On an SSAS OLAP "deploy", can anyone help point me to which Oracle data column is having the issue... I had no problems utilizing Oracle.net data source, until I switched the FACT table to utilizing an Oracle VIEW... I have removed all MDX calculations, and changed every numeric column I can find to a DOUBLE-precision, and have narrowed my view content down to only the current month, and still encountering the problem... The view worked before I tried to solve a zero-divide issue with a COALESCE/NULLIF combination of functions, as follows:
coalesce(nullif(producing_days_qty,0),days_in_month_qty)
...which defaults to the total count of days in the month, if the producing_days_qty is either NULL or zero... However, the OLAP cube does build, when I limit it down to a single customer
The problem could be in several places.
Overflow could happen in the way View is defined.
Another place to check is definition of the measure in Analysis Services measure group. See what is the data type of the measure you have. try and see if changing that to bigger datatype solves your problem.
Edward Melomed.
--
This posting is provided "AS IS" with no warranties, and confers no rights.
Yes, the Oracle OCI-22053 problem was solved by changing both the source datatype and the datatype of the measure to "DOUBLE"... Alternatively, the ROUND/TRUNC functions can be utilized to reduce the significant digits, as it seems to carry ALL the digits to the right of the decimal point, no matter how insignificant they appear...
Moral to the story: Analysis Services handshakes much better with Microsoft SQL Server, than with Oracle, even with a .Net connection...
No comments:
Post a Comment