What is Managed Service for Apache Airflow?
Managed Service for Apache Airflow, formerly Cloud Composer, is Google’s fully managed workflow orchestration service based on Apache Airflow. It lets you create, schedule, monitor, and manage workflow pipelines that span cloud and on-premises systems, without operating your own Airflow infrastructure. Workflows are defined as DAGs (Directed Acyclic Graphs) in Python.
Core Features
- Managed Airflow on GKE: Environments run on Google Kubernetes Engine, no dedicated Airflow infrastructure required
- DAG-based workflows: Define and version workflows as Python code
- Native GCP integration: Direct connection to BigQuery, Dataflow, Cloud Storage, and more
- Autoscaling and high availability: Automatic worker scaling and options for resilient deployments
- Networking and security options: Public or private IP, VPC integration, VPC Service Controls, IAM-based access control
- Multiple generations: Composer 3 as the current generation, Composer 2 still available, Composer 1 as a legacy option
Common Use Cases
ETL Pipelines
The service orchestrates daily or hourly ETL jobs. Data is extracted from various sources, transformed, and loaded into data warehouses like BigQuery.
ML Pipeline Orchestration
Training, validation, and deployment of ML models as an orchestrated workflow, for example in combination with the Gemini Enterprise Agent Platform (formerly Vertex AI) for end-to-end MLOps.
Cross-System Integration
Workflows connecting multiple systems: loading data into data lakes, generating reports, sending notifications.
Benefits
- Proven Airflow ecosystem with a wide range of operators
- No infrastructure to manage yourself
- Native integration with Google Cloud data services
- Centralized monitoring and alerting via Cloud Monitoring
Integration with innFactory
As a certified Google Cloud Partner, innFactory supports you with Managed Service for Apache Airflow: data pipeline architecture, migration of existing Airflow installations or older Composer generations, DAG development, and performance optimization.
Available Tiers & Options
Composer 1 (Legacy)
- Lower base cost
- Older Airflow versions
- No longer recommended for new environments
Composer 2
- Proven stability
- Autoscaling
- Being succeeded by Composer 3 as the current generation
Composer 3
- Current generation
- Support for newer Airflow versions
- Improved scaling and availability
- Requires migration of existing environments
Typical Use Cases
Technical Specifications
Frequently Asked Questions
What is Managed Service for Apache Airflow?
Managed Service for Apache Airflow, formerly Cloud Composer, is Google's fully managed Apache Airflow service. It orchestrates workflow pipelines across clouds and on-premises data centers.
Why was Cloud Composer renamed?
Google renamed the service to Managed Service for Apache Airflow in 2026 to more clearly communicate its alignment with the open Apache Airflow ecosystem. Functionality and APIs remain compatible.
What is the difference between the Composer generations?
Composer 3 is the current generation with improved scaling and support for newer Airflow versions. Composer 2 remains available, while Composer 1 is considered legacy and no longer recommended for new environments.
Can I migrate existing Airflow DAGs?
Yes, existing DAGs can be imported into the service. Adjustments are usually only needed for custom operators, local dependencies, or when moving to a new major Airflow version.
How does the service scale?
The service runs on Google Kubernetes Engine and offers autoscaling as well as options for highly available deployments; the number of workers adapts to current load.
Note: All product information on this page has been compiled with care, but is provided without guarantee and may be outdated or incomplete. Cloud services evolve rapidly — features, pricing, SLAs, and availability change frequently. Authoritative and up-to-date information can only be found on the official product page of Google Cloud (official documentation). This page does not represent an offer by Google Cloud.
