Mapping is an important step to understanding your data and where the data resides in your ecosystem. Mapping takes us from the known to the unknown and is effectively accomplished by using mapping tools, adopting best practices, and having a common understanding of how the mappings will be used. But mapping does take a considerable amount of time and requires a person with extensive knowledge in the source or target system or both.
Our team maps from an industry logical data model (core model) to access path building blocks -- then to semantic structures such as dimensions, or lower to higher level facts.
Access path building blocks (APBBs) are designed to help the semantic modeler develop dimensions and facts (which are specific and denormalized) for the semantic data model from the core data model (which is normalized and generalized). APBBs bridge the gap between the dimensional and normalized logical data models. To help the modeler in using APPBs, construction maps are included in the SMBBs to illustrate how a dimension can be built from the core data model (the appropriate Teradata industry data model) using the access path building blocks. The following picture is an example of an APBB construction map. The green boxes represent tables in the core data model (the Teradata Financial Services Data Model, in this example), the orange boxes are the APBBs, and the white boxes are the resulting dimensions.
In this example the construction map visually layouts the data needed to identify a person who is insured by a policy (join path) and the diagram can also be understood by business analysts that may not model or write SQL.
Performing mappings helps our team identify gaps in both models. In one case, core data may need to be added to the semantic data model in the other, the core data model may need to be expanded with data required by BI reports. The gap analysis identifies the missing pieces of data need to address the business requirements.
Initially, our group performed detailed attribute-to-attribute mapping, but then switched to higher-level, entity-to-entity mappings -- to save time. The time saved on detailed attribute-to-attribute mappings, many of which are obvious, was instead focused on describing the purpose of the mappings thru filter and join notes. In the picture above Gender Type is an entity while Gender Type Description (in Gender Type) is an attribute.
We found that we could save a considerable amount of time by reusing the mappings from areas such address or product in the same industry or across industries. And using a tool to perform the mappings makes them easily reusable, adaptable, and assists in standardizing around best practices.
Mappings help with many things, including:
• Focusing on specific areas in an ecosystem
• Finding and resolving information gaps as well as design gaps
• Creating the core layer views for the semantic layer
• Establishing a reusable base for common content such as location or product
• Supporting a more precise way of communicating and refining details during design and implementation
Our team finds mapping to be useful, reusable, and educational -- and is a worthwhile investment of our time.
For mapping we use Teradata Mapping Manager (TMM) found on Teradata Developer Exchange. Information on our Industry Data Models (iDMs) and Solution Modeling Building Blocks is on www.teradata.com.
Karen Papierniak is a Product Manager responsible for development of Teradata’s Industry Solution Modeling Building Blocks and Data Integration Roadmap portfolio -- that spans eight major industries and used by customers worldwide. Karen’s roles at Teradata have been in software development, systems architecture, and visual modeling while working in a variety of industries including retail and communications.