Skip to main content

1. The MDA Approach to Deriving Value

The primary goal of MDA is to transform models from passive diagrams into active assets that manage the complexity of modern, interconnected systems.

MDA produces value through several distinct mechanisms:

  • Communicate Ideas: Models serve as the primary vehicle for shared understanding.
  • Generate Artifacts: Automatically producing code, schemas, and documentation.
  • Run Analytics: Testing the design for complexity and dependencies before building.
  • Simulate / Execute: Running the model to validate behavior early.
  • Derive New Info: Extracting playbooks and responsibility matrices.
  • Structure Data: Providing the framework for organizing unstructured information.

Models as Communication Vehicles​

A model's primary function is to facilitate shared understanding. MDA enhances this by:

  • Standardization: Utilizing standard terms, icons, and notations (e.g., UML, BPMN) so that models are interpreted consistently.
  • Persistence: Treating models as semantic data that can be versioned, audited, and shared.
  • Reusability: Providing model libraries that contain common business rules, design patterns, and vocabularies.
Models as Corporate Memory

Models serve as a permanent record of design intent and requirements, ensuring that institutional knowledge survives even as team members change.

Derivation via Automated Transformation​

MDA uses transformations to convert source models into target artifacts. This reduces manual labor and ensures the implementation remains "in sync" with the design. A transformation typically requires:

  • A source model (the abstract design).
  • Parameters (configuration for the target environment).
  • Transformation rules (the logic of the conversion).
  • A target artifact (code, DDL, XML schemas, or another model).

Model Analytics​

Because models are structured data, you can perform deep analytics before a single line of code is written:

  • Validation and consistency checks.
  • Complexity and dependency metrics.
  • Impact analysis for proposed changes.

Model Simulation and Execution​

Models can drive simulation engines to validate system behavior early in the lifecycle. In some cases, models can be executed directly by specialized engines (e.g., executing BPMN via a workflow engine or SQL DDL via a database), providing the shortest path from design to production.

Deriving Information​

From a mature set of models, tools can automatically extract:

  • Process Playbooks: Step-by-step operating guides.
  • Responsibility Matrices: Clear mapping of roles and actions (e.g., RACI).
  • Documentation: Always-up-to-date technical and functional references.
  • Procurement Specs: Inputs for RFPs and contracts.

Structuring Unstructured Information​

While MDA focuses on structure, it provides the "containers" for information extracted from unstructured sources (emails, PDFs, etc.). By mapping unstructured data into an MDA-compliant model, you turn noise into actionable semantic data.