Specific steps where variables are processed

[Former Member]

Hi everybody,

Can someone tell me where are the variables processed when executing queries.

We are on 3.1 and our queries were defined with the Query Designer (QD). A template was also developed with the Web Application Designer, to standardize all our reports.

Within QD, variables are defined under the filters, free characteristics or columns (in our case). I always tough that when the user specifies a variable (defined under the filters of QD), it was executed as a REAL filter, which means that the OLAP processor was extracting the data using that variable, and any other filters. Then, the front end tool was applying the other variables and displaying them to the screen.

My colleague, which was at the latest conference in Orlando, is telling me that this is not what is happening. In fact, the OLAP is extracting all the data, using the fixed filters and then, the front end is applying the variables. Let me tell you that I have a hard time believing in that.

Does anyone know of any detailed documentation on How the variables are processed and at which step? The only one I found is related to the user exit.

Your help will be greatly appreciated.

GCB

[Former Member]

Hi,

please have a look on the following slides from the service market place:

http://service.sap.com/~sapidb/011000358700001236832003E/OLAPAggregation.ppt

Slide 14,...

Heike

[Former Member]

Keike,

Tks a lot, this is exactly what I was looking for. According to this one, it does what I always think it was supposed to do. The other document I have let us think that everything is read, then the variables are applied.

Guylaine

[Former Member]

Sorry, I misspelled your name Heike

[Former Member]

I got a complete description of the way the variables are processed from the BW experts. They will post an article on it in july or september.

Thanks to everyone,

GCB

[Former Member]

The Query defination is clearly shown in doc..

"Variables enable you to customize queries flexibly (parameterize the query) possible. If you use variables in the Query Designer, do not select any fixed characteristic values, hierarchies, hierarchy nodes, texts, or formula elements. Instead, set up variables as place holders. These are then filled with values when the query is running (when you insert the query into a workbook, when you refresh the workbook or when you execute the query on the Web. You can use one query definition as the foundation for many different queries if you use variables."

[Former Member]

Hi,

"These are then filled with values when the query is running "

As you know, there is many steps in the query execution. This does not "clearly show" in which of them it is used.

I also got this document saying that for variables, BW retrieves all of the data, then asks users which subset they want to work with. Looking at it with what my colleague told me is really scarry. However, with the detailed one I got from Heike, I can 'assume' (I know, I should not use this word) that it is what is actually happens if you specify many variables and they are not filled. Everything is retrieve, then the Query is used to format the output to the user.

Guylaine

[Former Member]

I think you would have to investigate using traces, e.g. using RSRT to check the SQL generated and the aggregates being sought using different types of variable.

I agree that is hard to believe that BW does not filter the data as early as possible, my guess is that what was described applies to special cases involving complex processing.

[Former Member]

Robert,

I tried it already, but there is a problem with the system. I always get an SQL error. I found the OSS note and sent the problem and solution to the support group, but... since BW is not the only system they support, I do not know how long it can take.

Anybody having documented information on that would be welcome.

Guylaine