A working reference for EA architects at every level: from analysts building their first capability map to architects governing enterprise-wide transformation programs. Definitions are written for clarity, not formality. Where a term connects to another, we say so.
Enterprise Architecture The discipline of aligning an organization’s business strategy with its operating model, technology, data, and processes in a coherent, governed way. EA makes the invisible structure of an enterprise visible: and manageable. It is both a practice and a body of artefacts. See also: Architecture Repository, Architecture Framework.
Architecture Repository The central store where all EA artefacts live: models, diagrams, decisions, standards, and reference data. In Sparx EA, the repository is a database (SQL Server, MySQL, Oracle, or PostgreSQL) accessed through the desktop client and Pro Cloud Server. A well-governed repository is the difference between EA as a one-off exercise and EA as an ongoing capability. See also: Pro Cloud Server (PCS), Sparx EA.
Architecture Framework A structured set of principles, methods, and models for producing, organizing, and governing architecture artefacts. Frameworks provide a common language and prevent every EA team from reinventing the wheel. TOGAF is the most widely adopted; others include DoDAF, FEAF, and BIAN. See also: TOGAF, Metamodel.
Metamodel The model of your modeling language: the rules that define which elements exist, what properties they carry, and how they relate to each other. In Sparx EA, the built-in metamodel can be extended or replaced using MDG Technology. Getting the metamodel right is foundational to AI-readable models. See also: MDG Technology, MDG Technology.
MDG Technology Model Driven Generation Technology: Sparx EA’s mechanism for defining and distributing custom metamodels, diagram toolboxes, and stereotypes across a team or enterprise. MDG profiles are the primary governance instrument for ensuring consistent modeling. When AI tools query your EA repository, well-formed MDG profiles are what make the data intelligible. See also: Metamodel, EA GraphLink.
Architecture Pattern A reusable, named solution to a recurring architecture problem. Patterns are not copy-paste blueprints; they are structured guidance that must be contextualized to each situation. Documenting patterns in the EA repository: with Sparx EA: turns institutional knowledge into a governed, searchable asset. See also: Reference Architecture.
Capability Map A structured view of what an organization does: expressed as discrete capabilities rather than organisational units or processes. Capabilities are stable over time even as structures and processes change, making them the ideal anchor for investment decisions and gap analysis. A capability map is usually the right place to start with EA. See also: Business Architecture, Value Stream.
Value Stream An end-to-end sequence of activities that delivers value to a customer or stakeholder. Value streams cross organisational boundaries, which is precisely what makes them useful for spotting inefficiency and misalignment. In ArchiMate, value streams are modelled in the motivation and business layers. See also: Capability Map, Business Architecture.
Business Architecture The layer of EA concerned with strategy, business models, capabilities, value streams, organisational structures, and processes. Business architecture provides the “why” and “what” that technology architecture must serve. Without it, technology investment decisions lack grounding. See also: Application Architecture, Capability Map.
Application Architecture The layer of EA that describes the software applications an organization uses, how they are structured internally, and how they interact with each other and with the business. Application architecture is where most legacy complexity becomes visible. See also: Business Architecture, Technology Architecture.
Technology Architecture The layer of EA concerned with the infrastructure, platforms, cloud services, and technology standards that host and connect applications. Technology architecture decisions have long lead times and high switching costs, making governance here especially important. See also: Application Architecture, Data Architecture.
Data Architecture The layer of EA that defines how data is produced, stored, governed, and consumed across the enterprise. Data architecture has grown in importance with the rise of AI: models are only as good as the data they reason over. See also: Technology Architecture, Ontology.
Architecture Principles Agreed statements that guide architecture decisions consistently across the enterprise. Principles are not rules: they are enduring preferences with rationale. A good principle names the trade-off it is making and why. Example: “We prefer cloud-native services over self-hosted equivalents, except where data sovereignty requires otherwise.” See also: Architecture Decision Record (ADR).
Architecture Decision Record (ADR) A short document that captures a significant architecture decision: its context, the options considered, the decision made, and the consequences accepted. ADRs are the institutional memory of your architecture program. Stored in Sparx EA, they become queryable artefacts rather than buried documents. See also: Architecture Principles, Architecture Repository.
Architecture Runway The extent to which the current architecture can support near-term feature and capability delivery without significant rework. A short runway forces reactive, expensive changes; a long runway enables agile delivery. Managing runway is a core responsibility of an EA practice embedded in product or program delivery. See also: Technical Debt, Architecture Roadmap.
ArchiMate An open, vendor-neutral modeling language for enterprise architecture, maintained by The Open Group. ArchiMate provides a layered structure (motivation, strategy, business, application, technology, physical, implementation) with well-defined element types and relationship semantics. It is the preferred language for cross-layer EA modeling. Sparx EA ships with full ArchiMate support. See also: TOGAF, UML.
BPMN Business Process Model and Notation: the ISO standard for modeling business processes. BPMN diagrams are readable by both business and technical stakeholders, making them a valuable communication tool. BPMN models well in Sparx EA and is often combined with ArchiMate: ArchiMate for structure, BPMN for process detail. See also: ArchiMate, UML.
UML Unified Modeling Language: the standard for object-oriented and software design modeling. UML covers structural diagrams (class, component, deployment) and behavioural diagrams (sequence, activity, state machine). Sparx EA was built on UML and remains one of the strongest UML tools available. See also: BPMN, SysML.
SysML Systems Modeling Language: a UML profile adapted for systems engineering. SysML adds requirements diagrams and parametric diagrams to UML’s repertoire, making it suitable for cyber-physical systems and defense or aerospace programs. Sparx EA has strong SysML support. See also: UML, DoDAF.
DMN Decision Model and Notation: the OMG standard for modeling business decisions and decision logic. DMN separates decision structure (decision requirements diagrams) from decision logic (decision tables). It is particularly useful alongside BPMN for modeling automated decision processes. See also: BPMN.
Notation The visual vocabulary of a modeling language: the shapes, lines, colors, and labels that represent model elements on a diagram. Notation is the surface; the metamodel is the meaning beneath it. Consistent notation is a governance concern, enforced in Sparx EA through MDG Technology. See also: Metamodel, MDG Technology.
Viewpoint A specification of the conventions for constructing a view: which concerns it addresses, which element types it includes, and which stakeholders it is intended for. The TOGAF Architecture Content Framework and ArchiMate 3 both define standard viewpoints. See also: Stakeholder View, Concern.
Concern A specific interest that a stakeholder has in the architecture: performance, security, cost, compliance, scalability. Architecture views are constructed to address specific concerns. Tracing concerns to viewpoints to views is the formal discipline of architecture description. See also: Viewpoint, Stakeholder View.
Stakeholder View A representation of the architecture that addresses the concerns of one or more stakeholders. A view is always produced from a viewpoint. Effective EA produces views that stakeholders actually use: which requires understanding what each stakeholder cares about. See also: Viewpoint, Concern.
Diagram vs Model A critical distinction: a diagram is a graphical representation; a model is the structured, semantically rich underlying data. In Sparx EA, diagrams are windows into the model: the same element can appear on many diagrams without being duplicated in the model. AI tools query the model, not the diagram. This distinction matters enormously for EA GraphLink integration. See also: EA GraphLink, Architecture Repository.
TOGAF The Open Group Architecture Framework: the world’s most widely adopted enterprise architecture framework. TOGAF provides the Architecture Development Method (ADM), a cyclic process for developing and governing architecture, along with content frameworks, capability frameworks, and reference models. TOGAF is a starting point, not a prescription. See also: ArchiMate, Architecture Framework.
ArchiMate See Modeling Languages section. As a framework entry: ArchiMate is maintained by The Open Group and is designed to work alongside TOGAF: TOGAF provides the method, ArchiMate provides the language. The two are complementary, not competing. See also: TOGAF.
DoDAF The US Department of Defense Architecture Framework: a framework designed for defense acquisition and systems engineering. DoDAF organizes views into viewpoints (All, Capability, Data and Information, Operational, Project, Services, Standards, Systems). Widely used in US defense and by allied defense organizations. See also: MODAF, NAF, SysML.
MODAF The UK Ministry of Defense Architecture Framework: a framework closely aligned with DoDAF but adapted for UK defense procurement and operations. MODAF has largely been superseded by NAF in UK and NATO contexts but remains in use in legacy programs. See also: DoDAF, NAF.
NAF The NATO Architecture Framework: developed to support NATO interoperability and coalition operations. NAF v4 was designed to align with TOGAF and ArchiMate, making it more accessible to commercial EA architects entering defense programs. See also: DoDAF, MODAF, ArchiMate.
FEAF The US Federal Enterprise Architecture Framework: a framework for managing EA across US federal government agencies. FEAF organizes architecture across performance, business, data, application, infrastructure, and security sub-architectures. See also: TOGAF, Architecture Framework.
BIAN The Banking Industry Architecture Network framework: an industry reference model for banking business architecture. BIAN provides a service landscape of banking capabilities and service domains, making it valuable for financial services EA programs as a starting point for capability mapping. See also: Capability Map, TM Forum ODA.
TM Forum ODA The TM Forum Open Digital Architecture: a reference architecture framework for telecommunications companies. ODA provides a component-based architecture model with well-defined business capability components and API standards. Analogous to BIAN but for telco. See also: BIAN, Architecture Framework.
Sparx EA Enterprise Architect by Sparx Systems: a desktop-based modeling tool and one of the most powerful and cost-effective EA platforms available. Sparx EA supports UML, ArchiMate, BPMN, SysML, TOGAF, and custom metamodels via MDG Technology. Its repository model (shared database) enables team collaboration at enterprise scale. See also: EA Repository, Pro Cloud Server (PCS).
EA Repository The shared database that stores all model content in a Sparx EA deployment. The repository is the single source of truth for all architecture artefacts: elements, relationships, diagrams, documents, baselines, and tagged values. Repository integrity is the foundation of everything else. See also: Pro Cloud Server (PCS), Sparx EA, Baseline.
Pro Cloud Server (PCS) Sparx Systems’ server-side component that provides browser-based access to Sparx EA repositories, WebEA (read-only web interface), integration services, and the connectivity layer used by EA GraphLink. PCS is the bridge between the desktop client and the wider enterprise ecosystem. See also: Sparx EA, EA GraphLink.
EA GraphLink Sparx Services’ connectivity product that exposes the EA repository as a live data source for AI and BI tools. EA GraphLink provides two interfaces: Interface A (GraphQL API for Power BI, Tableau, and other analytics tools) and Interface B (MCP Server for AI tools including Microsoft Copilot, Claude, and Salesforce Agentforce). See also: MCP Server, GraphQL, Pro Cloud Server (PCS).
Kernaro AI Hub A web-based portal that provides business stakeholders with guided, AI-assisted access to EA content: without requiring Sparx EA licenses. Kernaro AI Hub uses EA GraphLink’s MCP interface to retrieve live model data and presents it through a conversational interface. It is the stakeholder-facing layer of an AI-enabled EA practice. See also: EA GraphLink, Kernaro Assist.
Kernaro Assist An AI assistant embedded within the Sparx EA client, providing modellers with context-aware suggestions, pattern recommendations, and governance checks as they work. Kernaro Assist uses the local model context to provide relevant guidance rather than generic responses. See also: Kernaro AI Hub, EA GraphLink.
MDG Technology See EA Core Concepts section. As a platform entry: in Sparx EA, MDG Technologies are the primary mechanism for distributing governance: custom element types, stereotypes, tagged values, toolboxes, and validation rules: across the modeling team. Organizations without well-maintained MDG profiles find their models drift into inconsistency over time. See also: Metamodel.
MCP Server Model Context Protocol Server: a server that exposes structured data to AI agents and large language models via the Model Context Protocol standard. EA GraphLink’s Interface B is an MCP Server, allowing AI tools to query the EA repository as a knowledge source during reasoning tasks. See also: Model Context Protocol (MCP), EA GraphLink.
GraphQL A query language and runtime for APIs that allows clients to request exactly the data they need: no more, no less. EA GraphLink’s Interface A exposes the EA repository as a GraphQL endpoint, enabling precise, efficient queries from Power BI, Tableau, and custom applications. See also: EA GraphLink, Semantic Layer.
Package The primary organisational unit within a Sparx EA repository. Packages are hierarchical containers for model elements and diagrams, analogous to folders. Package structure reflects governance decisions: who owns what, what belongs where, and how artefacts are versioned. See also: Baseline, EA Repository.
Baseline A snapshot of a Sparx EA package at a point in time, used for change tracking and version comparison. Baselines enable “before and after” analysis of model changes and support architecture governance by providing an audit trail. See also: Package, Architecture Decision Record (ADR).
Model Search Sparx EA’s built-in capability for querying the repository using SQL-based searches, pre-built search templates, or custom scripts. Model Search is how architects interrogate the repository programmatically: finding all elements of a type, all relationships of a kind, or all elements with specific tagged values. EA GraphLink extends this capability to external tools. See also: GraphQL, EA GraphLink.
Model Context Protocol (MCP) An open protocol standard, developed by Anthropic, that defines how AI agents and large language models can connect to external data sources and tools. MCP enables AI tools to query live data: including EA repositories via EA GraphLink: rather than relying solely on training data. MCP is becoming the standard integration layer for enterprise AI. See also: MCP Server, EA GraphLink.
GraphQL See Sparx EA Platform section. As an AI & Integration entry: GraphQL’s precise query capability makes it well-suited to BI and analytics use cases where you want to retrieve specific model slices: all applications in a domain, all capabilities mapped to a strategic objective: without pulling the entire repository. See also: EA GraphLink, Semantic Layer.
Large Language Model (LLM) A machine learning model trained on large text corpora that can generate, summarize, classify, and reason over natural language. LLMs are the engine behind tools like Microsoft Copilot, Claude, and ChatGPT. LLMs can reason over EA model content when connected via MCP or similar protocols, but their output quality is directly tied to the quality and structure of the model content. See also: Model Context Protocol (MCP), AI Agent.
AI Agent A software system that uses an LLM to plan and execute multi-step tasks, often calling external tools or data sources as part of its reasoning process. EA-connected AI agents can answer questions like “which capabilities have no supporting applications?” or “what is the impact of decommissioning this system?” by querying the live EA model. See also: Large Language Model (LLM), EA GraphLink.
Retrieval-Augmented Generation (RAG) A technique for improving LLM accuracy by retrieving relevant documents or data at query time and including them in the model’s context window. EA GraphLink’s MCP interface enables a form of RAG for EA content: the AI retrieves relevant model elements before generating a response. Well-structured, MDG-governed models dramatically improve RAG quality. See also: Large Language Model (LLM), MDG Technology.
Semantic Layer An abstraction layer that translates raw data into business-meaningful concepts, metrics, and relationships. In EA, the semantic layer is the metamodel: the definitions that give meaning to model elements. A strong semantic layer makes EA data interpretable by both BI tools and AI tools without additional transformation. See also: Metamodel, GraphQL.
Architecture Intelligence The application of AI capabilities to EA practice: using LLMs and AI agents to query, analyze, synthesize, and extend architecture models. Architecture intelligence is not about replacing architects; it is about giving architects faster access to their own repository content and giving stakeholders self-service access to architecture knowledge. See also: EA GraphLink, AI Agent.
Digital Thread A connected data trail that links artefacts across the full lifecycle of a system or capability: from requirements through design, implementation, operation, and decommission. The EA repository, when properly integrated, can serve as the backbone of a digital thread. EA GraphLink surfaces this thread to AI and analytics tools. See also: Architecture Repository, EA GraphLink.
Ontology A formal representation of knowledge as a set of concepts, definitions, and relationships within a domain. In EA, the metamodel is a form of ontology. Knowledge graph technologies build on ontologies to enable richer reasoning over connected data. MDG Technology is the practical mechanism for defining and governing EA ontologies in Sparx EA. See also: Knowledge Graph, Metamodel, MDG Technology.
Knowledge Graph A structured representation of entities and the relationships between them, designed to support complex querying and reasoning. EA repositories are natural knowledge graphs: elements are entities, relationships are edges, and diagrams are views. EA GraphLink exposes this graph structure to AI tools via MCP. See also: Ontology, EA GraphLink.
Architecture Review Board (ARB) The governance body responsible for reviewing, approving, and maintaining architecture decisions within an organization. An effective ARB is lightweight enough to move at program pace but rigorous enough to prevent architectural drift. It should be chaired by a senior architect with executive sponsorship. See also: Architecture Governance, EA Center of Excellence (CoE).
Architecture Governance The framework of processes, roles, and standards by which architecture decisions are made, communicated, and enforced. Governance is what prevents EA from becoming a library of ignored diagrams. It includes the ARB, architecture principles, decision records, and compliance checkpoints in delivery processes. See also: Architecture Review Board (ARB), Architecture Principles.
EA Center of Excellence (CoE) A dedicated team or function responsible for setting EA standards, developing capability, managing the architecture repository, and supporting delivery teams. A CoE is the organisational home of the EA practice. Its size ranges from two people in a mid-sized organization to dozens in a large enterprise. See also: Architecture Governance, EA Program.
Architecture Maturity A measure of how systematically and effectively an organization practises enterprise architecture. Maturity models (such as TOGAF’s Architecture Capability Framework or Gartner’s EA Maturity Model) assess dimensions including governance, tooling, stakeholder engagement, and business alignment. Maturity assessment is a useful starting point for an EA program. See also: EA Program, EA Center of Excellence (CoE).
Technical Debt The accumulated cost of shortcuts, legacy decisions, and deferred refactoring in an organization’s technology estate. Technical debt is not inherently bad: sometimes it is a deliberate trade-off: but it must be made visible and managed. EA is the discipline that makes technical debt visible at enterprise scale. See also: Architecture Runway, Decommission.
EA Program The organisational program that establishes, runs, and matures the EA practice. An EA program is distinct from individual architecture projects: it is the sustained investment in the capability of doing architecture well. Without program-level commitment, EA initiatives tend to stall. See also: EA Center of Excellence (CoE), Architecture Maturity.
Stakeholder Register A structured list of all stakeholders relevant to an architecture program: their roles, interests, concerns, and influence. Managing a stakeholder register is basic program management, but EA stakeholder registers should also capture which architecture views and artefacts each stakeholder needs. See also: Stakeholder View, RACI.
RACI A responsibility assignment matrix: Responsible, Accountable, Consulted, Informed. RACI charts clarify who does what in governance processes: who approves a decision, who must be consulted, who is kept informed. In EA governance, RACI applies to architecture reviews, model updates, standard changes, and program decisions. See also: Architecture Governance, Stakeholder Register.
Architecture Roadmap A time-phased plan that shows how the architecture will evolve from its current state to a target state. An architecture roadmap is not a project plan: it is a sequenced, prioritized set of architecture changes, with dependencies and enabling conditions made explicit. It is one of the most important outputs of an EA program. See also: Architecture Runway, Technical Debt.
Decommission The planned retirement of a system, application, or technology component from the estate. Decommissioning is often neglected: organizations add new capabilities but rarely remove old ones. EA provides the dependency analysis and impact assessment that makes safe decommissioning possible. Tracked in Sparx EA as a lifecycle state on application elements. See also: Technical Debt, Architecture Roadmap.
This glossary is maintained by Sparx Services. Terms reflect current practice as of 2025. For questions or suggested additions, contact the Sparx Services team.
Talk to a Sparx Services architect about where your organization is on the journey and what the next stage looks like.