I can get indexes and make the extraction faster.
Any Ideas?
Looking forward.
BG
Answer:
I would go for the index.
Make an SQL trace in st05 here you will in which SQL statement the time is spend.
Copy the statment and put it in the the explain SQL.
There you will see with tables are being accessed.
Build the need indexes, im sure that the R/3 and basis team will asist you.
Test the indexes by using the explain SQL again.
When the indexes fullfill your requirements, try your extractor again.
_________________
Rasmus ;o)
- One small step for man, one giant leap for mankind