Copy cube

[Former Member]

Hello guys.

I need your help in telling me what is the best way to make a copy of a cube that has 10 million registers.

I'm in Oracle database, so I need to take care of table spaces also.

But I need to know how can I archive the load of data from the original cube to the copy cube in a successfull way.

Do I have to load all data in once, or in infopackage making filtering selections by 4 or 5 months ??'

the problem of filtering in the DTP is that is creatting me partitions in the fact table of the copy cube, and the loads take a long of time to finish (aprox. 4 hours, for a set of 360,000 registers). How can I improve the performance of loads. ???

THANKS IN ADVANCE.

[former_member184494]

Guillermo,

1. You can archive the data from the original cube and restore it into the new copy cube

2. As for data loads - each request in the fact table (F) is a separate partition - the F table is partitioned based on requests - this is done so that data loads can be recalled easily ( drop the partition )

3. Compress the request and then the partition will get dropped and the same will move into the E table which is partioned as per our needs.

Also before loading to the cube - delete the indices and then rebuild them after the data load. We routinely load 2 to 3 million records to the cube in about an hour or so.

In the performance tab of the cube you will find a button for "Create Index (Batch ) click on that and uncheck the check boxes and select continue ( say no for write transport request )

then click on Delete Index (Now ) and then once the indices are dropped - start the data load.

once it is complete - click on Create Index (Batch ) and select start.

Hope it helps...

Arun

[Former Member]

In which system are u ? BI 7.0 or BW 3.5?

[Former Member]

I'm in BI 7

[Former Member]

I'm in BI 7.

[former_member345199]

Hi,

You can use multiple options for this.

The first is to create a similar cube / copy of this cube and load from this cube to the new cube.

Refer

3.X

Update rules method

You can try datamart scenario

cube1 right click 'generate export datasource'

create update rules for cube2, assign with cube1

rsa1->source system->bw myself, right click 'replicate datasource'

rsa1-> infosource -> search 8cube1 name

(if not get, try right click root note 'infosource'->insert lost node(s) from that infosource, you will find assigned with datasource, right click and 'create infopackage', schedule and run.

7.0

Create transformation btn cubes.

Create DTP and run it.

Creating a similar / Copy of cube

While creating the new cube give the cube name in the "Copy from" section. It would copy all the characteristics and Key figures. It would even copy the dimensions and Navigational attributes

Another option is:

The steps for copying the contents of one cube to another:

1. Go to Manage -> Recontruct of the new cube.

2. Select the "selection button"(red , yellow, blue diamond button).

3.In the selection screen you can give the technical name of the old cube, requests ids you want to load, from & to date.

4.Execute and the new cube would be loaded.

Its all that easy!!!!!!

Refer this link:

Hope this helps.

Thanks,

JituK

[Former Member]

HI,

Copying a cube data to another cube consisits of two steps: creating a cube with same structure, and copying the source data to traget cube.

First step is very simple.

After that, you need to create a transformation for the target cube from the source cube and create a DTP. If you are in BI7, NO need to create a export datasource.

To load a large volume of data, dont schedule all the data in a single go rather use selection.

Thanks...

Shambhu

[Former Member]

hi,

generally coping a cube is done by creating a cube with same structure(copy from)..

then you have do load to the new cube. for this make the source cube as datamart eneabled.right click on source cube -> generate export datasource.

this option will create an infosource with name 8<source cube name>.

create transformation on new cube with source as infocube and give the name of the source cube.

create DTP and loading at once is not at all advisable.also before every load delete the index of the target cube(manage -> performance tab-> delete index)

in the filter of DTP specify 2 months in the month infoobject(0calmonth or 0fiscalyearperiod) run the infopack and if required do compression once the load is done for the target cube, this will reduce the space occupied.

Ramesh