on 2014 Jan 30 6:47 AM
Hi,
I would like to know how data from planning area is backed up to BW cube logically.
In planning area, we have around 80k CVC but when it is backed up to cube its in millions.
So theoretically does each CVC is multiplied by the fiscal period to have so many combinations? Also does key figure has any role to play here.
Please give your inputs.
Regards,
Alok
Hi Alok,
As far as I know, you need to create a Data Source for the planning Area as follows:
go to MSDP_ADMIN, find your planning Area, double click on it, and go to "Extras" tab as follows
Create a Data Source, give a data source name, your MPOS (for the Planning Area) by default appears and F8. One Important point is, if you maintain the following setting
The key figures without values are not extracted (this could help you reduce the number of records)
Create an Infopackage that extracts data from Data Source, that will get data to PSA and then further a transformation and DTP to drop all the data to target Infocube
Now coming to the question of numerous records, let me give you an example,
say you have CVC1 for Week 1. 2014 has some key figure values and in your selection you have used 4 fields, based on combinations, you will have (Number of weeks)*(number of CVCs)C4 combinations which is enough to get your 1 model to approximately 500 entries
Hope this helps you.
Regards
JB
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
Thanks JB for explaining in detail.
I understand the process which you mentioned to create data source, infopackage and transformation to send data to infocube.
We have exactly the same process in place to extract the data to PSA and then load to infocube via DTP.
On trying to understand number of CVCs, suppose if we introduce couple of new key figures in the planning area which spans across the entire time period - Will it have any impact on the increase in records being transferred to PSA?
Generally, number of CVCs would be multiplied by number of weeks. So will key figure play a role here.
Thanks
Hi Alok
There is already some good response to your question.
The No Extraction of Data Records without Key Fig Values indicator prevents the system from transferring data records in which all key figure values are zero (initial). This will surely help.
Besides I would like to mention that we should not take the backup of each and every key figure from DP PA. Backup should be limited to only those key figures for which numbers are entered by users in the planning book.
We should not include any key figure for which data is already exist in some other cube like (Sales History data etc.), and any macro calculated key figures
Above will be bring down the number of key figure to be backed up by at least to half of total KFs.
Also No. of periods in the past and future helps in further limiting the data volume.
Also run these infopackages and DTPs in parallel while loading data into PSA and then to DTP, this will help considerably to improve the performance of DP Planning area back up job. Also this helps in avoiding memory problems.
Hope this will help.
Thank you
Satish Waghmare
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
Hi Alok,
There might be a mis-communication, you cannot run infopackage and DTP in parallel.
Infopackage pulls data from Source to PSA, according to the selections you make, DTP extracts from PSA to the Target through a transformation.
Yes you can assign a parallel processing profile in Data Source properties. which will speed up data extraction to PSA.
Also you can maintain parallel extraction for DTP, suggest you not to go beyond 8 processes to avoid overload
Regards,
JB
Hello Alok,
Let me clarify -
1. Infopackages can be run in Parallel. It will help improve the performance. Since you mentioned you have 80K CVCs. You need to identify a way to group those CVCs into 4 or 5 groups. May be based on Sales Organization or Sales Region or Product Divisions or Customer Group. Then you can run these set of infopackages in parallel using process chain. Like Running all four infopackages in parallel to load the data for USA, LA, EU and APAC regions. This saves crucial time.
2. Same logic for DTPs. Running it the set of DTPs in parallel using Process chain.
3. DTP must not be run in parallel with Infopackage as data in the PSA needs to be ready prior to running the DTP. Here Data flow is Infopackge loads data from Planning Area into PSA. Then DTP loads data from PSA into Infocube(Data Target).
4. Parallel Processing Profile can be set in datasource. Please refer transaction code /N/SAPAPO/SDP_EXTR, Select the PA, DataSource Name and then Click on DataSource Properties
5. You can develop process chain like below -
Hope this will help.
Thank you
Satish Waghmare
Hi Alok,
Adding Key figures will not have an effect on the number of records.
Use the check box i mentioned, this could reduce the number of records extracting to PSA.
Regards,
JB
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
Hello Alok,
To the best of my knowledge, when you generate CVCs, there is already a cube in which these are stored. You can go to the transaction code listcube and type *mposname* and press F4 to get the cube name.
Also, when you take the data from Planning area to cube, the selection depends on the number of key figures selected as well. When you take the data to backup cube, how old data are you keeping it in the cube ?
Babu Kilari
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
Hi Babu,
Yes the underlying cube(MPOS NAME) contains the CVCs but they are not containing any keyfigure details. Number of CVCs in planning area (MC62) is matching the number of records in this internal cube.
We take livecache backup of planning area and copy it to BW cube(Full update). This includes Key figure data also.Time period data availbale in cube is from 2009 to 2015.
User | Count |
---|---|
8 | |
6 | |
3 | |
2 | |
2 | |
1 | |
1 | |
1 | |
1 | |
1 |
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.