on 12-11-2011 6:41 AM
Hi Guys ,
I have a question related to filling set up table .The scenario is :
I have added a new field to a data source and replicated the data source to BI system .I have added the same field in Info cube , DSO .
Now deleting and filling the data in set up table will take 4 days which is a log time and user cannot afford to give such a long time .Is there any other way to get the data to the new field then taking 4 days down time.
Regards
Santosh
Hi,
Try this, create one simple data source with full load with the fields - keys which you used in immediate standard DSO above PSA and new field you added in your data source. Let the new records to come via existing flow. For existing records, use the newly created data source which will update the records in DSO accordingly(update only those records which are in DSO using look ups in routine).
That will remove the need of filling setup table at cost of new simple datasource.
Thanks and Regards,
Prashant Vankudre
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
santhosh,
Why do you need down time to refill setup tables...
The simple steps are like this :
1. Stop collection jobs
2. Clear delta queues
3. Move datasource Transport
4. Replicate datasource
5. Move BW transports
6. INIT without data transfer
7. Restart collection jobs
8. Start regular deltas
9. Do a historical fill of the setup tables and do a repair full into BW.
You would need downtime from Steps 1 to 6 and this will not take 4 days. I would say that you would not need downtime atleast from Step 1 to 3 - this is because the collection jobs and the datasource structure have to match - or if downtime is not an option - move the transports and then do a historical fill immediately after the init is done to cover for the time the transport took.
Hi,
I think you need to couple of things
1. Discussion with business
2. Plan your data load
Just ask business which data is important for him means whether you can just load current period or latest quarter oir year data first.
Post that you can load you historical data slowly..
so you can have your downtime minimised to say 1 -2 days..
Thanks and regards
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
Hi Santosh,
You may like to create a generic datasource based on some key parameters (like Billing Document number and Item number in Billing application to identify unique record ) along with added information.
Then Replicate this newly created datasource and run init for this data source. This would ensure that all historical data is loaded with enhanced information. This generic data source is used only for loading enhanced information from historical data then on there is no use of this datasource.
Rest regular deltas and other things are explained by experts already.
Hope that helps.
Regards
Mr Kapadia
*Assigning points is the way to say thanks*
.
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
Hi,
I don't think that any downtime is needed at all in this scenario.
Following are the steps you need to proceed with for fulfilling your requirement(if historical data not required for enhanced field) :-
1. Fix ECC down time for 20 to 30 munities
2. Keep all objects on Quality system in ECC and BW.
3. Run Delta laods in BW for 2 to 3 times.So with this step you can clear SMQ1 and RSA7. Check the entries in RSA7, if it is ZERO then it is fine.
4. Move DS from ECC Quality to ECC PROD .
5. Replicate in BW.
6. Move all BW objects from BW Quality to BW Prod.
7. Delete Init load in InfoPackage level (Not in ODS/Cube).
8. Load Init without DataTransfer as data already
9. Then Run Delta.
10.Next day on wards deltas will come as usual.
If you need Historical data also.
1. Delete data in Cube.
2. Fix Down Time and load Init then Delta.
Navesh
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
User | Count |
---|---|
89 | |
10 | |
9 | |
9 | |
9 | |
6 | |
6 | |
5 | |
5 | |
4 |
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.