Architecture glossary

An activity diagram is a UML behavior diagram that shows the flow of actions in a process, including decisions, parallel work and where it starts and ends.

An architecture diagram is a visual overview of a software system that shows its main components, how they connect and where each one runs.

A block diagram is a high-level drawing of a system in which each major part is a labeled block and the lines between blocks show the signals that connect them.

BPMN, or Business Process Model and Notation, is a standard for drawing business processes with defined symbols for events, activities, gateways and flows.

The C4 model is a way to diagram software architecture at four levels of detail: system context, containers, components and code.

A class diagram is a UML structure diagram that shows a system's classes, their attributes and operations, and the relationships between them.

A component diagram is a UML diagram that shows the components of a software system, the interfaces each provides or requires and how they are wired together.

A context diagram shows one system as a single box, with the people and external systems around it and the flows between them, but nothing of its inside.

A data flow diagram (DFD) is a visual map of how data moves through a system, including where it comes from, how it changes, where it is stored and where it goes.

A dependency graph is a directed graph in which each node is a package, module, task or service and each edge shows that one of them depends on another.

A deployment diagram is a UML diagram that shows where software runs: hardware and environments as nodes, the artifacts on them and the links between them.

Diagram as code is the practice of describing a diagram in plain text so that a tool renders it, and the text can be versioned, reviewed and reused like code.

An entity relationship diagram (ERD) is a visual model of a database that shows its entities, their attributes and how they relate to each other.

A fishbone diagram, also called an Ishikawa or cause-and-effect diagram, lays out the possible causes of a problem by category to help find the root cause.

A flowchart is a diagram that shows the steps of a process or algorithm as shapes joined by arrows, with a diamond wherever the path depends on a decision.

A network diagram is a map of a computer network that shows its devices, such as routers, switches, firewalls and servers, and the links between them.

A process map is a diagram that shows the steps in a process, who does each one and the decisions along the way, from the first step to the last.

A sequence diagram is a UML interaction diagram that shows the messages passed between objects or services over time, in the order they happen.

A state diagram shows the states an object can be in, the events that move it between them and the actions taken on each transition, from start to finish.

A swimlane diagram is a flowchart divided into lanes, one per role, team or system, so every handoff shows up as an arrow crossing from one lane to the next.

A UML diagram is a visual model of a software system drawn in the Unified Modeling Language, a standard notation for describing structure and behavior.

A use case diagram is a UML behavior diagram that shows what a system does for its users, with actors, use cases and the system boundary between them.

A user flow is a diagram of the path a user takes through a product to complete a task, showing each screen, decision and action from start to finish.

A user journey map is a grid that shows what a user does, uses, feels and struggles with at each stage of reaching a goal, inside and outside your product.

A wireframe is a low-detail sketch of a screen that shows its layout, content and controls as plain boxes and lines, before any visual design is applied.

AI agent architecture is the structure of a system in which a language model plans, calls tools and uses memory to complete tasks with little human input.

An API gateway is a server between clients and backend services that handles authentication, rate limiting, routing and caching for every request in one place.

An architecture decision record (ADR) is a short document that captures one significant design decision, the context that led to it and its consequences.

Architecture patterns are reusable ways to structure a software system, such as layered, microservices or event-driven, each with known benefits and costs.

Client-server architecture splits an application into clients that send requests and a server that holds the data and logic and answers them over a network.

A design review is a structured check of a proposed software design by peers before it is built, to catch risks and record the decision while change is cheap.

Event-driven architecture is a design in which services publish events to a broker and other services consume them asynchronously instead of calling each other.

Microservices architecture builds an application as a set of small, independently deployable services that each own one business capability and its data.

Monolithic architecture builds an application as one deployable unit that holds the user interface, business logic and data access in a single codebase.

MVC architecture divides an application into a model that holds data and rules, a view that renders it and a controller that turns user input into changes.

RAG architecture, or retrieval-augmented generation, is a system design that fetches relevant documents and gives them to a language model to ground its answer.

Serverless architecture runs application code as managed functions and services that the cloud provider provisions, scales and bills per use.

A solution architect is the engineer who designs how applications, data and infrastructure combine to solve one business problem and guides its delivery.

Solution architecture is the design of how applications, integrations, data and infrastructure fit together to solve one specific business problem.

System architecture is the high-level structure of a whole system: its components, how they interact, where they run and the qualities it has to meet.

System design is the process of choosing a software system's components, data stores and interactions so it meets its functional and scale requirements.

Technical debt is the future cost of rework a team takes on when it chooses a quick solution now instead of a better one that would have taken longer.

Three-tier architecture splits an application into a presentation tier, an application tier and a data tier that run on separate hosts and talk over a network.

A CI/CD pipeline automatically builds, tests and deploys every code change through fixed stages from commit to production, with a check at each step.

Cloud computing architecture is the arrangement of front-end clients, the network and a provider's back-end compute, storage, database and management services.

Container orchestration automates placing, scaling, networking, healing and updating containers across a cluster of machines from a declared desired state.

Disaster recovery is the plan and infrastructure for restoring a service after a major outage, set by how fast it must recover and how much data it may lose.

High availability is the practice of keeping a service running through failures by removing single points of failure and failing over to a healthy copy.

A hub-and-spoke network connects workload networks (spokes) to one central hub that holds shared firewalls, VPN gateways and DNS, instead of a full mesh.

A hybrid cloud joins an on-premises data center to a public cloud with private connectivity and shared identity, so workloads can run on either side.

Infrastructure as code manages servers, networks and cloud services through versioned configuration files a tool applies, instead of manual console changes.

Kafka architecture is the structure of a Kafka cluster: brokers store topics as partitioned, replicated logs that producers write to and consumer groups read.

Kubernetes architecture is the structure of a cluster: a control plane that stores and enforces the desired state and worker nodes that run containers in pods.

A landing zone is the pre-built foundation of accounts, identity, networking, logging and guardrails into which cloud workloads are deployed.

A load balancer is a server or service that spreads incoming traffic across a pool of backend servers, checks their health and routes around any that fail.

A message queue is a buffer that holds messages from producers until consumers process them, so services can communicate without waiting on each other.

Multi-cloud is the use of two or more public cloud providers by one organization, with identity, networking and tooling shared across all of them.

Observability is the ability to understand what a system is doing from the logs, metrics and traces it emits, well enough to debug problems nobody predicted.

A reverse proxy is a server in front of backend servers that accepts client requests, terminates TLS, caches responses and routes each request by host or path.

A service mesh is an infrastructure layer of proxies and a control plane that secures, routes and observes traffic between services without changing their code.

Change data capture (CDC) is a technique that detects every insert, update and delete in a database and delivers each change to other systems as it happens.

A data fabric is an architecture that connects data across systems through a shared metadata layer, which automates how data is found, integrated and governed.

Data lake architecture is the design of a store that keeps raw data of any format in low-cost object storage, with a catalog and separate engines to query it.

Data lineage is a record of where a piece of data came from, every transformation it passed through and everywhere it is used, from source system to dashboard.

Data mesh architecture is a decentralized design in which domain teams publish their data as products on a shared platform, under common governance rules.

Data modeling is the process of defining how data is structured, related and stored, moving from business concepts through a logical model to a physical schema.

Data pipeline architecture is the design of how data moves from source systems to its users through ingestion, storage, transformation and serving.

A data warehouse is a central database that holds copies of data from many systems, organized for analysis and reporting rather than for running applications.

A database schema is the formal definition of how a database is organized: its tables, columns, data types, keys, constraints and indexes.

An ETL pipeline is a data workflow that extracts data from source systems, transforms it into a consistent shape and loads it into a warehouse for analysis.

Medallion architecture is a data design pattern that organizes a lakehouse into bronze, silver and gold layers, each holding a more refined version of the data.

A star schema is a data model with a central fact table of measurements joined to dimension tables that describe them, designed for fast analytical queries.

Cloud security architecture is the design of the controls that protect workloads in a cloud account, layered across identity, network, compute and data.

Single sign-on lets a user log in once with an identity provider and then reach many applications without signing in again, using a token or assertion.

Threat modeling is a structured way to find what can go wrong in a system before it is built, by mapping its parts, data flows and trust boundaries.

Zero trust architecture is a security model that verifies every request to every resource, using identity, device and context, instead of trusting the network.

AI governance is the policies, processes, controls and roles an organization uses to decide which AI systems it runs and to keep them safe and accountable.

Application dependency mapping records how applications connect to each other and to their infrastructure, so teams can see what a change or outage affects.

Application portfolio management inventories and scores an organization's applications to decide which to invest in, tolerate, migrate or eliminate.

Application rationalization is the process of assessing every application an organization runs and deciding which to keep, consolidate, replace or retire.

An architecture review board is a standing group of architects and specialists that checks significant designs against an organization's standards before build.

A business capability map is a one-page model of what an organization does, split into stable capabilities, used to plan where to invest in systems and change.

A configuration management database (CMDB) stores information about an organization's IT components, called configuration items, and the relationships between them.

Current vs future state architecture compares the systems an organization runs today with the target it is moving toward, so the gap can be planned.

An enterprise architect owns an organization's target architecture across business, data, applications and technology and the roadmap for reaching it.

Enterprise architecture is the practice of describing an organization's business, data, applications and technology as one system and planning how it changes.

IT asset management (ITAM) tracks the hardware, software and cloud resources an organization owns across their lifecycle, with cost, contracts and compliance.

An IT roadmap is a time-based plan of the initiatives, milestones and dependencies that take an organization's technology from today's state to its target.

Service mapping builds a top-down model of a business service that shows every application, middleware and infrastructure component the service depends on.

The ServiceNow CMDB is the configuration database inside the ServiceNow platform, and CSDM is its standard for which tables to use and how to relate them.

TOGAF is The Open Group's enterprise architecture framework, built around the architecture development method, a repeatable cycle for planning change.