
IT-Security Architecture Frameworks
How TOGAF, the Microsoft Cloud Adoption, Well-Architected and Security Adoption Frameworks, MITRE ATT&CK and the MCRA fit together, and how they help to build an enterprise-wide IT security strategy.
Contents
How to navigate IT-Security frameworks?
In a recent workshop about IT-Security & Security-Architecture, one of my customers asked how all of those frameworks - TOGAF, CAF, WAF, SAF, MCRA, MITRE ATT&CK, plus standards like NIST CSF, ISO 27001 or the CIS Controls - fit together. They rightfully said that, at first glance, the frameworks seem to compete with each other - but actually they don’t. Each one answers a different question at a different level of detail, and together they form a chain from the enterprise strategy down to a single detection rule in the SOC.
In this post I’ll try to give a brief overview of six frameworks, from generic to specific, and then show how they relate and how to use them together for an enterprise-wide security strategy:
- TOGAF - The Open Group Architecture Framework
- CAF - Microsoft Cloud Adoption Framework
- WAF - Microsoft Azure Well-Architected Framework
- SAF - Microsoft Security Adoption Framework
- MITRE ATT&CK - Adversarial Tactics, Techniques and Common Knowledge
- MCRA - Microsoft Cybersecurity Reference Architecture
The Frameworks in Brief
TOGAF
The TOGAF Standard (The Open Group Architecture Framework) is the best-known method and framework for enterprise architecture. The current version is the 10th Edition, which is split into the Fundamental Content (core concepts, the method and its artifacts) and the Series Guides (guidance for specific topics and ways to tailor the method).
Its core is the Architecture Development Method (ADM), an iterative cycle of phases with Requirements Management at the center:
- Preliminary Phase: Architecture Principles, Architecture Repository, Business Principles, Goals, and Drivers, Organizational Model for Enterprise Architecture, Request for Architecture Work, Tailored Architecture Framework
- Phase A: Architecture Definition, Architecture Vision, Capability Assessment, Communications Plan, Statement of Architecture Work
- Phase B: Architecture Requirements Specification, Architecture Roadmap
- Phases C and D: Architecture Deliverables
- Phase E: Implementation and Migration Plan
- Phase F: Architecture Building Blocks, Architecture Contract, Implementation Governance Model
- Phase G: Compliance Assessment, Solution Building Blocks
- Phase H: Change Request, Requirements Impact Assessment
TOGAF is vendor- and technology-neutral. It doesn’t tell you which firewall to buy, but how to get from business goals to a governed target architecture and roadmap. Also, security isn’t a separate phase: it is a concern in every phase, and The Open Group publishes guides and templates on integrating risk and security into the ADM.
Microsoft Cloud Adoption Framework (CAF)
The Cloud Adoption Framework is Microsoft’s guidance for decision makers who adopt Azure. It is organized into seven phases, each answering one key decision:
- Strategy: What should our Azure adoption look like?
- Plan: How will we prepare for it?
- Ready: How will we build our Azure landing zone?
- Adopt: How will we migrate, modernize and build workloads?
- Govern: How will we control our Azure environment?
- Secure: How will we protect our Azure environment?
- Manage: How will we operate and optimize Azure over time?
The main technical result is the Azure landing zone: a platform landing zone with central governance, security and shared services, plus workload landing zones that inherit the platform’s guardrails (Azure Policy, RBAC, networking). The Secure methodology runs through every phase and is aligned with Zero Trust.
Microsoft Azure Well-Architected Framework (WAF)
The Well-Architected Framework looks at a single workload. It is built on five pillars: Reliability, Security, Cost Optimization, Operational Excellence and Performance Efficiency. Each pillar provides design principles, a design review checklist and documented tradeoffs between the pillars.
The Security pillar is based on Zero Trust (verify explicitly, least privilege, assume breach) and the CIA triad (confidentiality, integrity, availability). It covers security readiness, confidentiality, integrity, availability and how to sustain the security posture over time, including threat modeling, an asset inventory, penetration tests and incident response.
WAF and CAF complement each other: Microsoft itself says that CAF is for decision makers and centralized controls across a portfolio, while WAF is for teams that design a specific workload.
Microsoft Security Adoption Framework (SAF)
The Security Adoption Framework is Microsoft’s operating model for security modernization based on Zero Trust. Unlike CAF and WAF, it isn’t limited to Azure: it covers hybrid, multicloud and multiplatform environments, including OT/IoT and AI.
SAF connects four components:
- Business scenarios define why security investments are needed (e.g. securing remote work or protecting critical assets).
- Security disciplines define who is responsible. There are nine of them in three categories: planning and oversight (Strategy, Integration and Governance; Security Architecture), technical strategy (Access and Identity; Infrastructure, Development, Data and OT/IoT Security) and operations (Security Posture Management; SecOps). See the disciplines overview.
- Technologies define what to protect: identities, endpoints, applications, data, infrastructure, networks and AI.
- Technical solutions define how protections are implemented and operated.
SAF doesn’t replace existing standards. It provides mappings to NIST CSF 2.0, the CIS Controls, The Open Group Zero Trust model, the CISA Zero Trust Maturity Model and the DoD Zero Trust Strategy.
MITRE ATT&CK
MITRE ATT&CK (Adversarial Tactics, Techniques and Common Knowledge) is a vendor-neutral, publicly available knowledge base of real-world attacker behavior. It has matrices for Enterprise, Mobile and Industrial Control Systems (ICS). The Enterprise matrix organizes tactics (the attacker’s goal, e.g. Initial Access, Privilege Escalation, Lateral Movement, Exfiltration) into techniques and sub-techniques (how the goal is reached), linked to known threat groups, software, mitigations and detection strategies.
ATT&CK is updated twice a year. Version 20 (October 2026) contains 15 Enterprise tactics with 227 techniques and 479 sub-techniques. Since v19, the former Defense Evasion tactic is split into Stealth and Defense Impairment, and v20 adds AI-related techniques such as prompt injection.
ATT&CK is not an architecture framework by itself but it is commonly referenced by other security frameworks. It is the attacker’s view, a shared vocabulary for threat modeling, detection engineering, red and purple teaming, and for measuring how well controls cover real attacks.
Microsoft Cybersecurity Reference Architecture (MCRA)
The MCRA is the most detailed layer: a large set of technical architecture diagrams (published as a PowerPoint deck, latest version June 2026) for the “hybrid of everything” estate. It covers Microsoft security capabilities, Zero Trust user access, security operations, operational technology (OT), multicloud and cross-platform capabilities, attack chain coverage, infrastructure and development security, data security, AI, and security roles. It also maps Microsoft capabilities to Zero Trust standards such as The Open Group’s Zero Trust Reference Model.
The MCRA is part of SAF. Microsoft suggests using it as a starting template for a target security architecture, as a comparison with what you already own (organizations often find capabilities they didn’t know they had), and as a learning tool.
How the Frameworks Relate
The frameworks differ in two ways: in scope (enterprise, cloud estate, workload) and in level of detail (method, principles, reference architecture, techniques). ATT&CK additionally switches perspective (defender vs. attacker). Put side by side, from generic to specific:
| Framework | Key question | Main audience | Typical result |
|---|---|---|---|
| TOGAF | Where does the enterprise want to go, and how? | Enterprise architects, CIO | Architecture principles, target architecture, roadmap |
| CAF | How do we adopt and run the cloud? | Cloud center of excellence, platform team | Cloud strategy, landing zones, governance policies |
| WAF | Is this workload well designed? | Workload team, solution architects | Design decisions, reviewed tradeoffs |
| SAF | How do we organize and prioritize security work? | CISO, security architects, business owners | Security operating model, scenarios, disciplines |
| MCRA | What does a complete security architecture look like? | Security architects and engineers | Target state, gap analysis |
| ATT&CK | What do attackers actually do? | SOC, red and blue teams, architects | Threat models, detection coverage |
Please note that it isn’t a strict hierarchy, though. A closer look shows two strands that cross each other:
- The general architecture strand narrows the scope: TOGAF looks at the whole enterprise, CAF at the cloud estate, WAF at a single workload. Each level works within the guardrails of the one above: a workload designed with WAF lives in a landing zone built with CAF, and the cloud strategy is one part of the enterprise architecture.
- The security strand gets more concrete: SAF defines the security strategy and operating model, MCRA turns it into technical capabilities, and MITRE ATT&CK describes the threats those capabilities must handle.
Where the strands cross, the general frameworks have their own security view: TOGAF has its security Series Guide, CAF the Secure methodology, WAF the Security pillar. These views refer to SAF and the MCRA for depth. CAF, for example, recommends the CISO and MCRA workshops for its security strategy. Seen as a matrix:
| Level | General architecture | Security |
|---|---|---|
| Enterprise | TOGAF ADM | TOGAF security architecture, SAF strategy discipline |
| Cloud estate | CAF | CAF Secure methodology, SAF |
| Workload | WAF | WAF Security pillar |
| Capabilities and products | Azure Architecture Center | MCRA |
| Threats | MITRE ATT&CK |
Two things connect all levels:
- Zero Trust is the common set of principles. CAF’s Secure methodology, the WAF Security pillar, SAF and the MCRA are all built on it, and The Open Group (the publisher of TOGAF) is standardizing a Zero Trust reference model that SAF and MCRA map to.
- MITRE ATT&CK is the common threat vocabulary. It informs threat modeling in WAF, prioritization in SAF’s posture management and SecOps disciplines, the attack chain views in the MCRA, and day-to-day detection engineering. Microsoft Sentinel, for example, shows detection coverage on the ATT&CK matrix.
From Frameworks to an Enterprise-Wide Security Strategy
So how do you use all of this without drowning in documentation? The following approach works top-down for planning and bottom-up for feedback.
1. Anchor Security in the Enterprise Architecture (TOGAF)
Security belongs in the enterprise architecture from the start, not in a separate document that nobody reads:
- Preliminary phase: define security principles (e.g. Zero Trust, “secure by default”) as architecture principles, and set up the governance guardrails.
- Phase A: capture the business risks and risk appetite as part of the architecture vision.
- Phases B to D: classify information and business processes, define trust boundaries, and include identity and security services in the technology architecture.
- Phases E to G: prioritize security work packages in the roadmap and check projects against the security architecture.
- Phase H: feed new threats and incidents back as change requests.
2. Build the Security Operating Model (SAF)
SAF translates the enterprise view into a security program: pick the relevant business scenarios, assign the nine disciplines to teams and roles, and choose a starting point (strategy, an urgent risk, a business initiative, or a standard you have to comply with). If you already use NIST CSF or the CIS Controls, the SAF mappings link their objectives to disciplines and technologies instead of creating a parallel structure.
3. Build Secure Platform Guardrails (CAF)
For the cloud, CAF turns policies into enforced platform controls: the platform landing zone centralizes identity, networking, logging and security services, and Azure Policy enforces the security baseline on every workload landing zone. This is where “secure by default” is implemented, because workload teams inherit the controls instead of building them again.
4. Design Secure Workloads (WAF)
Workload teams use the WAF Security pillar and its checklist in design reviews: threat modeling, data classification, encryption, least-privilege access, secure supply chain, and an incident response plan aligned with the central SOC. WAF also forces an honest look at tradeoffs, e.g. when security controls add complexity or cost.
5. Define the Target Security Architecture (MCRA)
The MCRA serves as a template for the target security architecture across identity, endpoints, data, infrastructure, OT and SecOps. Use it as a reference for comparing it with your current tools to show gaps and overlaps. It also includes third-party integrations, and not every capability is needed everywhere.
6. Validate Against Real Threats (MITRE ATT&CK)
Finally, check whether the architecture holds up against real attack scenarios:
- Identify the threat groups and techniques that are relevant to your industry.
- Map preventive and detective controls to ATT&CK techniques and look for gaps.
- Test with red/blue-team exercises and attack simulations.
- Track detection coverage over time, e.g. on the Sentinel MITRE ATT&CK page.
Feedback loop
The gaps found in step 6 flow back up: as posture management findings in SAF, as new guardrails in the CAF landing zone, as design changes in WAF reviews, and in the end as change requests in TOGAF’s Phase H.
Pitfalls
- Don’t adopt everything at once. Start with the level where the pain is greatest: an urgent risk (SAF, ATT&CK), a cloud migration (CAF) or a critical workload (WAF).
- Frameworks aren’t compliance. SAF states this explicitly: good security makes compliance easier, but alignment with a framework doesn’t prove compliance.
- Frameworks change, tools lag behind. CAF renamed “application landing zones” to “workload landing zones” in October 2026. ATT&CK v19 split Defense Evasion in April 2026, and at the time of writing, Microsoft Sentinel is aligned with ATT&CK v18.
Summary
The frameworks don’t compete, they build on each other. TOGAF sets the enterprise direction, CAF builds the cloud foundation, WAF designs the individual workloads. SAF organizes security across all of these levels, the MCRA describes the technical target picture, and MITRE ATT&CK checks it against what attackers actually do. Zero Trust and a feedback loop from operations back to architecture tie it all together into an enterprise-wide security strategy.
Sources
- The Open Group: The TOGAF Standard
- National Institute of Standards and Technology: The NIST Cybersecurity Framework (CSF) 2.0
- Center for Internet Security: CIS Critical Security Controls
- Microsoft Learn: What is the Cloud Adoption Framework?, CAF Secure methodology, What is an Azure landing zone?
- Microsoft Learn: What is the Azure Well-Architected Framework?, Security design principles, Security checklist
- Microsoft Learn: Microsoft Security Adoption Framework (SAF), Security disciplines, Security standards and frameworks
- Microsoft Learn: Microsoft Cybersecurity Reference Architecture (MCRA)
- MITRE: ATT&CK, ATT&CK updates
- Microsoft Learn: MITRE ATT&CK coverage in Microsoft Sentinel
- The Sherwood Applied Business Security Architecture: What is SABSA?
- AWS Docs: AWS Well-Architected Framework
- Google Cloud Docs: Google Cloud Well-Architected Framework