Uploading of historic data into BW from legacy

[Former Member]

Hi all,

I would like some inputs from all the BW experts in the forum.

Here is the scenario:

My client has all the data from 2001 in R/3 and BW. But prior to 2001, the data was maintained in a legacy system. Now my client wants to move the data prior to 2001 into BW and use the data for reporting purposes. What are the possible options to move the data into BW?

The data in legacy system is not as descriptive and elaborate as in BW.It doesnt have all the attribute fields.

So,do i need to create a new cube, with the availiable fields and upload the data?

Is there any way that i could upload the data in the existing cubes?

Which one is the appropriate method of uploading data,DB connect or flatfile upload? there are approx. 10 million records

Any insight would be appreciated.

Thaks in Advance.

Mav.

[former_member181887]

Hi Maverick,

I would suggest that you create a new ODS Object or InfoCube for the old data. I would start with the model of the new data and make changes as the data dictates. You can then create a MultiProvider on the new and old data for comprehensive reporting.

With regard to the technology for getting the data. I think the decision is based on a couple of considerations:

1. DB/OS of the system where the old data resides

2. How much transformation is required for the data to be loaded.

Number 1 will tell you if DB Connect can be used.

Number 2 will tell you if you need to create simple or complex transformation routines or if it would be better to transform the data before getting to SAP BW.

DB Connect would not be the best solution if you have massive transformation requirements. It may be better to do this on extraction to a flat file or ETL tool.

If you chose flat files, I would split the files in smaller chunks of data for loading. It is easier to manage and achieve parallelism that way. Also if you can presort the flat files, it is faster too!

I hope this helps,

Mike

[Former Member]

hi Mike,

That was a helpful insight. Thanks. Waiting on some more replies and different prespectives.

Thanks,

Mav.

[Former Member]

Hi Maverick,

As you say data is coming from Legacy system, that too 10 millions records, DBConnect is a good method to extract large volume of data for following reasons.

1) It has to connect to Legacy System for Long Time

2) It may lock other process to access those tables, if any background jobs are running in Legacy System.

3) Transformation is not easy when you extract and upload into BW.

4) Transfer from one server to another server, it will take long time.

Better Extract those records in Legacy System and do the Transformation there itself,You can ask whoever is maintaining those systems to do the transformation as per you design.

And also ask them to Sort it and give to multiple files may be group by Month, or Year.

[Former Member]

HI Raja,

Thanks for the reply, but i what i didnt understand is .. what is ur recommendation? U said that DBConnect is a good method but at the same time, in the reasons given below, they didnt sound like supporting the statement.

As far i understand ur reply, even u say that flat file extraction is a much better way provided the transformations are done in the kegacy system according to the data model of BW.. Pls lemme know if i got it wrong?

Thanks for the input,

Mav.