The Architecture Description Standard (ADS) defines how to write a Solution Architecture Document (SAD) — the single document that describes how a technology solution is designed, built, and operated.
This standard defines the required structure and content for Solution Architecture Documents (SADs). The architectural views and quality attributes within a SAD constitute the High Level Design (HLD) of the solution.
A Solution Architecture Document conforming to this standard serves the following functions:
Communication — Provides a shared understanding of the solution architecture across all stakeholders
Governance — Enables the architecture governance process to evaluate the design against quality criteria
Traceability — Establishes links between design decisions, business requirements, quality attributes, and applicable standards
Reference — Acts as a living document that accurately describes the current-state architecture
Evidence — Demonstrates that the solution is well-architected against industry-recognised frameworks
This standard applies to the documentation of any technology solution architecture, regardless of:
Deployment model — cloud-native, on-premises, hybrid, multi-cloud, or SaaS
Scale — from single applications to multi-application ecosystems
Industry — the standard is organisation-agnostic and sector-neutral
Lifecycle stage — new builds, migrations, modernisations, or incremental changes
The standard is extensible — organisations can add custom sections, map to internal tools and standards, and define governance gates without modifying the core structure. See Adopting ADS for a rollout guide.
The architecture is described through complementary views, each addressing different stakeholder perspectives. These are based on Kruchten’s 4+1 Architectural View Model — a widely-used framework that organises architecture documentation into Logical, Process, Development, and Physical views, plus Scenarios that tie them together — extended here with dedicated Data and Security views.
Together, these views form the High Level Design (HLD) of the solution:
View
Viewpoint
Primary Stakeholders
3.1 Logical View
Application architecture, components, patterns
Architects, Developers
3.2 Integration & Data Flow View
Data flows, integrations, interfaces
Integrators, Architects
3.3 Physical View
Deployment, hosting, networking, environments
Infrastructure, DevOps
3.4 Data View
Data stores, classification, privacy, lifecycle
Data Architects, Compliance
3.5 Security View
IAM, encryption, monitoring, threat model
Security, CISO, Compliance
3.6 Scenarios
Key use cases, architecture decision records
All Stakeholders
No single view provides a complete description of the architecture. Together, the views provide a holistic picture addressable by all stakeholder concerns.
Describes how the solution is developed, deployed, operated, and eventually retired — including CI/CD, migration strategy, resourcing, release management, and exit planning.
Captures the constraints, assumptions, risks, dependencies, and issues that shaped the design. Documents key architecture decisions (ADRs), guardrail exceptions, and compliance traceability.
Reading order: A completed SAD tells a story. It begins with the business context and who cares about the solution (Sections 0-2). It then describes the architecture itself through multiple views (Section 3). It evaluates how well the architecture performs against quality attributes (Section 4). It explains how the solution is built, deployed, and operated (Section 5). Finally, it records the decisions, risks, and governance that shaped the design (Section 6).
The diagram below shows how the three tiers work together. Architectural views describe what the solution is. Quality attributes assess how well it performs. Lifecycle and governance cover how it is built, run, and governed.
This standard assigns each section a documentation depth — Minimum (SHALL), Recommended (SHOULD), or Comprehensive (MAY) — that defines when it must be completed. This enables progressive adoption: begin with Minimum for early-stage designs, expand to Comprehensive for critical and regulated systems.
See Conformance and Usage for the full depth table, terminology definitions, and governance gate mapping.
Architecture documentation is inconsistent across the industry. Teams use a patchwork of templates — some too abstract to guide real work, others too platform-specific to generalise. After analysing widely-used frameworks and real-world architecture documents, gaps appear in every single one.
No existing template provides guidance on how much detail is needed for different project scales. ADS defines three tiers (Minimum, Recommended, Comprehensive) with RFC 2119 keywords (SHALL / SHOULD / MAY) so teams right-size their effort.
Machine-Readable Schema
Every template reviewed is prose-only. ADS provides a JSON Schema with atomic, enumerated fields — enabling automated completeness checks, cross-SAD comparison, and multi-format generation (Markdown, Word, web).
Dedicated Security View
Most frameworks have no dedicated Security View. Security is either buried in cross-cutting concepts, treated as a quality pillar, or absent entirely. ADS treats security as a first-class architectural view (Section 3.5).
Dedicated Data View
Most frameworks lack a structured Data View. Data architecture is scattered across other views or missing. ADS provides a dedicated Data View (Section 3.4) covering data models, flows, classification, retention, and sovereignty.
Cloud WAF-Aligned Quality Attributes
No existing documentation template aligns quality attributes to the cloud Well-Architected Frameworks. ADS maps cross-cutting quality attributes (Section 4) to AWS, Azure, GCP, Oracle, and IBM.
A structured, fill-in template that removes the blank-page problem. Atomic fields mean less ambiguity. Depth levels mean you write only what’s needed for your project’s scale.
Architecture Governance
A consistent standard to review SADs against. Machine-readable schema enables automated completeness checks across a portfolio of architecture descriptions.
Delivery Teams
Clear, structured architecture documentation that’s navigable by role — infrastructure engineers read the Physical View, security teams read the Security View.
Organisations
Comparable architecture descriptions across projects. Portfolio-level analysis of quality attributes, risk posture, and compliance coverage.
ADS is open-source and hosted on GitHub. Contributions are welcome — submit issues, suggest improvements, or open a pull request. The standard content is licensed under CC BY 4.0 and the source code under MIT.