BACKGROUND PURPOSE:
Data Modeling, Data Integration
(ETL/DI), Business Intelligence (BI) and other design tools supported by MIMB
may not always save models which are valid. Indeed, some of these models may
still be under development (not yet completed), or the model validation may be a
separate step (design tool menu entry, separate tool, or part of the model
publication). Therefore, MIMB import bridges may be dealing with semantically
incorrect or incomplete models which the MIR repository can save, but they may
break the export bridges. Having a centralized model validation (or consistency
check) between the import and export phases allows to continuously improve the
consistency check algorithms and avoids having each export bridge retest for the
same model validity issues.
INPUT PARAMETERS:
The user has the choice of 3 levels of
consistency check:
o "No Consistency Check": If the user purposely decides to turn off all consistency check (e.g. when the tool is not available to fix the source model), then the imported model may break the export bridge, or produce an invalid model that may break the target tool. MITI assumes no responsibility when consistency check is turned off.
o "Basic Consistency
Check": This level corresponds all minimal consistency checks required to
export, including (but not limited to) the detection of disconnected
relationships, foreign keys not connected to any primary or alternate key,
etc.
Note 1: the source model may pass the model validation provided by the
original tool, but still may fail the basic consistency check which may include
more substantial testing required by the target tools.
Note 2: the basic
consistency check may be more rigorous than needed by the export bridge or
target technology/tool that will be used. In such case, some of the warnings or
errors can be ignored.
Note 3: the basic consistency may not be complete
enough to catch all the model validation tests of the target tools. Therefore,
the user should still run any model validation or consistency check available in
the target technology/tool.
o "Detailed Consistency Check": This level includes a more verbose version of the above basic consistency checks, plus more advanced semantic checks specific to the technology of the source model: RDBMS (e.g. attributes of a PK cannot have a null value), ETL/DI metamodels, or ROLAP/MOLAP BI metamodels (e.g. a hierarchy must have a root). This level is sometimes used as a substitute to the lack of model validation capabilities in the original tool, but is not required to use the MIMB export bridges.
OUTPUT RESULTS:
Consistency checks will report (in the
log) "WARNINGS" on the source model, or "ERRORS" that are fatal
and will prevent the export to any tool. The solution is to fix the original
model, and then import it again.