cancel
Showing results for 
Search instead for 
Did you mean: 

working of delta in BW

Former Member
0 Kudos
209

Hello,

can somebody explain me how delta load works in BW.

starting from R/3 system, when a document is considered to be delats and the complete follow till BW. then how it behaves for an ODS and how it behaves for a CUBE.

In an ODS how it behaves for key figures which are overwrite and key figures which are addition.

also what is the role of reversal indicator and recordmode.

regards

Meps

Accepted Solutions (0)

Answers (3)

Answers (3)

Former Member
0 Kudos

Hi Meyappan,

IO_ORD IO_SREG IO_OFF IO_MAT IO_QUAN 0UNIT 0CALDAY

ORD1 K.L MALAYSIA MAT001 2 CS 20050304

ORD2 KLANG MALAYSIA MAT002 100 CS 20050305

ORD3 S/PORE S/GAPORE MAT003 10 CS 20050526

ORD4 J.B MALAYSIA MAT004 2 CS 20050529

The Diagram above shows, before Original data before changes.

IO_OR IO_SRE IO_OFF IO_MAT IO_QUAN 0UNIT 0CALDAY 0R/MODE

ORD1 K.L MALAYSIA MAT001 4 CS 20050304

ORD2 KLANG MALAYSIA MAT002 200 CS 20050305 X

ORD3 S/PORE S/GAPORE MAT003 20 CS 20050526 D

ORD4 J.B MALAYSIA MAT004 4 CS 20050529 R

The Diagram above, shows, after data changes.

A load request can be processed in parallel when data are sent in multiple packets from source system. Each packet will use one SAP work process. This parallel processing generates an ORA-54 error when multiple process compete for the same table lock on the ODS tables.This table lock is essential so that the Overwrite option is selected will not cause data consistency problems when data load executes. Loading data from PSA to ODS object or an InfoCube is handled by SAP work process, which prevents the table locking problem.

Data can be loaded into an ODS object in multiple packets, through parallel processing.Data request are identified uniquely by a combination of several keys fields.

After loading PSA data into ODS Object, the table/BIC/XXX_XXXX contains new data. If you view the second table, second row does not appear anymore, which indicates that BW detected the value "X" in the 0RECORDMODE column and the skipped the second row.

TQ

Kumar

Former Member
0 Kudos

Hi Meyappan,

You use 0RECORDMODE value X to mark rows to be skippep in the delta load. You use D & R to delete and remove rows respectively from active ODS table, if the rows can be identified by table primary key. Otherwise, D & R work just like X, they do not effect either the change log or the active data table.

This is how ODS delta load works, if you have a table lists four rows of the original "sales data -Eg", what happens to the data in the ODS object and IfoCube after loading the table data into them is, has an extra column contains the 0RECORDMODE values, also sale quantity doubled.

TQ

Kumar

Former Member
0 Kudos

Nantha,

Lets have an example now.

DocNo - Item Qty Request ReveralInd

Doc 1 - 10 600.000 REQ1 -

Doc 1 - 10 600.000 REQ1 -

Doc 1 - 10 600.000 REQ1 -

Doc 1 - 10 600.000- REQ1 - X

Doc 1 - 10 600.000 REQ2 -

Doc 1 - 10 600.000- REQ2 - X

Doc 1 - 10 600.000 REQ3 -

These are the records in PSa and this is updating an ODS what will be the result of QTY in ODS, ODS has a primary key of DocNo, Item.

regards

Meps

Former Member
0 Kudos

Hi Meps,

I don't know if yours is a complete and true overview of this record in your PSA, but it seems strange having:

DocNo - Itm Qty Req 0REC

Doc 1 - 10 600 REQ1 -

Doc 1 - 10 600 REQ1 -

Doc 1 - 10 600 REQ1 -

that are three records, for the same item, that describe for three time the same situation !!!

Anyway, if you are using overwrite method, your finla result will be 600.000, in summation 1.800.000 !

Hope it helps!

Bye,

Roberto

Former Member
0 Kudos

hi,

generally speaking, the delta loading works two ways:

1. checking the change documents created for particular objects - like for article/material delta.

2. (um, not too sure about this - but i understand that it exists) data queueing via qRFC - changes are recorded in a storage queue, and BW collects information from here.

ryan.