DEA Architecture Model (OpenDEAM) Framework

OpenDEAM (Open Digital Enterprise Architecture Model) is the root authority for the TechNeHub Labs DEA architecture: layers, building blocks, entity allocation, and relationships.

What this is

One file — model/opendeam-model.yaml — defines:

1. Architecture parts: 5 layers → building blocks (first-level components).
2. Object allocation: every first-level entity mapped to exactly one layer + building block (MECE), with its catalog repository. Abstract entities may be realised by subclasses in other layers (ADR-0002 D3).
3. Relationships: typed edges between entities (rel_type + cardinality).
4. Orthogonal allocators: the ECF 7×7 city-block matrix (owned by dea-metaframework), and the Measurement Dimension (cross-cutting, owned here) — referenced and exercised, never treated as layers.

Everything else in the organisation — dea-metamodel schemas and viewer graph, catalog metamodel-pointer.yaml files, README badges, the Pages viewers, dea-web-viewer — is a derived consumer of this model, version-pinned and CI-validated.

The five layers

Layer	Name	Scope	Entities
L1	Ecosystem & Value Network	External	6
L2	Strategic & Governance	Intent & Rules	12
L3	Business Operating Model	Internal	16
L4	Digital & Intelligence	Data & Brain	10
L5	Technology & Execution	Systems & Infra	7

Plus 2 dimension entities (Concept, Performance Metric) allocated to the orthogonal dimensions below, not to any layer — 54 first-level entities in total (v0.6.0).

Plus two orthogonal dimensions: the Measurement Dimension (Performance Metric — measurable entities declare measured_by, metrics declare scope_layers; ADR-0002 D1) and AI & Automation Governance (AI-driven entities may declare governed_by: [Risk/Control/Regulation]; ADR-0003 D6). See the ADRs for why these are dimensions, not layers.

Derived from the enterprise axiom: "An enterprise is any bounded entity that persists by exchanging value with its environment." See docs/allocation-rationale.md.

Terminology (binding)

• Layer — architecture layer L1–L5. Defined only here. Measurement is NOT a layer (v0.2.0+) — it is an orthogonal dimension.
• Tier — repository-stack position T0–T3. Never call a tier a "layer".
• Building block — named grouping of entities within a layer.
• Entity — first-level architecture object, exactly one layer + one building block (unless abstract or a dimension entity).

Consumption contract

Consumers pin a model tag and validate against it:

# In a consumer repo's workflow (e.g. dea-catalog-processes)
jobs:
  allocation:
    uses: technehub-labs/dea-architecture-framework/.github/workflows/validate-against-model.yml@v0.6.0
    with:
      model_ref: v0.6.0
      pointer_file: metamodel-pointer.yaml

• PRs that drift from the pinned model (wrong layer, stale alias, unknown entity) fail CI.
• A central drift-detection cron diffs all consumers against the model and opens issues (Phase 4).

Change control

Change	Requirement
New/removed layer, new dimension, re-allocation	ADR + minor version bump
New entity / building block / relationship	PR + patch version bump
Every PR	scripts/validate_model.py passes

See docs/ADRs/ — 0001 governance · 0002 structural upgrade (v0.2.0) · 0003 comprehensiveness expansion (v0.3.0) · 0004 L2 cleanup & vocabulary purge (v0.4.0) · 0005 metamodel governance + Resource & Information/Knowledge (v0.5.0).

Downstream sync status (2026-08-16)

• v0.5.0 (ADR-0005) added Resource (abstract) with three specializations (Financial / Physical / Intangible, L3), Information Asset and Knowledge Asset (L4), and the entity_role / completeness_contract governance fields.
• Consumed downstream by dea-metamodel v0.6.0–v0.9.0 (CR-001 canonical metamodel, CR-002 relationship ontology, CR-003 normalization, CR-004 core ontology, 2026-08-16). The CR-004 Core/Profile split introduces 10 new core anchors that are not yet in this root model — promoting them upstream is the candidate content of the next minor release (v0.6.0, ADR forthcoming).
• Known pin drift, documented not silent: T2 catalog metamodel-pointer.yaml files currently pin v0.2.1 (auto-generated, do-not-edit). They regenerate from this repo's fan-out when the next root-model release lands — not by hand-edit in catalog repos.

Repository map

T0  dea-architecture-framework   ← this repo (root model)
T1  dea-metamodel                ← machine schemas, viewer graph (consumer)
T2  dea-catalog-*                ← entity instance catalogs (consumers)
T3  dea-cli, dea-web-viewer, technehub-labs.github.io  ← tooling & viewers (consumers)
⟨orthogonal⟩  dea-metaframework  ← ECF 7×7 city-block matrix (referenced allocator)

Structure

├── model/opendeam-model.yaml          # THE source of truth
├── schemas/opendeam-model.schema.json # JSON Schema for the model
├── scripts/
│   ├── validate_model.py              # schema + referential integrity (CI)
│   └── validate_consumer.py           # consumer pointer drift check
├── docs/
│   ├── ADRs/                          # 0001–0005 (see Change control)
│   ├── allocation-rationale.md        # 5→6 layer mapping + boundary rules
│   └── roadmap.md
└── .github/workflows/
    ├── model-ci.yml                   # validate on every PR/push
    ├── validate-against-model.yml     # reusable consumer validator
    └── fan-out.yml                    # dispatch model changes to consumers

License

Apache 2.0 — see LICENSE.