Loading data u2013 data mart scenario

[Former Member]

We are loading from a cube to 20 other cubes with one info package.

If I created say 2 info packages and loaded to 10 cubes from each info package, would this reduce the runtime for loading and if so what do I need to ensure is available ie

No of processes used to load to each cube (does anyone know what this is and whether this is background, dialog) etc

Thanks

[Former Member · Accepted solution]

Hi,

This depends on many things,....

1) No of available resources(dialog/bkgd) in ur system.

2) The type of rules u have defined...

3) The data packet size u have defined at the IP level.

4) The fastness of moving idocs i.e in SM58...

5) The extraction job and the performance(deleting indexes) of the target cubes...

Now whenever u start a load then the extraction job will be running in BKGD only but the processing of each data packet runs in dialog.So the extraction job always be finishing in lesstime but the processing and updating takes time..this u can monitor it in SM51.

And then the dailog processes can be monitored at IP level>scheduler>default data's transfer there ucan chk the no of dialog processes....There itself u can increase the dat apackert size also....

and if ur in BI use DTP's it always runs in BKGD so it access very fast as there is no concept of TRFC i.e SM58 in DTP's...u can increase the batch/parallel processes also...

If the load is huge as ur loading to 20 cubes i suggest to run one after the other if ur using 2 packages...

Use SPRO>Display IMG> to Maintain Control Parameters for Data Transfer

--

rgds,

Edited by: Krishna Rao on May 14, 2009 1:00 PM

[Former Member · Accepted solution]

Thanks. What factors can increase/control the speed of moving idocs

[former_member200211 · Accepted solution]

Hi,

Generally, if the PSA is having huge amount of data, moving of IDOCs takes time. So in that case, we need to delete PSA data and proceed further.

Thanks

Lavanya.

[Former Member · Accepted solution]

Hi,

this mostly depends on the free dialog processes available in the system...

Before running this load once discuss with BASIS team by giving the details of the load.....

Howmany records are there in ur load?

--

rgds,

[Former Member · Accepted solution]

Thanks there are about 1 million records being transferred and depending on the cube being updated less than 50,000 records being added

Just a thought otherwise can timelines be compared to transferring 20 million records since it is loading to 20 cubes

[Former Member · Accepted solution]

Hi,

1 million records to each cube is not a big amount, i guess the load will be fine within less time only...

Don't forget to delte the indexes before loading..while running keep refreshing SM58 session and chk how the Idocs are moving...

--

rgds,