Hi,

Following are the possible changes in cube/dimension.

Dimension: Addition/Deletion or updating of some of the rows
(members/attributes)
Cube : Addition/updating of some of the rows in the fact table.
Roll-up formulea: Changes in these formulea.
Member display name : Change in display name by changing the formatting
function being used on the member display column.

For each of these changes, need to find out the mode of processing and
then create the DTS package accordingly. So instead of creating such a
complex DTS package depnding on the type of change, I just want to use
ProcessDefault on dimension (if any change) and cube. Could somebody
elaborate on ProcessDefault more? i.e. how much reliable is that
processing mode?

Thanks in advance
Gayatri.

The definition on ProcessDefault is: It will bring any object to the fully
processed state.
It will not detect any of the changes you have done to the underline data.
It will not detect any of addition\deletion of the members in the dimension
and so on..
In case and changes needs to be made to the dimension please keep in mind 2
processing options:

ProcessUpdate - will read all the data for the dimension and will compare it
to the existing dimension structure allowing for adding/deleting members and
changing display names.

ProcessAdd for dimension will allow for adding members. It is bit
complicated to implement please see my earlier reply to thread "ProcessAdd
on a dimension" started on 2/22/06

Adding new data to partitions. You can create a new partition for only new
rows and run ProcessFull on it.
Here is whitepaper can help you on the subject:
http://msdn.microsoft.com/library/de...asprocarch.asp

Edward.
--
This posting is provided "AS IS" with no warranties, and confers no rights



"gayu" <gayatri.kulkarni (AT) gmail (DOT) com> wrote

ProcessDefault doesn't know about the changes in source data. It only knows
the current state of the AS object.

If you do ProcessDefault on an unprocessed dimension, it will do
ProcessFull. If it was already processed, it will do ProcessUpdate. For
partitions, ProcessDefault should map to ProcessFull. Etc.

The server can't infer what changes may have happened to your data since the
last processing operation and therefore you should be specific when it is
necessary...

Thanks,
Akshai
--
This posting is provided "AS IS" with no warranties, and confers no rights
Please do not send email directly to this alias. This alias is for newsgroup
purposes only.

"gayu" <gayatri.kulkarni (AT) gmail (DOT) com> wrote

Correction: "If it was already processed, it will do ProcessUpdate." should
read: "it will not do anything".

As Edward's posting indicates, ProcessDefault will essentially bring each
object into a processed state.

Thanks,
Akshai
--
This posting is provided "AS IS" with no warranties, and confers no rights
Please do not send email directly to this alias. This alias is for newsgroup
purposes only.

"Akshai Mirchandani [MS]" <akshaim (AT) online (DOT) microsoft.com> wrote

Thanks a lot for your guidance.
Also, if the requirement is like, its possible that older measures data
can be updated in the fact table then here the cube must be reprocessed
in ProcessData mode?

Regards,
Gayatri.

If your data changes in a particular partition, then you would typically
fully reprocess that partition (ProcessFull) . If all that happens is data
is "added", then you could do ProcessAdd.

ProcessData implies that data is re-processed but aggregations and indexes
are not built -- so you would have a perf degradation if you didn't
explicitly build indexes.

Thanks,
Akshai
--
This posting is provided "AS IS" with no warranties, and confers no rights
Please do not send email directly to this alias. This alias is for newsgroup
purposes only.

"gayu" <gayatri.kulkarni (AT) gmail (DOT) com> wrote