Glossary
Glossary
Short definitions of the terms used across the services and notes.
- Agentic system
- Software in which one or more AI agents plan, use tools and collaborate to reach a goal, within defined limits and oversight. Related service.
- AI agent
- A program that uses a language model to decide which steps to take and uses tools and data to carry them out, within defined limits. Related service.
- AI engineering
- Applying software engineering practices to software built on language models: architecture, tests, evaluations, observability, security and controlled costs. Related service.
- Ansible
- An automation tool that configures servers and networks from readable files describing the desired state, with no agent on the nodes. Related service.
- API
- An interface through which one program uses another, with a contract stating which requests are allowed and which responses come back. Related service.
- Argo CD
- A GitOps tool for Kubernetes: it compares the cluster state with the one in Git and realigns it, with a web interface.
- Assessment
- An independent review of architecture, infrastructure, security and cost, ending with priorities and next steps. Related service.
- Cloud native
- A way of building and running applications with containers, small services, automation and infrastructure described as code, so they can change often and safely. Related service.
- Container
- A package containing an application and its dependencies that runs isolated from the rest of the system, identically everywhere.
- Data protection by design
- A GDPR principle (Art. 25): measures to protect data are decided in the design and, by default, only the minimum necessary is processed.
- DevSecOps
- A practice that puts security checks inside the development and delivery cycle instead of adding them at the end.
- Evaluations (evals)
- Automated tests that measure the quality of answers and behaviours of an AI system on known cases, repeated at every change. Related service.
- FinOps
- A discipline for making cloud costs visible and controllable, involving both engineers and decision makers.
- Gateway API
- A set of Kubernetes APIs describing how traffic enters the cluster; it succeeds Ingress and separates the roles of those who run the network and those who publish applications.
- GitOps
- A method where the desired state of infrastructure and applications lives in Git and a controller applies and maintains it. Related service.
- Helm
- A package manager for Kubernetes: it bundles an application's resources into a configurable package (chart).
- Infrastructure as code
- Describing servers, networks and services in versioned files so infrastructure can be created, reviewed and repeated like software.
- Jenkins
- An open source automation server for building, testing and releasing software with pipelines described as code. Related service.
- Kubernetes
- An open source platform that starts, scales and heals containerised applications across a group of machines. Related service.
- LLM
- Large language model: an AI system trained on large amounts of text that generates and understands natural language.
- MCP
- Model Context Protocol: an open protocol through which an AI agent discovers and uses external tools and data in a standard way. Related service.
- MQTT
- A lightweight publish-subscribe messaging protocol widely used for sensors and IoT devices. Related service.
- Observability
- The ability to understand what is happening in a system from its signals: metrics, logs and traces.
- PCI DSS
- A security standard for organisations handling payment card data: networks, access, encryption, monitoring and periodic checks.
- RAG
- Retrieval-augmented generation: the model first retrieves relevant passages from your documents and uses them to answer, citing sources. Related service.
- RBAC
- Role-based access control: each person can do only what their role allows.
- Red Hat OpenShift
- Red Hat's enterprise Kubernetes platform, with built-in security, update and developer tooling. Related service.
- Refactoring
- Restructuring existing code to make it easier to understand and change, without altering its behaviour. Related service.
- Replatforming
- Moving an application to a new platform with few changes, for example to containers or managed services. Related service.
- SBOM
- A list of the components and dependencies of a piece of software, useful to know what it contains and to react to vulnerabilities. Related service.
- Scalability
- The ability of a system to handle more load by adding resources, without redesigning it.
- SLO
- Service level objective: a measurable target, for example on availability or response time, that guides reliability decisions.
- Software engineering
- A discipline that brings method to development: requirements, architecture, tests, review, delivery and maintenance, for software that lasts and can be changed. Related service.
- Terraform and OpenTofu
- Tools for describing and creating cloud infrastructure as code; OpenTofu is the open source project forked from the same base.
- Zero trust
- A security model in which no user or system is trusted by default: every request is verified and gets only the permissions it needs.