Optimizing Dimension and Cube Processing - SQL Server Analysis Services 2012 Cube Development - page 172

Learning advanced dimension

processing options

In the previous recipe, you processed a few dimensions after altering the degree of
parallelism. You can specify additional options to handle any errors you might en-
counter during processing from the same dialog using the Dimension key errors tab.
You should handle a majority of the data quality issues in the Extraction, Transform-
ation, and Loading ( ETL ) layer of your application. However, in some cases it's ac-
ceptable to ignore processing errors, particularly during the development stage when
your ETL routines might be half-baked. The following table lists the remaining dimen-
sion processing options you can specify:

Processing
option Description

Key error
action

The default action in case of an error is to convert the attribute key to
the unknown member. The other alternative is to discard the record.
Though this is a viable alternative during the development phase, you
should be careful with discarding dimension records in production
since this will also cause Analysis Services to exclude partition data
for any ignored dimension records, and by doing so would lead to in-
accurate reports.

Processing
error limit

The Ignore errors count value ignores all errors. The other
value is Stop on Error . This value allows you to specify the num-
ber of errors to ignore before you stop processing or stop logging er-
rors.