Indexing in Explorer against SAP BW (MDX - UDT)

[Former Member]

Hi All,

In Explorer we are connecting to SAP BW database via UDT universe (MDX) created using BEX queries on top of SAP BW, now these queries are too big and contains lot of calculation. (>40+ charac., 10+ Key figures, Calculations etc).

We are facing a challenge while doing the indexing in BI explorer.

Let's say we have data for the month of Jan, Feb, March, Apr etc.

If I am doing indexing for an individual month the indexing is passing, but if I try to index for the whole year I get a memory dump in the SAP BW system.

Is there any way we can index only single month at a time and merge the data for each month at BO side.

Like a delta load, run the query for Jan and then for Feb and data get appended to same information space.

If something like this exists we could use some more filtering options and perform the indexing

We are currently on SAP BI4.1 SP7 Patch 1

Thanks,

Vivek

[Former Member]

Hi,

Any thoughts on the above possibility?

Thanks,

Vivek Chauhan

[former_member376301]

Hi,

Maybe I will say a stupidity....

I don't know BW very well, but maybe you could create an universe (UNX or UNV) on BW with a prompt for Month facet. Next you could clone 12 times the same IS containing this prompt and edit them in order to set the correct month for each IS.

Regards,

[Former Member]

Hi Arnaud,

Thanks for your response, Well this is what we are trying:-

1. Put more filters
2. Break down IS with some logical way !!!

But the idea was to have everything in 1 information space.

So is explorer can understand that last indexing with JAN was successful, next indexing on same information with FEB will just append the data. Is it possible in any way ?

Thanks,

Vivek

[former_member376301]

I'm sorry to say to you that it is not possible.

For Explorer : 1 IS == 1 index (in fact, there may have more than 1 index because index is timestamped.). Explorer cannot execute severals indexing for one IS (with different value parameters - month value - ) and 'aggregate' index files into one. It is not so agile.

Explorer cannot do 'incremental' indexing

(It could be an improvement to suggest ...)

Regards,

[former_member376301]

To add some info about your issue...

Why your BW generates 'memory dump' ? Maybe you should increase memory or else (look at Sizing Guide if you have it) ?

I'm not an expert on BW ...

Regards,

[Former Member]

Thanks for the suggestion Arnaud, we are trying to get that also.

Just another thought Is below method possible in any way:-

Bex query 1 ----> Universe 1

                                                         Information Space (Universe 1 & Universe 2)

Bex query 2 ----> Universe 2

Create two BEX queries, then create two universes and create 1 information space on top of two universes.

I know we cannot create a single information space on two different universes. but do you have any suggestion if by any mean we can achieve this?

Regards,

Vivek

[former_member376301]

Did you try to create 1 universe (UNX) in Multisource mode with IDT  ?

You can create an Universe with a data foundation using severals connections (even if they are the same connections) and may be you could customize each of them with yours Bex queries (?)

Regards

[Former Member]

Actually Jco connection with SAP BW cube is not working in our setup !!! There is some technical issue.

Again and requirement is to create information space on two cubes.

So I thought if we can achieve this using UDT ?

Regards,

Vivek

[former_member376301]

It is not possible to create 1 IS on top of severals Universes/cubes/datasources.

But did you try to 'insert universe into universe' with UDT ?

If Universe1 and Universe2 are 2 differentes Bex queries, you could 'link' universe1 to universe2

(maybe n sub bex queries are better than 1 big bex query ?)

With UDT -> [Insert] -> [Universe...]

Another idea (worse) could be to have several IS and all of them could be integrated into 1 ViewSet (1 exploration view per month/IS). I don't know how do you want to use/see yours datas ....

Regards,