Guidelines for Validating a Logical Data Model

From MIKE2 Methodology

One of the key activities of ensuring the LDM was successfully designed to meet the requirements is through the process of validating the LDM contents and relationship as aligned to business requirements, questions, reports and or queries. This is necessary in order to reach mutual agreement on the LDM content and accuracy, and to proceed to the next step, physical design of the data structures.

Formalise the Validation Process

A formal data modelling validation meeting needs to be arranged between the data modeller and required end user staff. If following a requirements or hybrid requirements driven approach, the need for IT related staff should be minimal, as the objective is to ensure it addresses business (logical) requirements verses physical (data base) or data access requirements (SQL complexity or performance). Such items are catered for in the physical data model design. The scope of the activity to pursue is identified below.

• Prepare an overview of the history (purpose/activity), approach, scope, content and known issues/challenges faced. This will be presented during the validation workshop.
• A review of the essential entities and or Dimensions and Facts and their contents in the LDM and the relationships between them will be described.
• Where there are hierarchies involved, consider creating a mini prototype to demonstrate how a hierarchy within the LDM is referenced to depict a data grouping. This will help visualize what the end result will look like. Hierarchy structures should be explored as this will have implications as to design requirements of the ETL design and loading activity and if OLAP tools are involved.
• Select business requirements examples (reports/queries), will be applied by the end user to the LDM to see if they can be supported by the LDM.
• Add-hoc questions on the model contents, structure, relationships, definitions should be pursued.
• Any issues along with action items will be documented within the review minutes.

Note: Data types and lengths will still require verification along with enhancement of metadata further describing the LDM entities and attributes.

Preparing for the Validation Session

Adequate preparation is essential to ensure the LDM validation session is highly effective and will not require multiple validation sessions to occur as a result of inadequate attendance, lack of information, or content. Examples of this are:

End user:

• Prepare validation questions and queries (i.e. leverage reports, etc), from items previously provided to the Data Modeller and used to derive the LDM. This will be used to provide focus when navigating through the LDM to confirm that all the necessary information

’ s needed are present in the LDM. This will also identify any gaps in the LDM that may need to be addressed still

• Review the LDM regarding the relationships and attributes. Obtain a cursory perspective if it is a adequate representation of the information needed to support the requirements which were provided as input into the LDM.

Modeller:

• Ensure relevant end user resources are there to validate the requirements given and to accept (or approve) the LDM as per its current state.
• Fully attributed LDM Diagram to support the data elements, hierarchies and business relationships in support of the business requirements.
• Entity/Attribute definition report.
• Summary listing of Data Modelling considerations (of interest) and overview of methodology employed.
• Sample of reports used as input in to the LDM
• Sample business questions or queries used as input in to the LDM

Get Confirmation that Participants Accept the LDM

It is imperative that the LDM validation session is documented to include who was in the meeting, what was discussed, issues, if there are gaps, how, when and by who will resolve the gaps.

If the consensus by the end user is that the LDM in principle address their requirements from a content and accuracy perspective, capture this within the meeting notes to infer acceptance of the LDM. This is one of the few mechanisms, which can influence gaining agreement (sign-off) of the LDM schema and help freeze (stop) request to expand the model with additional "last minute" requirements. Any additions to the model after agreement can be considered as change requests.