Last time we began to look at the opportunity for integrating a newly developed master data repository with existing business applications supported by a data warehouse. That meant understanding the requirements for integration and verifying that those requirements are met.
Verifying the observance of requirements involves specifying the variables that are measured, specifying the method of measurement, and specifying a level of acceptability for the measure. When it comes to the requirements for integrating a data warehouse with an MDM system, this involves identifying where the data warehouse applications touch master data, noting the type and the constraints of the relationship, and then determining that the master data dependence relationships are observed. Those dependence relationships are basically whether the master data entity’s values are searched for and located, read, and/or updated.
These assertions and dependences are in addition to typical expectations for data quality. Rather than looking at the level of trust of the specific values, they suggest expectations regarding the level of integration across the different consuming applications. Some examples are:
- Consistency – Are the data values associated with master data entities that are stored in the data warehouse comparably consistent to the corresponding master data entity values in the master data repository?
- Synchronization period – How frequently is there a data refresh performed between the two systems?
- Timeliness – Is the data in the data warehouse delivered from the master data repository within the specified time period (e.g. on a daily basis)?
- Coherence – Are other copies of data values taken from the master data repository coherent with the copy in the data warehouse?
The real questions focus on whether data warehouse applications access a master data entities and the parameters for ensuring the collection of accesses are consistent and whether results can be replicated. Next time: master data model requirements.