Glossary

Agentic orchestration

What is agentic orchestration?

Agentic orchestration puts AI agents inside a modelled process, so the process decides where an agent may act and records what it did.

Camunda uses the term for AI agents that work as steps of a BPMN process. In Camunda 8 the agent runs through the AI Agent connector. Around it, the model keeps the deterministic steps: service tasks, decisions in DMN and user tasks for people. Inside an ad-hoc sub-process the agent may choose among the tools the process offers, and nothing else. Each step and its variables are recorded in the process history. An audit can then follow what the agent did and which rule allowed it.

API Gateway

What is an API gateway?

An API gateway is a single entry point in front of your APIs: it checks who calls, applies policies such as rate limits and routes each request.

A gateway sits between the applications that call an API and the services that answer it. It enforces security, such as API keys, OAuth tokens and mutual TLS, applies throttling and quotas, and can transform or mediate a request on the way. In IBM webMethods, API Gateway takes this role beside Integration Server, which provides many of the services behind it. The gateway also records traffic, so you can see which consumer uses which API. A developer portal often sits beside it, where API consumers find the documentation and request access.

Business Process Model and Notation (BPMN)

What is BPMN?

BPMN is a standard notation for business processes: one diagram that people can read and a process engine such as Camunda can execute.

Business Process Model and Notation is maintained by the Object Management Group (OMG). A model shows events as circles, tasks as rounded rectangles, gateways as diamonds and the sequence flows between them. Behind the picture sits an XML file, and that file is what an engine runs. A service task calls a system, a user task waits for a person, and a business rule task evaluates a DMN decision. A token moves along the flows and marks where each running instance stands. Because the engine executes the model itself, the diagram and the running process do not drift apart.

Configuration Management Database (CMDB)

What is a CMDB?

A CMDB records the configuration items of your IT estate, such as servers, applications and services, and the relationships between them.

Each record in a CMDB is a configuration item (CI): a server, a database, an application, an integration or the business service built on them. The relationships matter as much as the items. They show which services are affected when a server fails, and which systems a change will touch. In ServiceNow the CMDB sits at the centre of ITSM: incidents, problems and changes refer to its CIs. A CMDB is only useful while it is current, so discovery tools and integrations feed it, and owners keep each class of item correct.

Connector

What is a connector in Camunda?

Connector: In Camunda 8, a connector is a reusable building block that links a process to an outside system, such as a REST API, without a custom job worker.

Outbound connectors call a system from a task in the process. Inbound connectors start a process or let it continue when an event arrives, such as a webhook or a message. A modeller configures a connector through an element template: the fields for the endpoint, the authentication and the mapping of the result. Camunda ships connectors for common protocols and services, and a team can write its own with the Connector SDK. The Camunda AI Agent connector is one of them: it lets an AI agent act as a step of a BPMN process.

Data Migrator

What is the Camunda Data Migrator?

The Data Migrator is Camunda’s tool for moving process data, above all running process instances, from a Camunda 7 database into Camunda 8.

A Camunda 7 to 8 migration has three parts: the models, the code and the data. The models are converted and deployed to Camunda 8 first. The Data Migrator then reads the running instances from the Camunda 7 database and starts each one in Camunda 8 at the step it had reached, with its variables. History is a separate question: what must be kept, for how long and where. Instances that will finish soon can also run out on Camunda 7, while new ones start on Camunda 8. The choice depends on how long your instances live.

Data Protection Impact Assessment (DPIA)

What is a DPIA?

A DPIA is the assessment GDPR Article 35 requires before processing personal data that is likely to result in a high risk to people’s rights and freedoms.

A DPIA describes the processing and its purpose, tests whether it is necessary and proportionate, assesses the risks to the people concerned and sets out the measures that address them. GDPR names cases where it is required, such as systematic and extensive evaluation of people, including profiling, that leads to decisions with legal or similarly significant effects. AI systems often fall into these cases, and automated decisions also meet the limits of Article 22. If a high risk remains after the measures, the controller consults the supervisory authority before the processing starts.

Related: EU AI Act, ISO/IEC 42001

Decision Model and Notation (DMN)

What is DMN?

DMN is a standard for modelling business decisions, most often as decision tables, so that people can read the rules and an engine can evaluate them.

Decision Model and Notation comes from the Object Management Group, like BPMN. A decision table lists input columns, output columns and one rule per row, with a hit policy that says which rows apply. Conditions are written in FEEL. A decision requirements diagram shows how decisions depend on each other and on their inputs. In Camunda, a business rule task in a BPMN process calls a DMN decision and uses its result. Keeping the rules in DMN lets a team change a threshold without touching the code, while the process model stays as it is.

End of support

What does end of support mean?

End of support is the date from which a vendor stops standard support for a product version: no more fixes or security patches under its standard policy.

End of support is a date in a vendor’s policy, not a legal deadline. The software keeps running after it. Without extended support, new defects and vulnerabilities stay unfixed, and the risk is yours. Many vendors sell extended support for a period after the date. For IBM webMethods 10.15, standard support ends on 31 October 2027, initial extended support runs to 31 October 2028, and help with usage and known defects only to 31 October 2030. Some vendors call the date end of life: Camunda 7 Community Edition reached it in October 2025.

Enterprise Service Bus (ESB)

What is an enterprise service bus?

An enterprise service bus is a central integration layer that routes, transforms and mediates messages between applications, so they need not connect to each other directly.

The ESB grew with service-oriented architecture (SOA). Instead of many point-to-point links, each application connects once to the bus. The bus translates formats, maps data, routes each message by its content and handles retries and errors. IBM webMethods Integration Server, MuleSoft and TIBCO have all been used in this role. The pattern brings control and reuse, and it brings a risk: a single team and a single platform through which every change must pass. Many estates combine an ESB core with APIs, events and an iPaaS.

EU AI Act

What is the EU AI Act?

The EU AI Act, Regulation (EU) 2024/1689, sets rules for AI systems placed on the market or used in the EU, by the risk each system carries.

The AI Act entered into force on 1 August 2024. It prohibits some practices, sets duties for high-risk systems, such as risk management, data governance, technical documentation, logging and human oversight, and requires transparency for some others. The prohibitions and the AI literacy duty apply from 2 February 2025, and the general date of application is 2 August 2026. The AI Omnibus, Regulation (EU) 2026/1744, in force since 27 July 2026, moved the high-risk dates: the rules for Annex III systems apply from 2 December 2027, and those for AI in products under Annex I from 2 August 2028.

Friendly Enough Expression Language (FEEL)

What is FEEL?

FEEL is the expression language of the DMN standard, and Camunda 8 uses it for every expression in BPMN and DMN models.

Friendly Enough Expression Language was defined with DMN by the Object Management Group, to be readable by analysts and exact enough for an engine. It has types for numbers, strings, dates, durations, lists and contexts, and built-in functions for text, lists and dates. In Camunda 8 an expression starts with an equals sign, for example “= amount > 1000” on a sequence flow. Conditions, input and output mappings, timer values and decision rules all use it. A Camunda 7 to 8 migration rewrites the JUEL expressions of Camunda 7 in FEEL.

Integration Platform as a Service (iPaaS)

What is iPaaS?

An iPaaS is an integration platform that its vendor runs as a cloud service, where you build and operate integrations without running the platform yourself.

With an iPaaS the vendor installs, patches and scales the platform, and your team designs the integrations: the connectors, the mappings, the APIs and the flows between applications. It suits connections between cloud applications well, and many platforms add a runtime you install inside your own network to reach the systems there. IBM Integration SaaS is IBM’s offering in this space, and MuleSoft Anypoint Platform can also be used this way. The questions to settle early are where the data passes, who holds the keys and how the platform’s release cycle fits your own.

Integration Server

What is IBM webMethods Integration Server?

Integration Server is the runtime at the centre of IBM webMethods: it runs the integration services, adapters and packages that connect your applications.

IBM webMethods, a Software AG product until IBM acquired it on 1 July 2024, is built around Integration Server. Its services are written in the webMethods flow language or in Java and grouped in packages, which are deployed and versioned as units. Adapters connect it to databases, ERP systems and files, and Trading Networks runs on it as a package. It runs on your own servers or in containers on Kubernetes and Red Hat OpenShift. Standard support for IBM webMethods 10.15 ends on 31 October 2027.

ISO/IEC 42001

What is ISO/IEC 42001?

ISO/IEC 42001 is the international standard for an AI management system: how an organisation governs the AI it develops, provides or uses.

ISO/IEC 42001:2023, published in December 2023, is a management-system standard. It sets requirements for AI policies and roles, risk and impact assessments, controls across the AI life cycle, internal audits and continual improvement. It governs the organisation around the AI, not a single model. An organisation prepares its management system against the standard and then has it audited; the certificate comes from an accredited body. The standard is voluntary and does not by itself show conformity with the EU AI Act, but its records can support that work.

IT Service Management (ITSM)

What is ITSM?

IT Service Management is the way an organisation designs, delivers and supports its IT services, through processes such as incident, problem, change and request management.

Incident management restores a service quickly; problem management finds the cause behind repeated incidents; change management plans and approves changes to systems; request management handles what users ask for, from access to hardware. Frameworks such as ITIL describe these practices. ServiceNow ITSM implements them on the ServiceNow AI Platform, with the CMDB at its centre. ITSM works best when it is connected: an alert from monitoring opens an incident, a change ticket triggers a deployment, and the integration layer carries the data between ServiceNow and the systems it manages.

Jakarta EE

What is Jakarta EE?

Jakarta EE is the set of open specifications for enterprise Java, such as Servlet, Persistence and Validation, developed at the Eclipse Foundation.

Jakarta EE continues Java EE under the Eclipse Foundation. With the move, the package names changed from javax.* to jakarta.*, so code that imports javax.servlet or javax.persistence has to change its imports and its dependencies. For Spring applications that change came with Spring Boot 3. Spring Boot 4 moves on to Jakarta EE 11, with Jakarta Servlet 6.1, together with Spring Framework 7 and Jackson 3. A service still on Spring Boot 2 therefore makes both moves on its way to Spring Boot 4. Tools such as OpenRewrite can automate the import changes.

Related: Spring Boot, OpenRewrite

Java Unified Expression Language (JUEL)

What is JUEL?

JUEL is an open-source implementation of the Unified Expression Language for Java, which Camunda 7 uses for the expressions in its BPMN models.

In Camunda 7, JUEL expressions are written between a dollar sign and braces, and they appear in many places: conditions on sequence flows, the inputs and outputs of tasks, listeners, and service tasks bound to Java beans. Those bean calls run Java code inside the engine. Camunda 8 does not evaluate JUEL: its engine reads FEEL, and Java logic runs in job workers outside the engine. A Camunda 7 to 8 migration therefore lists every JUEL expression and sorts them. Simple conditions translate into FEEL; bean calls become job workers or connectors.

JavaDelegate

What is a JavaDelegate?

A JavaDelegate is a Java class that implements a Camunda 7 interface of the same name and runs the logic of a service task inside the process engine.

In Camunda 7 the engine is often embedded in a Java application, and a service task calls a JavaDelegate through its execute method. The delegate reads and writes process variables, and its work runs in the engine’s own transaction: if it throws an error, the transaction rolls back to the last wait state. Camunda 8 has no such call. Its engine creates a job, and a job worker outside the engine fetches the job, does the work and completes it. Moving a JavaDelegate to Camunda 8 means rewriting it as a job worker and planning for retries and idempotent work.

Job worker

What is a job worker in Camunda 8?

A job worker is a program outside the Camunda 8 engine that takes the jobs a process creates for its service tasks, does the work and reports the result.

When a process instance reaches a service task, the engine creates a job of a given type, such as check-credit. A worker subscribed to that type activates the job, reads its variables, calls the systems it needs and completes the job with new variables, or fails it so the engine can retry. Workers run as separate applications and scale on their own, in any language that can call the Camunda API. Java teams can build them with the Camunda Java client, also inside a Spring Boot application. The worker and the engine share no transaction.

Related: JavaDelegate, Zeebe, Connector

Long-Term Support (LTS)

What does LTS mean for Java?

A Long-Term Support release is a version that its vendors support with fixes for longer than the releases around it, so production systems can stay on it.

Java ships a feature release every six months, and some of them are marked LTS. Java 21 and Java 25 are LTS releases; JDK 27, a later release, is not. Many teams in production move from one LTS release to the next and skip the ones between. How long support lasts depends on the vendor of the JDK build you use, so the date to plan against is in that vendor’s policy. Spring Boot 4 needs at least Java 17, and an upgrade plan usually targets Java 21 or Java 25.

Related: Spring Boot, End of support

OpenRewrite

What is OpenRewrite?

OpenRewrite is an open-source refactoring tool: its recipes change source code automatically, for example to upgrade Spring Boot or to move from javax to jakarta.

OpenRewrite reads the code into a lossless semantic tree that keeps the formatting, applies a recipe to it and writes back only what changes. Recipes cover dependency versions, renamed classes and properties, changed APIs and test frameworks, and they can be combined. It runs through a Maven or Gradle plugin, so a team can try a recipe on one service, read the diff and then apply it across many. It does not replace the review or the tests: some changes still need an engineer, and each run should end with a build and the full test suite.

Related: Spring Boot, Jakarta EE

Orchestration Cluster

What is the Camunda Orchestration Cluster?

The Orchestration Cluster is the core of Camunda 8 from version 8.8: the Zeebe engine, Operate, Tasklist and identity, deployed and run as one unit.

Before 8.8, Zeebe, Operate and Tasklist were separate components with their own APIs. From 8.8 they form one Orchestration Cluster with a single REST API and a shared identity for its users and clients. Zeebe executes the processes, Operate shows running instances and lets you resolve their incidents, and Tasklist gives people their user tasks. Web Modeler and Optimize remain separate components. For teams on Camunda 8 Self-Managed, an upgrade to 8.8 therefore changes the deployment layout. Standard support for Camunda 8.8 ends on 13 April 2027.

ServiceNow AI Platform

What is the ServiceNow AI Platform?

The ServiceNow AI Platform is ServiceNow’s single platform, with one data model and workflow engine, on which its products such as ITSM, ITOM, CSM, HRSD and SPM run.

Every ServiceNow product shares the platform’s database, its CMDB, its workflow engine and its access rules, so an incident, a change and a customer case can refer to the same records. The platform carries AI features, such as generative assistants and agents, beside its classic workflows. For the systems around it, it offers REST APIs and IntegrationHub. In an enterprise estate it rarely stands alone: an integration layer such as IBM webMethods connects it to the applications it manages and to the processes in Camunda.

Spring Boot

What is Spring Boot?

Spring Boot is an open-source Java framework for stand-alone Spring applications, with auto-configuration, an embedded web server and production features such as health checks and metrics.

Spring Boot sits on top of Spring Framework and picks sensible defaults from the libraries on the classpath, so a service starts with little configuration. Microservices, integration adapters and Camunda job workers are often written with it. Each minor line has a published open-source support period. Spring Boot 3.5 left open-source support on 30 June 2026, with 3.5.16 as its last open-source release, and Spring Boot 4.0 leaves it on 31 December 2026; the current line is 4.1. Spring Boot 4 requires Spring Framework 7 and Java 17 or later, and moves to Jakarta EE 11 and Jackson 3.

Strategic Portfolio Management (SPM)

What is ServiceNow SPM?

Strategic Portfolio Management is ServiceNow’s product line for planning portfolios, demand, projects and agile work on the same platform as IT operations and service management.

SPM (formerly ITBM) has carried its name since ServiceNow renamed the product line on 13 April 2022. It covers demand and idea management, portfolio and project planning, agile delivery, resource planning and the alignment of investments with goals. Because it runs on the ServiceNow AI Platform, a project can refer to the same applications and services that ITSM and the CMDB record. A portfolio decision can then be traced to the systems and changes it touches, and its data can flow to finance and planning tools outside ServiceNow.

Trading Networks

What is IBM webMethods Trading Networks?

Trading Networks is the B2B component of IBM webMethods: it manages trading partners and routes, validates and tracks the business documents exchanged with them.

Trading Networks runs on Integration Server. Each partner has a profile with its identifiers and delivery methods. Document types recognise incoming messages, such as EDI, XML or flat files, and processing rules decide what happens to each one: which service runs, whether a reply is due, how it is delivered. Transactions are logged, so a team can answer the question “did the order arrive?” with a record. In an upgrade, partner profiles, document types, rules and their history are assets to inventory and move with care, because partners notice every change.

Universal Messaging

What is IBM webMethods Universal Messaging?

Universal Messaging is the messaging server of IBM webMethods: it carries publish-subscribe channels and queues between Integration Server and other applications, including over JMS.

Universal Messaging decouples the sender of a message from its receivers: a service publishes an event once, and every subscriber gets it, straight away or when it reconnects. It supports durable subscriptions, clustering for availability and the JMS API, so Java applications can use it directly. In IBM webMethods it took over the role of the older webMethods Broker. Moving from Broker to Universal Messaging is often part of a webMethods upgrade. Publishable document types and triggers can usually stay, but the connections, the clustering and the operations change, and each needs a test.

Zeebe

What is Zeebe?

Zeebe is the process engine of Camunda 8, and from version 8.8 the engine inside its Orchestration Cluster: it executes BPMN processes and hands work to job workers.

Zeebe was built for horizontal scale. It splits the work across partitions, replicates each partition across brokers and keeps its state as a stream of events, not in a relational database. It does not run your code: when a process reaches a service task, Zeebe creates a job and waits for a worker or a connector to complete it. That is the main difference from the Camunda 7 engine, which can call Java code inside its own transaction. Operate reads the exported data to show running and completed instances. Camunda 8 runs self-managed on Kubernetes or as SaaS.

Which version are you on?

Tell us the platform, the version and what has to change. We reply with questions or a proposed next step.

Talk to an architect