Kestra is an open-source orchestration platform for data, platform and software engineers, with workflows declared in YAML, self-hosted or in Kestra Cloud.
Kestra gives engineering teams one engine for scheduled and event-driven work across data pipelines, infrastructure jobs and AI. Kestra Technologies builds it, a company founded in 2021 by Emmanuel Darras and Ludovic Dehon. Each workflow is a flow written as declarative YAML, while its tasks run business logic in any language. The core is open source, and Kestra 2.0 is the current release. A flow starts from a cron schedule, a webhook, a real-time event or an update to an upstream asset. Any task can declare its own retries, and an errors path catches what still fails. The topology view, the Gantt tab and the Logs tab follow each execution task by task, and Replay restarts a run from any chosen task. Backfills replay missed schedule intervals, assets carry lineage and freshness checks, and every save keeps a revision that can be rolled back. Blueprints supply validated flow examples, the plugin developer guide covers custom plugins, and an MCP server opens Kestra to AI agents. Kestra is written for data engineers, software engineers and platform engineers, including teams that are moving off Airflow. It suits organizations that want one orchestrator shared by many teams, in fields from finance and healthcare to retail and public services. Groups that must stay on their own servers can run it on premises or fully air-gapped. It comes in three editions: the open-source edition, Kestra Enterprise Edition for governed use across an organization, and Kestra Cloud, the managed service. Data teams chain ingestion, dbt builds, Spark jobs and quality checks into one workflow, and send a Slack message when a run fails. Platform teams standardize Terraform, Ansible and CI/CD workflows across hybrid environments under one set of controls. Flows deploy from Git as soon as changes merge, after testing locally in a single Docker container. Task runners send work to Kubernetes or to AWS, Azure or Google Batch, while AI teams orchestrate agents and RAG pipelines under the same governance.
Pricing
View plans-
Open Source
Free
- Self-hosted, no subscription needed
- Unlimited flows and executions
- 2,300+ plugins
- Declarative workflows, event-driven scheduling, business logic in any language
- Queue on PostgreSQL or MySQL
- MCP Server for AI agents
- Worker Groups not included; all work runs in one default pool
-
Enterprise Edition
Custom pricing
- Price on request; annual subscription, per instance
- Unlimited flows, tasks and executions
- Runs in a cloud, on premises or air-gapped
- SSO, LDAP and SCIM
- Role-based access control and policies
- Promote between instances
- Audit logs and cases
- Multi-tenant and storage isolation
- Worker Groups
- Task runners on containers or VMs
- Queue on Postgres, MySQL, Kafka, Redis or AMQP
- Plugin versioning
- Log shipper and external log data store
- Secrets Manager
- Enterprise plugins
- Apps and reusable inputs
- Standard Support included, by email, with a 24-hour P0 response, 8x5
- Higher support levels add a dedicated Teams or Slack channel and P0 responses of 6 or 1 hours, up to 24x7
-
Cloud Edition
Custom pricing
- Access on request; usage counted in task runs and Cloud runner time
- Kestra runs the compute for Python scripts, shell commands and dbt runs
- Time on your own Kubernetes cluster or batch service is not billed by Kestra
- Loops and retries add task runs
- Groovy script tasks cannot run on Kestra Cloud
-
Expert Advisory Services
Custom pricing
- Price on request; an add-on to Enterprise support
- Included at 6 hours a month on the 24x7 support level
Features
- Not included: Pipelines defined as Python code
- Included: DAG graph view of task dependencies
- Included: Cron and interval schedules
- Included: Sensors and event-driven triggers
- Included: Per-task retries with backoff
- Included: Partitioned backfills
- Included: Asset-level lineage across pipelines
- Included: Local development environment with parity
- Not included: Branch deployments per pull request
- Included: Git-driven deploys
- Included: Run failure alerts to Slack and PagerDuty
- Included: Task-level run history and logs
- Included: Dynamic task mapping at runtime
- Included: Hybrid execution on your own compute
- Included: Managed serverless workers
- Included: Freshness and asset checks
- Included: Open source orchestrator core
Additional Features
- Declarative YAML flows
- 2,300+ plugins
- Business logic in any language
- Cron schedule triggers with time zones
- Webhook and real-time event triggers
- Asset event triggers
- Constant, exponential and random retry policies
- Custom error paths for failed tasks
Best for
- Data engineers running ingestion, dbt and Spark in one engine
- Platform engineers standardizing Terraform and Ansible workflows
- Teams that want workflows declared in YAML rather than in Python code
- Teams replacing Airflow
- Organizations that must deploy on premises or air-gapped
- Enterprises needing SSO, role-based access and audit logs across teams
- Teams moving off VMware Aria Automation
- Companies running data, infrastructure and AI workflows in one orchestrator
Use cases
- Scheduling nightly ingestion and dbt builds
- Starting a workflow from a webhook or a real-time event
- Backfilling missed schedule intervals
- Automating Terraform and Ansible runs across hybrid environments
- Orchestrating AI agents and RAG pipelines
- Replaying a failed execution from the failing task
- Alerting Slack when any flow in a namespace fails
- Promoting flows from development to production
- Running Python scripts and shell commands as tasks
- Migrating scheduled pipelines off Airflow
- Tracking dbt models as assets with lineage
- Finding assets that missed their expected update
Screenshots & Videos
Explore Kestra in action
User Reviews
Compare with Similar Software
Click to compare side by side with Kestra
