Question:
Hello,
When you extract transactional data from R/3, would you need to extract data that exists as master data on the BW side??
For example, the InfoObject 0EMPLOYEE has the employee number and other attributes such 0EMPLGROUP, etc. When I extract my transactional data, would I need to also include the employee number and employee group from the R/3 side to be loaded into my infocube??
I am getting requirement to extract the employee numbers and the employee group for the employees along with the transactional data.
How does this work?? I would think that you only need to extract the employee number along with the transactional data and not the employee group. How are the transactional data linked to the master data?
Thanks
Tess
Answer:
You extract transaction data and master data separately. You have an option, when you update transaction data, to update even when no master data exists.
Master data may be modified from time to time in R/3. So you must update master data as well but maybe on a less frequent basis than transaction data (which may be daily, hourly etc.)
You can see all attributes of master data in the reporting tool, provided they have been activated in the InfoCube. If master data is not updated on a timely basis you would not get to see the attributes, but only, as in your example, employee number. Master data infosources are in general attributes, texts and hierarchies.
Hope this gets you on your way.
_________________
Keep smiling
Answer:
For transactional data then, I would need to also extract the employee number as well because the transactional data contains the hours worked, for example, for each employee.
Essentially, then I would still need to extract the employee number for the corresponding hours worked that is considered transactional data. But the employee group need not be extracted because that can be referenced by the master data attributes of the infoobject 0EMPLOYEE.
Is my understanding and approach correct??
Thanks
Tess
Answer:
You are understanding is correct.
Other than that SR has answered very well.
Answer:
Hi,
Take care that in some update rules (for example 0HR_PA_CM_2), master data (active) is read to fill infoobjects in the cube, so HR master data should first be loaded before transactional data.
Best Regards,
Olivier
Answer:
Dear guest, you should consider reading some stuff around the so-called "star schema".
Ch
_________________
_
There are only 10 types of people in the world :
those who understand binary and those who don't.
Answer:
Dear CHC,
I read the STUFF but my colleagues are adamant that the employee group should be extracted along with the transactional data for each employee's hours worked. Thank you for your suggestion.
Now if you would want to help. How would I go about in telling my colleagues and users that we don't need to extract the employee group along with the transactional data?
Tess
Answer:
Having employee group as both master data and transactional data does not make sense as you are duplicating data. Select either one or the other but I would go for master data as you can share such information across cubes. However, your data model is dependant on your report requirements, data extraction, performance...etc.
Each situation is unique and there is no universal answer....unless you're using 100% business content but who uses that anyways
Regards,
EL
Answer:
I think you may need to answer them by doing exactly what they wanted do. They might be asking the right thing, because, I consider this as data modelling question.
This is what I would do.
1. First explain the following to them.
1.1 If the employee group is retrieved with the transaction code you can not save it to the 0EMPLOYEE info object. So, you want to extract with the transaction and DO NOT use it in the transfer rule.
1.2 If they say, they want to store it in the cube, then you should explain to them, that they will see 2 types of records, one without employee group, and one with #. The totals will match.
1.3 If they agree with this, then you might consider the time taken to modify the cube to include the gorup and reactivate. ( remember it exports the data, inactivates the cubes, drops the tables applies the changes and activates the cube ( rebuilding the tables) and import the data ). I am not sure in a large infocube, what would be the effect of the change????
1.4 Step 1 & 2 you must try on a dummy cube with less data and show the results to them. To do step 3, you might take the backup prior to changing the cube, if it is a large cube.
2. Consider this second scenario. If the employee changes the employee group often, then, you might want to see the results for different employee groups iinstead of current employee group. So keeping employee group in info object does not serve the purspose.
In my opionion, if you explain to them clearly and with a demo, they may decide either way.
Hope I have not provided too much information. Please remember that this is good modeling question and we like to hear from you.
Answer:
The remark from William is very important : all depends on the time dependency !!!.
If employee group is a time-dependent attribute of employee (which it is in the BCT), then it can be very useful to have the characteristic as well in master data as in infocube to give a solution for various scenario's, such as :
historical truth
today is yesterday
...
what you do depends very much on the user requirements
kind regards,
bene
Answer:
My understanding is that the master data load that we perform weekly should take care of the time dependency for the employees and their corresponding employee group (Employee status), that is if we have been loading it correctly; therefore, using only the master data should satisfy the model for the users.
I have seen that the navigational attributes can be defined for the employee group and it will then appear as data in the infocube if I review the data from the data content infocube level. In a way, if we extracted the employee group/status along with the transactional data into the infocube and also use the navigational attributes, we would see both columns with the same data.
What do you think?
Thanks
Tess
Answer:
Hi Tess,
Your general understand is correct.
You could have a lot of fun answering irate customers, if you use both nav. attribs as well as the object! It happened to me once when I turned on the nav. attrib. and forgot to delete the object from the infocube. People started using the object instead of the nav. attrib (because both show up under dimensions in the query design view) and were gettign unexpected results.
With best regards.
_________________
Keep smiling
Answer:
Bene, I don't understand your remark :
if the employee group is time dependent, I don't see the value-added of loading employee group directly into the cube ?
I understood that you should load it directly into the cube only if there is no time dependency for the considered attribute.
But perhaps am I wrong, of course, as usual...
Ch
_________________
_
There are only 10 types of people in the world :
those who understand binary and those who don't.
Answer:
I would conclude then that using the attributes 0EMPLOYEE_0EMPLGROUP as navigational attributes from infoobject 0EMPLOYEE should satisfy the requirement of reporting based on the employee group/status.
Another issue we have is that the 0EMPLOYEE_0EMPLGROUP is a one byte value. The users want to have a meaningful description to choose for reporting. How can this be achieved? Would the user need to choose the numerical code or the description of the empoyee goup/status code??
Thanks
Tess
Answer:
Go go BE admin workbench and select 0EMPLGROUP ( Hope this is your attribute)
Go to Business Explorer tab - General Settings- select Display and select Text ( I believe Text is short description ).
This will provide a selection of employee group by Short descrtiption instead of code.