Cette page est affichée en anglais. Une traduction française est en cours.
ScriptsJul 16, 2026·3 min de lecture

Camunda — Open-Source Process Orchestration for Microservices

A workflow and decision automation platform that orchestrates complex business processes across microservices using BPMN and DMN standards with a visual modeler and scalable Zeebe engine.

Prêt pour agents

Installation agent prête

Cet actif peut être installé après choix du runtime, vérification du plan et exécution de la commande adaptée.

Native · 98/100Policy : autoriser
Surface agent
Tout agent MCP/CLI
Type
Skill
Installation
Single
Confiance
Confiance : Established
Point d'entrée
Camunda
Commande d'installation directe
npx -y tokrepo@latest install c38eb233-80d0-11f1-9bc6-00163e2b0d79 --target codex

À exécuter après confirmation du plan en dry-run.

Introduction

Camunda is an open-source process orchestration platform that helps developers coordinate complex workflows spanning multiple microservices, human tasks, and external systems. Built on the BPMN 2.0 standard, it provides both a visual process designer and a high-throughput execution engine called Zeebe for production workloads.

What Camunda Does

  • Executes BPMN 2.0 workflow definitions across distributed microservices
  • Provides a horizontally scalable workflow engine (Zeebe) based on an event-sourced architecture
  • Evaluates business rules using DMN (Decision Model and Notation) decision tables
  • Offers a web-based process modeler for designing workflows visually
  • Includes operations dashboards (Operate) and task management (Tasklist) UIs

Architecture Overview

Camunda 8 centers on the Zeebe engine, a partitioned, replicated stream processor that persists workflow state using an append-only log inspired by Raft consensus. Workers connect via gRPC to poll for jobs and report completions. The platform includes Operate for monitoring process instances, Tasklist for human task management, and Optimize for process analytics. Each component runs as a separate container and communicates through an Elasticsearch-backed event store.

Self-Hosting & Configuration

  • Deploy using Docker Compose for development or Helm charts for production Kubernetes clusters
  • Configure Zeebe broker partitions and replication factor for throughput and fault tolerance
  • Use the Camunda Modeler desktop app to design BPMN and DMN diagrams before deployment
  • Connect job workers in Java, Go, Node.js, Python, or any language via the gRPC or REST API
  • Set up identity and access management through the included Identity component with Keycloak

Key Features

  • Horizontally scalable Zeebe engine handling millions of workflow instances per cluster
  • Language-agnostic job workers connect via gRPC or REST to execute service tasks
  • Visual BPMN process modeler with validation and simulation capabilities
  • Built-in connectors for REST, GraphQL, Kafka, RabbitMQ, and cloud services
  • Process instance migration for updating running workflows to new BPMN versions

Comparison with Similar Tools

  • Temporal — Code-first workflow engine; Camunda is BPMN-visual with a graphical modeler
  • Apache Airflow — DAG-based data pipeline orchestration; Camunda handles event-driven business processes
  • Netflix Conductor — JSON-based workflow definitions; Camunda uses the BPMN standard with broader tooling
  • n8n — Low-code automation for integrations; Camunda targets complex enterprise process orchestration
  • Step Functions — AWS-native state machines; Camunda is cloud-agnostic and self-hostable

FAQ

Q: Is Camunda free to use? A: Camunda 8 has a Community Edition with the core Zeebe engine under Apache 2.0. Operate, Tasklist, and Optimize have a proprietary license for production use beyond limited free-tier terms.

Q: What languages can I write workers in? A: Official client libraries exist for Java, Go, and Node.js. Community clients are available for Python, C#, Ruby, and Rust.

Q: Can I migrate from Camunda 7 to 8? A: Camunda provides a migration guide and tools. The architecture changed significantly (from embedded Java engine to distributed Zeebe), so migration requires reworking process applications.

Q: How does it handle failures? A: Zeebe automatically retries failed jobs with configurable retry counts and backoff. Incidents are raised for manual resolution in the Operate dashboard.

Sources

Fil de discussion

Connectez-vous pour rejoindre la discussion.
Aucun commentaire pour l'instant. Soyez le premier à partager votre avis.

Actifs similaires